Ubc - 2015 - February - Ferdous - Md. Saimoom PDF
Ubc - 2015 - February - Ferdous - Md. Saimoom PDF
Ubc - 2015 - February - Ferdous - Md. Saimoom PDF
by
in
(Electrical Engineering)
(Okanagan)
January 2015
A radio frequency (RF) heating system has been designed to heat electrolytic liquids
and sludges in a pressurized test vessel. The RF heating system is designed to heat a 400 ml
application for the heating system is to conduct experimental work on the pretreatment of
waste activated sludge, a byproduct of wastewater treatment plants. The dielectric properties
of waste activated sludge were analyzed and it was determined that ionic conduction heating
at a low frequency would be more efficient than microwave heating at a frequency of 2.45
GHz. The RF heating system was tested with sludge and salt water loads over a temperature
range of 20°C to 120°C. Power transfer efficiency from the generator to the load was
excellent and exceeded 86% over the entire temperature range for both loads.
An important part of this work was the design of the RF applicator. The applicator
consists of two circular electrodes mounted inside a Teflon cylinder to uniformly heat the
load. The Teflon cylinder has seals and can heat loads under pressure. The load cylinder is
enclosed in a coaxial housing and the geometry of the load cylinder and enclosure are
implemented in LabView to adjust the generator power for a target ramp rate or to maintain a
steady state temperature. Extensive electromagnetic simulations were done to verify and
optimize the applicator design. Analytic, simulation and experimental results are compared
ii
Table of Contents
Abstract .......................................................................................................................... ii
List of Tables................................................................................................................. vi
Chapter 1: Introduction............................................................................................... 1
2.2.1 Conductivity............................................................................................ 10
iii
2.3.3 Dielectric Properties of Waste Activated Sludge.................................... 24
3.4 Analytic and Simulated Results for Power Transfer Efficiency ..................... 55
iv
4.2.4 Thermal Gradient between the Electrodes .............................................. 88
v
List of Tables
Table 3.2 Comparison of the electrical properties of salt water and waste
Table 3.3 Equivalent circuit and geometrical parameters for WAS and
Table 3.4 Capacitance and inductance of the connecting coaxial cable. ............................... 52
Table 3.5 Intrinsic capacitance and inductance for the heating apparatus. ............................ 54
Table 4.2 Ramp rate of the RF heating device with 0.03315 N salt water load. ................... 84
Table 4.3 Comparison of ramp rates for 0.03315 N salt water and 4.5% WAS. ................... 86
vi
List of Figures
Figure 2.1 Conventional and electromagnetic heating temperature profiles across the load. . 9
Figure 2.3 Ionic current flow through a salt water electrolyte. .............................................. 12
Figure 2.4 Ohmic heating from conduction current (left) and ionic current (right). ............. 13
Figure 2.6 Equivalent circuit for a medium with complex permittivity. ............................... 16
Figure 2.7 The real part (blue) and imaginary (part) of the complex permittivity
Figure 3.3 Electric fields for a parallel plate structure (top) and a coaxial
vii
structure (bottom).................................................................................................. 37
Figure 3.4 Electric fields in the environment around a parallel plate applicator. .................. 38
Figure 3.8 RF heating system: (a) block diagram with equivalent circuit;
Figure 3.9 Equivalent circuit representation of (a) WAS and (b) salt water solution. .......... 48
Figure 3.10 Impedance of salt water and WAS as a function of temperature. ...................... 50
Figure 3.11 Power transfer efficiency of salt water and WAS load
Figure 3.12 Connecting coaxial cable geometry: (a) cross section; (b) side view. ............... 52
0.03315 N salt water (a) and 0.069 N salt water (b). .......................................... 55
Figure 3.14 Impedance of the RF heating system for a 0.03315 N salt water load. .............. 56
Figure 3.15 Impedance of the RF heating system for a 0.069 N salt water load. .................. 57
Figure 3.18 Electric field across the load as function of distance (arc length)
viii
for (a) 0.03315 N salt water load and (b) 0.069 N salt water load....................... 59
Figure 4.1 Pictures of the experimental test bed: (a) RF heating system
Figure 4.4 Photo of the top electrode assembly in the heating apparatus. ............................. 69
Figure 4.7 Photo of the heating apparatus and the corresponding equivalent circuit model. 75
with a 0.03315 N salt water load for a frequency range of 10-15 MHz. .............. 78
Figure 4.10 RF generator voltage (a) and load temperature (b) as a function of time........... 80
Figure 4.11 LabView front panel control interface: (a) temperature graph
Figure 4.12 Ramp rate measurement for salt water load with 250 V (blue line),
200 V (magenta line), 150 V (red line) and 100 V (yellow line)
Figure 4.13 Ramp rate measurement for a 4.5% WAS load with 100 V input to the PA. .... 85
Figure 4.14 Settling time for a target temperature of 120 ̊C (blue line:
ix
Figure 4.15 Settling time for a target temperature of 30 ̊C (blue line:
Figure 4.16 Thermometer tip positions for measuring temperature across load. .................. 89
Figure 4.17 Marks on the mercury thermometer used to set insertion depth in the load....... 90
Figure 4.18 Thermal profile across 0.03315 N salt water load for a target
temperature of 30°C and 40°C with 250 V (blue line), 200 V (red line),
Figure 4.19 Thermal profile across 0.03315 N salt water load for a target
temperature of 50°C and 60°C with 250 V (blue line), 200 V (red line),
Figure 4.20 Thermal profile across 0.03315 N salt water load for a target
temperature of 70°C and 80°C with 250 V (blue line), 200 V (red line),
Figure 4.21 Thermal profile across 0.03315 N salt water load for a target
Figure 4.22 Temperature gradient across the load due to heat conduction loss. ................... 95
Figure 4.23 Temperature gradient across the load due to heat conduction
0.03315 N salt water load for two different ramp rates. ..................................... 97
plus the coaxial cable (overall system) with a 0.03315 N salt water load. ........ 98
x
Figure 4.26 Measured power transfer efficiency with a 0.03315 N salt water load. ............. 99
Figure 4.28 Measured power transfer efficiency for the RF heating system
Figure 4.30 Power transfer efficiency comparison for the RF heating system
xi
List of Abbreviations
DC Direct current
EM Electromagnetic
HA Heating applicator
MW Microwave
OH Ohmic
PA Power amplifier.
RF Radio frequency
xii
Acknowledgements
I would like to express my sincere thanks and gratitude to my supervisor, Dr. Thomas
Johnson for his generous support, guidance and encouragement throughout my research.
patience that has helped me to focus and overcome many obstacles throughout my research.
I heartily thank Dr. Cigdem Eskicioglu and Dr. Kenneth Chau for supporting and
I also sincerely thank Dr. Jake Bobowski and Ehssan Hosseini for their altruistic help
Finally, I especially want to thank my parents for their selfless support and
xiii
Dedication
xiv
Chapter 1: Introduction
enhance the bio-gas production from sludge generated by wastewater treatment plants. The
Bioreactor Technology Group at UBC is leading this work and several years ago a
pre-treat the water treatment sludge. More precisely, the sludge is called wasted activated
sludge, and the acronym WAS will be used interchangeably with the word ‘sludge’.
The motivation for heating WAS is to breakdown the organic particles in the sludge
and preferably breakdown cell membranes to release intracellular material into the liquid
phase. After pretreating, the sludge is moved to a digester where anaerobic micro-organisms
convert proteins and complex carbohydrates into methane gas that can be captured and used
as an energy source. Pretreatment methods also have other advantages such as increasing the
throughput rate through the digester which reduces the volume of the digesters requirement,
killing pathogens and reducing the residual organic mass left after the digester phase.
WAS is a mixture of liquids and organic solids called bio-solids. The mass ratio of
solids to liquids is controlled and two common ratios are 4.5% and 18% where the ratio
specifies the percent content of solids in the total mass of the sludge.
Although the original intent of this research project was to investigate methods to heat
waste activated sludge, the work has led to the design of a RF heating apparatus that in
general can be used to heat liquids and sludges provided the material has certain electrical
properties that are compatible with the frequency of the generator used in this system. The
1
Within the environmental and civil engineering research community, experiments to
develop improved pretreatment methods to enhance the production of bio-gas from WAS has
been going on for many years. Amongst the various methods which have been evaluated to
pretreat WAS, microwave pretreatment and conventional heating are included. In microwave
pretreatment, the sludge is heated in a commercial microwave heating system used for
laboratory applications. These systems heat loads at a frequency of 2.45 GHz and are very
similar to domestic microwave ovens in that the principal heating mechanism relates to the
dielectric relaxation of water molecules. Conventional heating is a process where the load is
heated by a source using thermal convection. Ovens are an example of conventional heating
systems. The motivation for this research project is to answer the broader question: what is
In 1977, Haug et al. [1] demonstrated that adding a thermal pretreatment stage before
anaerobic digestion enhances the bio-degradability of organic materials. The enhanced bio-
degradability in this context means improved bio-gas production in the anaerobic digesters as
well as improvements in the ability to dewater the sludge and reduce the volume and cost of
disposing of the residual solids. The experiments conducted in Huang’s study used
conventional heating.
Over the past several decades many other pretreatment processes have been evaluated.
These methods include chemical and enzyme additives, high pressure homogenization,
microwave heating and exposure to pulsed electric fields [2-5]. The motivation for all these
pretreatment methods is to make organic materials present in the sludge more biodegradable
2
by physically reducing the size of organic structures and puncturing or rupturing cell
membranes [4]. Amongst all the pretreatment methods, thermal pretreatments have been very
Studies have shown that both conventional and microwave heating pretreatment
methods can improve the bio-gas production from WAS. In a recent study by Mehdizadeh et
al. [6], a comparison was made between conventional heating and microwave heating at 2.45
GHz. The study showed, that under identical heating profiles, similar amounts of bio-gas are
produced. Therefore, there is no current evidence that the heating method of sludge has a
significant impact on bio-gas production. More important factors appear to be the thermal
ramp rate and the operating temperatures involved in the pretreatment process.
From a system perspective, the energy input to the heating system and the efficiency of
converting the energy to heat the load are very important. The evaluation of heating
efficiency involves both the electrical design of the heating system as well as an evaluation
of the heating uniformity across the load. There has been much less work focused on
investigating efficient heating methods and the question of how best to heat WAS has
As will be explained in Chapter 2, a radio frequency (RF) heating system for treating
WAS was selected with an operating frequency of 13.56 MHz. Therefore, it is relevant to
applications include food and agriculture processes and wood drying [7].
RF heating methods were first employed in food processing as early as 1940 [8]. RF
heating has been used to dehydrate vegetables and fruits [9], bake cookies and crackers [10],
3
cook meat products [11], kill navel orangeworms in walnuts [12] and reduce microbial action
in fruit juices [13, 14]. Advantages of RF over conventional heating are improved product
Most RF heating systems in these applications use a frequency of 27.12 MHz which
corresponds to an allocated frequency band reserved for industrial, scientific and medical
(ISM) applications. Two other ISM frequency bands which can be used for RF heating are
13.56 MHz and 40.68 MHz and a few papers have reported on RF heating systems in these
the published work focuses on optimizing the process rather than optimizing the electrical
design of the heating system. Also, the majority of these systems are designed to heat loads
up to moderate temperatures usually below 100ºC. One of the requirements in this research
project was to heat liquid and sludge loads to temperatures up to 120ºC under constant
volume conditions. Consequently, a heating apparatus that can work under high pressure is
required.
The primary objective of this research project was to design an electromagnetic heating
applicator that can efficiently heat WAS. After designing and implementing the heating
apparatus, the equipment will be used to conduct experiments to measure how effective the
conventional heating, this research began with a study of the electrical properties of WAS.
The dielectric properties of WAS were measured over a wide frequency range from 100 kHz
4
to 40 GHz and these results were published several years ago by other members of the UBC
research team [17]. The dielectric study provides a fundamentally different starting point for
the electrical properties of the material and then selecting an optimum heating frequency.
Selecting an optimum heating frequency involves more than a study of material properties
and other important requirements are uniform heating throughout the load volume, efficient
high power RF generators, efficient matching between the generator and the load, and
constraints in terms of frequency bands which can be used for industrial heating applications.
After selecting a frequency, the next objective was to design an applicator which could
heat liquid and sludge loads over a temperature range of 20ºC to 120ºC. The minimum load
volume for the apparatus was 400 ml and the volume and temperature range requirement
thermal ramp rates up to 15ºC per minute and a temperature measurement system that can be
different mechanisms which can create heating in WAS are reviewed. It is shown that WAS
has significant ionic conductivity and that the bio-solids have a dielectric dispersion called β
dispersion related the electrical characteristics of cell membranes in the organic material.
models of load impedance as a function of temperature are derived based on matching the
5
ionic conductivity of WAS with an equivalent salt water load. The analytic models are used
to design an applicator which creates a uniform electric field between two electrodes. The
electrodes are mounted in a Teflon pressure vessel and the entire pressure vessel is enclosed
matching the load impedance to a 50 Ω high power RF generator. The final design is
Chapter 4 focuses on experimental results. The RF heating applicator was built and
tested. The RF heating test bed consists of a RF generator, an applicator, and software to
measure load temperature and control RF generator power. The applicator was tested with
salt water loads and samples of WAS. Loads were heated from room temperature to 120ºC
and impedance measurements of the applicator and load were made to verify the design.
Excellent results have been obtained and the RF heating applicator shows significant promise
and other potential applications for the RF heating system are described in the concluding
chapter.
6
Chapter 2: Selecting a Frequency for the RF Heating System
The theory of how electromagnetic fields interact with biological materials is essential
to the design of the heating system for WAS. A time varying electromagnetic field has both
an electric and magnetic field and the relations between the two fields are described by
Maxwell’s equations. From the heating perspective, heat is generated in dielectric and
metallic materials principally by the interaction of the electric field and charge.
heating using electromagnetic fields are conductivity and permittivity. Conductivity and
frequency sensitive. The frequency dependence then raises the question: what frequency is
are other constraints which affect the selection of a heating frequency for WAS. The
penetration depth of the electromagnetic wave affects the heating profile through the
material. Since penetration depth varies inversely with frequency, a low frequency source has
advantages over high frequencye. Other practical constraints include regulatory issues which
limit specific frequency bands for heating applications as well as the cost and power
efficiency of the amplifiers used to generate the high frequency signals to heat the material.
electrical properties of materials and how the properties affect the heating process. The
electrical properties of materials are frequency dependent and these relations are described
for materials relevant to the WAS application. Finally, a frequency of 13.56 MHz is selected
7
for this design and a discussion of the factors which were considered in the selection of this
frequency is presented.
created in the material, heating methods can be broadly classified into two categories:
the material and thermal energy transfers by convection through the material from a region of
where heating coils are the source and heat is conducted through the material created by the
temperature gradient between the source and load. Drawbacks of conventional heating are
uneven heating across the load because of the thermal conduction gradient and slow heating
rates. Uneven heating arises from the conventional heating can be made quite even by
inserting multiple electrodes inside the load, which requires intervening into the materials.
electromagnetic heating process, the electric field interacts directly with the molecules in the
load and increases the kinetic energy of the molecules to heat the load. Electromagnetic
waves propagate with very high velocity, and providing the electric field penetrates through
the load, uniform and near instantaneous heating can be obtained. Because the field
penetrates the entire volume, the term volumetric heating is sometimes used to distinguish it
from conduction heating. The primary drawback of electromagnetic heating is that its
temperature profiles for conventional and electromagnetic heating processes are shown in
Figure 2.1.
8
Molecular vibration
Temperature Temperature
Distance Distance
Conventional heating Volumetric heating
Figure 2.1 Conventional and electromagnetic heating temperature profiles across the load.
When an electromagnetic field interacts with a material, power can be dissipated. The
by Maxwell’s equation are linked by three constitutive parameters which relate to the
material properties and the fields that can exist in the material. The constitutive parameters
properties, conductivity and permittivity are relevant to heating for non-magnetic materials.
9
2.2.1 Conductivity
purely real number and ideally measured at DC (0 Hz)1. The conductivity of a material
relates to the movement of free charge. When an electric field is present in a material, the
electric field exerts a force on free charge, and the charge drifts through the material. The
F qE (N). (2.1)
Examples of free charge are electrons that move through a metal lattice in a conductor or free
The movement of electrons through a conductor is called electronic current flow and
does not involve the net movement of mass. Electrons in the valence band of a metal are
easily removed by a low energy electric field and they drift from atom to atom. The
movement of free charge is created by electron-hole pairs and the atomic lattice is unchanged
Figure 2.2. Current flow through a metallic conductor does not involve any electrochemical
reactions at the electrodes (anode and cathode) and the only chemical reaction is in the
1
Experimental measurements of conductivity at DC can be complicated by
electrochemical reactions between the electrodes and the material being measured. Therefore,
DC conductivity may be extrapolated from low frequency measurements where
electrochemical reactions do not contribute significantly to experimental errors in the
measurement of conductivity.
10
Figure 2.2 Conduction current flow through a metallic conductor.
Ionic current flow is distinct from electronic current flow. In ionic current flow, ions
(atoms or molecules with a net charge) physically move and there is a net movement of mass.
Under a static (DC) electric field, ions will drift in an electrolyte and eventually current flow
An illustration of ionic current flow through salt water (NaCl) is shown in Error!
Reference source not found.. Salt water is an electrolyte and salt molecules separate into
positively charged sodium cations (Na+) and negatively charged chlorine anions (Cl-). In an
electric field created by the potential difference between the two electrodes, ions drift to the
charged electrodes. The positive electrode is called the cathode and attracts negatively
charged chlorine ions, while the negatively charged electrode is called the anode and attracts
positively charged sodium ions. At the electrodes, a chemical reaction is required to generate
free electrons which can then create electronic current flow through the wires to the battery.
11
The electrochemical reaction at the electrodes creates chlorine gas at the cathode and
hydrogen gas at the anode. As times goes by, due to chemical reactions at the electrode
interfaces, all the ions in the electrolyte will eventually be depleted. Neutrally charged
molecules build up at the electrodes and this leads a condition known as electrode
polarization.
undesirable because they produce gas and the electrochemical processes can be reduced by
converting to an alternating current source with a sufficiently high frequency. For salt water,
the electrochemical reactions are negligible above 10 MHz. Biological material has
12
significant electrolytic properties and ionic current flow is relevant to the heating mechanism
collisions with atoms and molecules generate heat. The amount of energy released depends
on the mobility of the charge carriers (conductivity) and the intensity of the electric field.
Heat generated by conduction current, either electronic or ionic, is called ohmic heating. An
Anion
- Electron - Na+ Cl-
H 2O Na+
Na+
Nucleus + - + - Dipolar molecule
Cl-
+ - - - - - H 2O
Cl- H 2O
+
Cl- Cation
+ - + -
H 2O Na+ Cl- Na+
-
Resistor Electrolyte
Figure 2.4 Ohmic heating from conduction current (left) and ionic current (right).
Ohmic heating is described by Ohm’s law. The conduction current density Jc through a
J c DC E (A/m2) (2.2)
13
pc DC E (W/m3)
2
(2.3)
The total power dissipated in a volume is found by integrating the power density over the
volume. In general, the electric field varies as a function of the spatial position ( x, y,z ) in the
volume and a triple integral is required to calculate the total power dissipated in the material.
In this work, the integral is evaluated numerically using an electromagnetic modeling tool
heated, the conductivity increases because of the increased mobility of charges. If the
conductivity changes significantly with temperature, then the net energy absorbed by a load
depends on time and the current temperature of the load. The temperature dependence of
When a material with bound charge is placed in an electric field, a polarization charge
( P ) is induced within the material. The polarization charge can be visualized as the atomic
realignment of charge where electrons are attracted in a direction of increasing potential and
positive charge is repelled in a direction of decreasing potential. The atomic structure and
chemical bonds in the material may prevent the charge from moving and the charge
redistribution is localized within the atomic lattice; hence the name bound charge.
water. The orientation of polar molecules is affected by the presence of an electric field and
there can be a net alignment of molecules which also creates a polarization charge in the
material.
14
When material is exposed to a time varying electric field, the polarization charge
oscillates as the polarity of the electric field changes direction. As frequency increases, the
polarization charge cannot maintain perfect synchronization with the oscillating electric field
and the polarization charge starts to lag behind the polarity of the electric field. The phase lag
means energy from the electric field is absorbed by the material and dissipated as heat. This
type of energy loss is called relaxation loss. Relaxation loss is frequency dependent and the
physical characteristics of this loss mechanism are captured in the imaginary component of
the dielectric constant . Figure 2.5 illustrates the dielectric relaxation loss mechanism.
Electric Electric
field field
-
+ - +
Dipolar
molecule
orientation Following Following
delay delay
+
- + -
Low frequency High frequency (~22 GHz)
Figure 2.5 Electric dipole synchronization with frequencies; left: perfect synchronization at low
To show that the imaginary part of the complex permittivity results in dissipation,
H = ( j ) E (2.5)
H = E + j ( - j ) E
(2.6)
( ) E + j E
15
The term contributes to an effective conduction current ( J c ) that leads to dissipation in
the material. The real part of complex permittivity is associated with a reactive current flow
The distinction between conduction current flow and displacement current flow in a
material can be used to construct an equivalent circuit model of the medium. The circuit
Electric B= 1
X G= 1
+ field, E Plate R
+Q A
area, A ωε r (ω)ε 0 ( +ωε(ω)ε ) A
d r 0
d
-
-
+
Lossy
-
d
+
Dielectric
-
-
Susceptance
+
+
due to
+
Dielectric
-
-Q
Charge
-
Conductance due to
+
dielectric loss
Assume the plate area is large and fringing fields are small such that the electric field
between the plates is uniform. The capacitance of the parallel plate structure is
A
C (F) (2.7)
d
16
where is the complex permittivity, A is the area of the plates and d is the distance between
the plates. The admittance of the capacitor is Y ( ) jC which expands into
( j ) A
Y ( ) j
d
(S) (2.8)
A A
j Gd ( ) jBd ( )
d d
The admittance Y ( ) of the dielectric material can be modeled by a shunt conductance
Gd ( ) and a shunt capacitance with susceptance Bd ( ) . If the material also has finite
A
Gc (S) (2.9)
d
for the parallel plate structure. Note that unlike the conductance in the dielectric model where
Y ( ) Gc ( ) Gd ( ) jBd ( )
( ) A A (S) (2.10)
j
d d
This equation shows that an equivalent circuit model for modeling conductivity and complex
permittivity is a shunt conductance and capacitance. The equivalent circuit model is very
useful and enables the design of matching circuits to couple power to the material. The
17
0 A
Y ( ) j [ ( ) j( + )] (S) (2.11)
d
r r
0
Let C0 0 A / d which is the capacitance of the test structure assuming the dielectric is a
Y ( )
r ( ) - j ( r( ) + ) (S) (2.12)
jC0 0
The equation shows that the material properties as function of frequency can then be obtained
from measurements of admittance as function of frequency. The real component of the right
hand side of (2.12) is the real relative permittivity, r ( ) , and the imaginary component is
dependent on both the imaginary component of the complex permittivity, r( ) , as well as
the conductivity of the material. The conductivity dominates the low frequency
characteristics while the imaginary part of the complex permittivity dominates the high
frequency characteristics. Examples plots of the right hand side of (2.12) are shown later.
The complex permittivity of a material depends on frequency and there are models that
have been developed to relate the characteristics to the atomic structure of the material. An
important model that is widely used is called the Debye model. The Debye model leads to
dispersion which means that the permittivity does not have linear characteristics as a function
frequency.
As described earlier, polarization charge oscillates in a time varying electric field. For
low frequencies, polarization charge can follow the field variation, but as frequency is
increased, the polarization charge starts to lag the variation in the field. The lag between the
18
polarization charge and the electric field can be modeled in the time domain by a relaxation
time constant.
Let D(t) be the electric flux density as a function of time and assume the flux density
pulsed to create the flux density change in the material. Similar to the transient response of a
circuit with a conductance in shunt with a capacitance, the response can be modeled as an
exponential with a relaxation time constant . The flux density is then given by
t
D(t ) D ( D D0 )(1 e ) (2.13)
Equation (2.13) is called the Debye dispersion equation. The equation shows that in one time
constant, the flux reaches 63.2% of its final value, and in 5 , the flux is 99.3% of the final
value D0 . For example, water has a relaxation time of 7.2 ps at a temperature of 30ºC.
The Debye dispersion equation can also be written in the frequency domain by taking
the Laplace transform of (2.13) and then evaluating the transform along the imaginary axis
( j in the complex frequency plane) to obtain the corresponding frequency response. The
( s ) (2.14)
r ( ) r ( ) j r( )
1 j
where, s is the steady state (DC) dielectric constant, is the dielectric constant at optical
4.9, and = 7.2 ps. A plot of the real and imaginary parts of the complex permittivity of
19
80
70
Complex relative permittivity ( r )
60
50
40
30
20
10
0 6 7 8 9 10 11 12
10 10 10 10 10 22.1 10 10
Frequency (Hz) GHz
Figure 2.7 The real part (blue) and imaginary (part) of the complex permittivity of water at 30ºC.
As shown, significant dispersion in the dielectric properties occurs in a frequency range from
2 GHz to 200 GHz. Below 2 GHz, r s and above 200 GHz, r . Around 22 GHz, the
imaginary part of the permittivity peaks and this frequency corresponds to the frequency
where the most power is dissipated in the water. A microwave oven operates at a frequency
of 2.45 GHz and balances dissipation (heating), penetration depth (uniform heating) and the
dispersion regions which relate to multiple relaxation time constants. Also, high ionic
Biological materials are examples of materials with more complex characteristics and these
20
2.3.2 Electrical Models for Cells and Tissues
The electrical properties of cells and tissues have been the subject of much research
[18]. From this research, equivalent circuit models for cells and tissues have been developed.
Equivalent
Resistance
Equivalent
capacitance
Extracellular fluid
Intracellular fluid
Cell membrane
Plant and animal tissues are heterogeneous cellular structures and tissues are composed of
many cells of varying types and sizes. Each cell within a tissue is encapsulated by a cell
membrane that separates intracellular fluid from extracellular fluid in the tissue. If two
parallel plate electrodes are inserted into the extracellular fluid, current can be passed through
the tissue and cells. The current flow is frequency dependent because the medium is
21
heterogeneous with different electrical properties for different parts of the tissue. The
extracellular and intracellular fluids are electrolytic and an ionic conduction current can flow
through the fluids. The cell membrane is permeable but has much lower conductivity and
membrane can be modelled as dielectric sandwiched between the inner and outer ionic
(conductive) mediums. Therefore the cell membrane is analogous to a parallel plate capacitor
structure and current flow through the cell membrane is primarily displacement current rather
than conduction current. Based on the circuit model, at DC and very low frequencies, ionic
conduction current can flow between the electrodes, and the current path is primarily through
extracellular fluid. As the frequency increases, a displacement current can flow through the
cell membrane and the net current flow between the electrodes is composed of both
conduction and displacement currents. The current flow paths are illustrated in Figure 2.9.
22
The current flow through the cell membrane is more complex than shown in the
simplified model of Figure 2.8. At the exterior interface between the cell membrane and the
extracellular fluid, and at the interior interface between the cell membrane and the
intracellular fluid, there is a wet dielectric boundary. In general, each medium at an interface
ε1 σ1 ε2 σ2
R1 R2
C1 C2
d l
Relative
permittivity
dl
Frequency f = 1
dl
(Hz) 2πτdl
The conductivity in each medium can be modeled as a conductance and the permittivity can
be modelled as a capacitance. For a parallel plate structure, the capacitance and conductance
of the mediums is given by (2.9) and (2.11), respectively. The conductance and capacitance
in each medium are connected in shunt and the electrical circuit has a time constant,
23
dl C / G . The time constant is similar to the relaxation time constant in the Debye model
except that it arises from a different physical phenomenon where free charge carriers
interface is called the Maxwell-Wagner effect [18]. The Maxwell-Wagner effect creates an
dl (2.15)
dl dl j
0
dielectric dispersion that is similar to a Debye dispersion and an example of the dispersion
The primary load to be heated is WAS which is a complex mixture of organic solids
immersed in an inorganic liquid composed mostly of water and electrolytes. The solid
content in WAS is specified by a mass ratio of solids to liquids and two common mixture
ratios are 4.5% and 18%. The solid content in WAS, also called floc, is composed of a
heterogenous mixture of inorganic and organic solids where the organic solids have
A dielectric study of 4.5% and 18% WAS was made by the UBC research group
several years ago [17]. The results of these measurements have been analyzed and an
24
150
100
( 'r )
50 Bulk water
Dominant biological characteristics characteristics
0 6 7 8 9 10 11 12
10 10 10 10 10 10 10
Frequency (Hz)
4
10
( "r+/0 )
0
10 6 7 8 9 10 11 12
10 10 13.56 10 10 10 22.1 10 10
MHz Frequency (Hz) GHz
The electrical characteristics are shown in the form of (2.14) where the real part is r and the
imaginary part is r DC 0 . At low frequencies below 300 MHz, ionic conductivity
dominates the imaginary component of the dielectric characteristics. The liquid is composed
of electrolytes such as salt which makes the sludge conductive. The ionic component
corresponds to the term DC 0 and this leads to a straight line slope on a log-log plot.
Within the 1-10 MHz range, dispersion is evident in the real part r of the complex
permittivity. The response can be compared to the graph shown in Figure 2.10 (Maxwell -
Wagner effect). The β dispersion is attributed to the organic nature of the bio-solids in the
sludge. From 10 MHz to 5 GHz the real part of the permittivity is approximately flat and
equal to the permittivity of water (~80). At a frequency of 22 GHz, the Debye relaxation
associated with water is clearly visible. This leads to a decrease in the real part of
permittivity and a corresponding peak in the imaginary component. The power absorbed by
25
water molecules is maximized at a frequency of 22 GHz but the penetration depth of the
electromagnetic wave is very shallow. Microwave ovens balance penetration depth with the
optimum frequency for heating water and a frequency of 2.45 GHz is used in most
The imaginary part of the complex permittivity determines how efficient the material
can be heated using electromagnetic waves. Microwave heating at 2.45 GHz falls in the
valley between the low frequency slope and the dispersion at 22 GHz. Below 100 MHz, it is
clear that heating is even more efficient than the dispersion peak at 22 GHz. Furthermore, the
Based on the electrical properties of WAS, a low heating frequency appears to be the
most optimal in terms of maximizing the dissipated power density in the load. A low
frequency also has advantages in terms of a long wavelength which leads to deep penetration
depth. According to the slope in Figure 2.11, one might be tempted to conclude that the best
heating frequency would be 0 Hz; in other words DC. The issue with heating at DC is that
electro-chemical reactions are necessary at the electrode interfaces to create free electrons
which can flow through the external circuit. This was illustrated earlier in Error! Reference
source not found.. The electro-chemical reactions have other byproducts including the
generation of gases and polarization charge on the surface of the electrodes. Therefore, there
are issues with operating at too low a frequency. A summary of the different heating
mechanisms and the associated dielectric characteristics for WAS is shown in Figure 2.12
26
Figure 2.12 Complex permittivity of 4.5% waste activated sludge (figure based on data obtained
from [17]).
27
Figure 2.13 Complex permittivity of 18% waste activated sludge (figure based on data obtained
from [17]).
significantly reduced for frequencies above a few MHz [17] (region B in the figures). There
are also limitations in the available frequency bands which can be used for industrial heating
28
applications. The appropriate frequency bands for heating are associated with frequency
bands reserved for industrial, scientific and medical (ISM) applications. The lowest
frequency band which falls near the low end of the MHz frequency range is at 13.56 MHz.
The corresponding free space wavelength at 13.56 MHz is 22 meters which means the field
The other advantage of operating at a frequency of 13.56 MHz is that it falls very near
the β dispersion associated with cell membranes (region B). Since the objective of the
heating process is to disassociate tissues and cells, absorption of power in the cell membranes
is beneficial and this heating mechanism is absent in region C. However, the heating
associated with dispersion is expected to be much less than ionic conduction heating and the
More discussion on the relative power dissipation for ionic heating and dielectric relaxation
heating from β dispersion is discussed in the next chapter. Therefore, based the preceding
29
Chapter 3: Design of the RF Heating System
models of the load impedance as a function of temperature are derived based on matching the
ionic conductivity of WAS with an equivalent salt water load. The analytic models are used
to design an applicator which creates a uniform electric field between two electrodes. The
electrodes are mounted in a Teflon pressure vessel and the entire pressure vessel is enclosed
matching the load impedance to a 50 Ω high power RF generator. The final design is
simulated and optimized in an electromagnetic simulator and then compared with the analytic
model.
After selecting a frequency of 13.56 MHz for the RF heating system, detailed electrical
WAS were obtained from another study which was carried out by the UBC research group
[17]. From this study, values for conductivity and complex permittivity of WAS are known.
Although heating WAS is the final objective of this project, it is very useful to work
with a controlled load with well-known electrical characteristics. Salt water, with a controlled
molarity, can be prepared to have similar conductivity to WAS. Advantages of salt water
accurate DC conductivity, and a load that is free of bio-hazards which makes handling and
testing easier. There are also excellent literature references which have empirical equations
for predicting how conductivity and complex permittivity vary with temperature. The
30
temperature characteristics of salt water are used in the design to predict how the impedance
match between the load and the generator will vary over temperature. The electrical
properties of WAS and an equivalent salt water load are presented next.
WAS samples consist of a liquid with a controlled amount of bio- solid content. The
samples are specified by a ratio of the mass of the solids to the mass of the liquid and two
ratios, 4.5% and 18%, are commonly available from wastewater treatment plants. Based on
the experimental measurements presented in [17], the electrical properties of WAS are
Table 3.1 Electrical properties of waste activated sludge at a frequency of 13.56 MHz and at a
temperature of 25°C.
r 92.5 138.8
Important limitations of this data should be noted. First, the electrical properties were
measured at a single temperature of 25°C. Second, the measurements are based on a small
number of samples and some variation in the electrical characteristics is expected. Third, the
extraction of r from the original measurement data reported in [17] is sensitive to
measurement error at a frequency of 13.65 MHz because r is much less than / 0 .
Despite these limitations, within the scope of this project, the reported measurements are
used as a basis for the applicator design. As shown later in Chapter 4, experimental results
31
differ from the design target, but the variation is within acceptable margins and efficient
The total amount of power which is dissipated in WAS at a frequency of 13.56 MHz
can be found by modifying equation (2.3) in Chapter 2 to include dielectric heating losses
1 1
p ( DC r 0 ) E ( DC r) 0 E (W/m3)
2 2
(3.1)
2 2 0
1
pc DC E (W/m3)
2
(3.2)
2
1
pd r 0 E (W/m3)
2
(3.3)
2
Using these equations and Table 3.1, the ratio of ohmic heating to total heating can be
calculated. For 4.5% sludge, ohmic heating accounts for 99.4% of the dissipation in the load,
while ohmic heating accounts for 96.2% in 18% sludge. Clearly, at frequency of 13.56 MHz,
heating. As described earlier, it is convenient to have a controlled load for testing the heating
apparatus. Given that heating in WAS is dominated by ionic ohmic heating at a frequency of
13.56 MHz, salt water is a good choice for creating an equivalent load with similar heating
properties. Stogryn et al. [20], presented a method of determining the amount (molarity) of
sodium chloride salt required in pure distilled water to achieve specific dielectric
32
characteristics. A model for the complex permittivity of a saline water solution ( SW ) as a
function of temperature (T) in degrees Celsius, salinity (N) in moles per liter of solution and
frequency (ω) is
0 _ SW (T , N ) - _ W DC _ SW (T , N )
SW (T , N , ) _ W j
1- j SW (T , N ) 0 (3.4)
In this equation, _W is the high frequency permittivity of water and equal to 4.9. The other
terms in the equation are dependent on temperature and salinity, and the equations for these
25 T (3.11)
Using the salt water model equations, the molarity of the solution can be found to have
a dissipation at 13.56 MHz which is equivalent to WAS. For 4.5% WAS, the equivalent
molarity (N) is 0.03315 N, and for 18% sludge, the equivalent molarity is 0.069 N. The
electrical properties of the equivalent salt water loads are summarized in Table 3.2.
33
Table 3.2 Comparison of the electrical properties of salt water and waste activated sludge at a
temperature of 25°C.
r DC r DC
r
0
water
water
The electrical property equations (3.4) through (3.11) are very useful to model the
impedance changes in the load as a function of temperature. In order to gain insight into the
temperature sensitivity of conductivity and r , the equations are plotted for a temperature
range of 20°C to 120°C in Figure 3.1 and Figure 3.2, respectively. The temperature range
conductivity of 0.03315 N and 0.069 N salt water changes significantly and increases by
approximately a factor of four from 20°C to 120°C. The relative permittivity of salt water
changes less and decreases from approximately 79 at 20°C to 55 at 120°C. The temperature
34
variations of the material properties are used later to find the best compromise in terms of
3
0.069 N salt water
0.03315 N salt water
2.5
DC conductivity (S)
1.5
0.5
0
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
80
0.03315 N salt water
0.069 N salt water
75
Real part of relative permittivity ( 'r)
70
65
60
55
50
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.2 Real part of relative permittivity of salt water solution as a function of temperature.
35
3.2 Heating Applicator Physical Design
The physical geometry of the heating applicator is very important and affects the
heating profile across the load as well as the load impedance. Many different design concepts
were evaluated and a summary of the methodology employed to arrive at the final design is
summarized.
Uniform heating across the load requires a uniform electric field. An electric field is
created by applying a potential difference across two electrodes, and of all possible
geometries, a parallel plate structure is the only geometry that creates a uniform electric field.
An example of a parallel plate structure is shown in Figure 3.3. Although the ideal parallel
structure has a uniform electric field, a practical parallel plate structures has a fringing
36
Electric field
Uniform electric
field
A
VAB
d E= =constant
d
B
Non-uniform
electric field
r
B A
1
E
r
(E field decays w ith (r))
Figure 3.3 Electric fields for a parallel plate structure (top) and a coaxial structure (bottom).
Other geometric configurations of electrodes could be used and the geometry could be
scaled to minimize the variation of the electric field throughout the volume of the load. For
example, in Figure 3.3, a design with coaxial electrodes is shown. For this geometry the
electric field between the two plates varies as 1/r where r is the radial distance from the
center of the cross-section. If the inner conductor diameter is large such that the distance
between the electrodes is small, then the field variation can be reduced. The disadvantage of
37
the coaxial design is that there would be a large volume in the center which cannot be used.
Although the parallel electrode structure provides a starting point for the design it is
not ideal. First, the electrodes need to be connected to a coaxial connector to interface to the
RF generator. The electrode connections and wiring introduce parasitic inductance and
capacitance that must be minimized and controlled. Second, the load is a liquid and needs to
pressure as the maximum operating temperature is above 100°C. Third, very high electric
fields are used to heat the load and the radio frequency fields can radiate into the surrounding
completely shield the load and the electrodes. These factors are illustrated in Figure 3.4.
Inner conductor
A Electric field in between
parallel plate capacitor
+
d Leakage electric field
- spread into space
Coaxial cable
Electric field inside
coaxial cable
B
Outer conductor
Figure 3.4 Electric fields in the environment around a parallel plate applicator.
As a way to control the electric fields from the connector interface to the parallel plate
applicator, a hybrid design is created which encloses the parallel plate structure in a coaxial
38
structure. The design concept is shown in Figure 3.5. At the top of the enclosed structure is a
coaxial cable which connects to the RF generator. The cable is physically attached to the
applicator through a coaxial connector. The outside of cylinder is grounded and completely
envelopes the load. The bottom electrode is connected to the outside shield through an
elevated post. The top electrode is attached by a second post which connects to the center
conductor of the coaxial cable. Since the outside structure is a cylinder, the electrodes are
circular disks. The fringing fields inside the enclosure are controlled by adjusting the distance
39
Figure 3.5 Parallel plate applicator enclosed in a coaxial cylinder.
There are thermal and mechanical considerations that must be considered in selecting
the materials for the heating applicator. Starting with the load, a vessel is required to contain
the liquid load and attach the electrodes to the load. The container must be a dielectric and
Teflon is selected for the load container. Teflon has excellent electric characteristics with
very low dielectric losses. Teflon also has excellent thermal properties. The thermal
conductivity of Teflon is 0.25 W/m/K and has approximately half the thermal conductivity of
40
water, 0.58 W/m/K. Because the load container is operated under pressure, the wall thickness
of the Teflon pressure vessel must be large to handle the mechanical stress.
The electrodes contact the load and they need to be electrically conductive. Suitable
materials include copper or aluminum. Aluminum electrodes are selected as aluminum has
lower thermal conductivity (238 W/m/K) than copper (401 W/m/K) which reduces thermal
losses through the electrodes. Aluminum is also lighter than copper, 2700 kg/m3 versus 8960
kg/m3. The electrical conductivity of aluminum (3.77×107 S/m) is lower than copper
(5.9×107 S/m), but still has much higher conductivity than the load and ohmic losses in the
electrodes are negligible. Another consideration is electrode corrosion. In this application, the
significant issue.
The outside enclosure needs to be electrically conductive and the diameter was selected
based on the availability of standard sizes of aluminum piping. In this design, 8 inch diameter
piping with a wall thickness of 5/16 inch is used. The posts used to provide electrical
The optimization of the heating applicator geometry including load matching was done
using an electromagnetic simulator (COMSOL). The final dimensions for the heating
applicator are shown in Figure 3.6 and Figure 3.7. Two different designs are made to match
to the two different loads: 0.03315 N salt water or 4.5% WAS and 0.069 N salt water or 18%
WAS. The electrical circuit model of the load corresponding to the final geometry is
41
Inner conductor
Outer conductor
.1 2.54
.22 Inner conductor
3.25
6.27 (Aluminium)
Outer conductor
1.27 (Aluminium)
Air
12.06
24.6 Load
25.8
10.45 Dielectric
.42 .69 (Teflon)
6.27
1.48 1.28
.632
20.24
(All digits are in cm)
Figure 3.6 Dimensions and material specifications of the heating applicator for a 0.03315 N salt
water load.
42
Inner conductor
Outer conductor
.1 2.54
Inner conductor
.22 (Aluminium)
2.73
6.27 Outer conductor
(Aluminium)
1.27 Air
Load
17.12
29.66 Dielectric
30.86 (Teflon)
10.45
.42 .69
6.27
1.48 1.28
.632
20.24
(All digits are in cm)
Figure 3.7 Dimensions and material specifications of the heating applicator for a 0.069 N salt
water load.
The RF heating apparatus is designed to transfer power from a source to a load. The
source in this system is a high power RF amplifier and the load is salt water or WAS. The
overall heating efficiency of the system depends on the how much power is transferred from
the source to the load. Electrically, power transfer from the source to load is maximized
43
providing the impedance of the load is conjugately matched to the impedance of the
amplifier. Since the load impedance in this system changes with temperature, the amount of
power delivered to the load varies with temperature. When the source and load are not
matched, there is a mismatch loss which reduces the efficiency of transferring power to the
load. In the design of the RF heating applicator it is important to analyze the mismatch loss in
the system over temperature and maximize power transfer over the operating temperature
range.
The analysis of the power transfer from the source to load is done by converting the
load and the applicator into an equivalent circuit model. The equivalent circuits are derived
from the geometry of the load and applicator as well as the electrical properties of the load.
In the next section, a brief summary on the concept of matching and transfer is described.
This is followed by a description of equivalent circuit models for the load and the RF
applicator. After the electrical models for the load and RF applicator are available, power
transfer from the amplifier to the load is analyzed. The equivalent circuit models provide a
way to optimize the geometry of the RF applicator to maximize power transfer to the load.
A model of the RF heating system is shown in Figure 3.8 (a). The source consists of a
13.56 MHz oscillator followed by a 1 kW power amplifier. The amplifier is designed to drive
a 50 Ω load. The RF heating apparatus and load are generically called the load. The load
properties and conductivity of the load change with temperature. If the load is 50 Ω, then the
maximum available power Pav from the amplifier is delivered to the load. When the load
impedance is not 50 Ω, there is a mismatch loss and the total power delivered to the load is
44
PL (1- L ) Pav (W)
2
(3.12)
Z L - Z0
L ( ) (3.13)
Z L Z0
when Z L Z 0 . Under this condition, PL Pav . If the reflection coefficient is not zero because
of an impedance mismatch between the source and load, then PL Pav . The power flow
diagram of the RF heating system is shown in Figure 3.8 (b) and a transfer efficiency of
PL
Transfer efficiency ( ) 100% (3.14)
Pav
45
Load
LCC L HA
R S =50Ω
ZEQ ZHA CHA Zload Rload Cload
CCC
AC
13.56 MHz
oscillator and
power amplifier Coaxial cable(CC)
RF generator
(a)
Pav PL
Pref
(b)
Figure 3.8 RF heating system: (a) block diagram with equivalent circuit; (b) power flow diagram.
46
3.3.2 Equivalent Model for the Load
As shown in Figure 3.5, the load is a cylindrical volume sandwiched between two
circular plates. The plates have area A and are separated by distance d . With reference to
(2.11), the dielectric and conductivity characteristics of the load can be separated into three
components: 1) the real part εr of the complex permittivity, 2) the imaginary part r of the
complex permittivity, and 3) the DC conductivity DC . Each of these three parts can be
associated with an electrical model. A lossless parallel plate capacitor is created by the
r 0 A
C (F) (3.15)
d
A shunt conductor modeling dielectric losses is created by the imaginary component of the
A
GDE ( r 0 ) (S) (3.16)
d
where is the frequency of the oscillator in the heating system. A second shunt conductance
A
GDC ( DC ) (S) (3.17)
d
The admittance is complex and the real part is conductance ( GL ) and the imaginary part ( BL )
is susceptance.
The equivalent circuit for the salt water and WAS loads are shown in Figure 3.9. The
salt water load consists of two circuit elements: a capacitance created by the dielectric
47
constant of water and a shunt conductance to model the ionic conductivity of the solution.
WAS is modeled by three circuit elements: a capacitance created by the real part of the
dielectric constant, a shunt conductance associated with the imaginary part of the dielectric
constant which relates to the β dispersion characteristics of the biological material, and a
second shunt conductance corresponding to the ionic conductivity of WAS. The two
conductances in the circuit model for WAS can be combined into a single shunt conductance
(a)
GWAS Gsalt
Bsalt BWAS
(b)
Figure 3.9 Equivalent circuit representation of (a) WAS and (b) salt water solution.
As shown in (3.15) through (3.18), the electrical model is dependent on the ratio of the
area of the plates to the separation of the plate, A / d . The physical scaling ratio A / d can
therefore be used to adjust the admittance or impedance ( Z L 1/ YL ) of the load. Since the
electrical properties of the load are temperature dependent, the physical scale ratio is selected
48
to create the best matching condition at the mid-point of the temperature range. The
operating temperature range for the heating system is from room temperature (20°C) to
At 70°C, the physical scale ratio is adjusted such that the real part of the load
impedance is equal to 50 Ω. The residual imaginary part of the load impedance, which is
inductance so that the overall impedance of the load is ideally 50 Ω at a temperature of 70°C.
The design values for the plate area and plate separation are summarized in Table 3.3 for a
load volume of 400 mL. The table also includes circuit values for the load.
Table 3.3 Equivalent circuit and geometrical parameters for WAS and salt water loads at 25°C.
salt water
WAS
water
Although the design so far includes only the load, it is insightful to evaluate the
transfer efficiency at this point. The impedance and transfer efficiency of the analytic load
model is shown in Figure 3.10 and Figure 3.11. It turns out that the imaginary component of
the load impedance (reactance) is quite small relative to the real part of the load impedance
and the transfer efficiency is nearly 100% at a temperature of 70°C. Over the entire operating
49
temperature range, the power transfer efficiency is greater than 84.5%. The results confirm
that using the physical area and separation of the plates shown in Table 3.3 provide a good
120
R0.03315N
100 X0.03315N
R4.5%WAS
80 X4.5%WAS
R0.069N
Impedance ( )
60
X0.069N
40 R18%WAS
X18%WAS
20
-20
-40
20 25oC 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
100
98
Power transfer efficiency in (%)
96
94
92
90
84
20 25oC 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.11 Power transfer efficiency of salt water and WAS load as a function of temperature.
50
3.3.3 Electrical Model for the Enclosure
The coaxial load enclosure and the cable interface from the RF applicator add
additional capacitance and inductance to the system. Since the total load impedance
presented to the amplifier is measured at the connector interface to the amplifier, the
additional capacitance and inductance modifies the net load impedance. An equivalent circuit
model for the complete heating apparatus is shown in Figure 3.8 (a). The load is modeled by
a shunt resistor Rload and shunt capacitor with reactance X load . The load model corresponds to
The coaxial enclosure that encapsulates the load also has stray capacitance and series
inductance. The capacitance is primarily associated with the top plate which is electrically
isolated from the grounded outside cylinder that shields the load. The metal post which
connects the top electrode to the center conductor of the coaxial connector and the metal post
that connects the bottom electrode to ground have a small amount of inductance. The
capacitance and inductance of the enclosure are modeled by an LC section with inductance
The last section of the equivalent circuit model is a LC section which models the
capacitance and inductance of the coaxial cable which connects the amplifier to the heating
apparatus. A short length of the RG-58/U coaxial cable is used for the connection and
analytical models for the capacitance and mutual inductance of the cable are readily available
from any introductory book on electromagnetics [21]. In this design, the cable length is 30.48
cm, the inner conductor has a radius of 1.1 mm and the dielectric has a radius of 3.57 mm. A
diagram of the cable geometry is shown in Figure 3.12. The dielectric is Teflon with a
51
= 2. With these values, the capacitance and inductance of the
relative permittivity of TF
rout 3.57 mm
rin 1.1mm
(a)
30.48cm
(b)
Figure 3.12 Connecting coaxial cable geometry: (a) cross section; (b) side view.
0 rout 0
2 TF
LCX = ln( ) (nH) CCX = (pF) ZCX= LCX (Ω)
2 rin r
ln( out ) CCX
rin
All element values of the equivalent circuit model are known except for the enclosure
capacitance and inductance. There are no readily available equations to calculate these values
52
because the geometry involved is complex. An alternate way to determine the equivalent
circuit values is to find the energy stored in the electric and magnetic fields and then
The energy in the fields can be obtained from electromagnetic simulations of the
enclosure. A simulation tool called COMSOL is used for the simulating the fields in the
enclosure. Once the fields are found, the electric energy We and the magnetic energy Wm can
be found. The stored energy is partitioned between the load which corresponds to the fields
within the cylindrical load between the electrodes, and the stray fields inside the enclosure
but outside the load cylinder. If the total electric energy stored inside the heating apparatus is
We and the total electric energy stored within the load cylinder is We,load , then
1
We - We _ load CHAVm 2 (J) (3.19)
4
In this equation CHA is the effective capacitance of the enclosure (see Figure 3.8) and Vm is
In this work, there are two different physical design configurations corresponding to
the design of a heating applicator for 4.5% WAS and 18% WAS. The equivalent salt water
loads are 0.03315 N and 0.069 N, respectively, and from electromagnetic simulations using
salt water loads, the capacitance, CHA , is estimated to be 9.6 pF and 11.3 pF, respectively.
A similar calculation can be made to estimate the inductance of the enclosure. Unlike
the calculation of the equivalent capacitance of the enclosure where the load and enclosure
both have capacitance, the load does not have significant inductance and the inductance is
primarily related to the enclosure only. Recall that the load models in section 3.3 do not
include any inductive elements - only resistance and capacitance. Therefore, the total
53
magnetic energy integrated over the interior volume of the enclosure can be used to estimate
the equivalent inductance of the enclosure. The relation between magnetic energy Wm and the
1
Wm LHA I m 2 (J) (3.20)
4
where I m is the peak current of the source used in the simulation. Using this method, the
estimated inductance of the enclosure for 4.5% WAS is 117 nH and for 18% WAS is 143 nH.
The estimated values of the capacitance and inductance of the enclosure were obtained
from electromagnetic simulations using a salt water load at a temperature of 70°C. The
equivalent model values are ideally constant and independent of the load temperature.
However, the stray electric field does change slightly because the load capacity changes and
there is a small variation in the equivalent component values of the enclosure depending on
the temperature used in the simulation. The results are summarized in Table 3.5. As shown
the maximum variation in capacitance is 0.5 pF and the maximum variation in inductance is
3.7 nH. The variations are within 5% of the average values, and in the following sections, the
equivalent circuit model for the enclosure uses the average values to calculate transfer
efficiency.
Table 3.5 Intrinsic capacitance and inductance for the heating apparatus.
Load LHA (nH) LHA (nH) LHA (nH) CHA (pF) CHA (pF) CHA (pF)
54
3.4 Analytic and Simulated Results for Power Transfer Efficiency
The equivalent circuit models developed in the previous section are used to calculate
analytic values for the impedance and power transfer efficiency as a function of temperature.
In this section, the analytic results are compared with a full electromagnetic simulation of the
heating applicator.
The COMSOL models for the two different heating applicator designs are shown in
Figure 3.13. The DC conductivity is larger for the higher molarity (0.069 N) load;
consequently, the area of the electrodes is reduced and the spacing between the electrodes is
increased relative to the 0.03315 N design to maintain a nominal 50 Ω load impedance. The
temperature.
(a) (b)
Figure 3.13 COMSOL models of the heating applicator for 0.03315 N salt water (a) and 0.069 N
55
By measuring the ratio of the phasor voltage over the phasor current at the coaxial
connector interface, the equivalent impedance of the heating applicator can be determined
from the simulation. The simulation results for the two applicator designs are shown in
Figure 3.14 and Figure 3.15. The figures include the corresponding analytic results derived
from the equivalent circuit models for the enclosure and load. As shown, the simulated and
100
Analytical (resistance)
Analytical (reactance)
Simulation (resistance)
Simulation (reactance)
50
Impedance ( )
-50
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.14 Impedance of the RF heating system for a 0.03315 N salt water load.
56
150
Analytical (resistance)
Analytical (reactance)
Simulation (resistance)
Simulation (reactance)
100
Impedance ( )
50
-50
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.15 Impedance of the RF heating system for a 0.069 N salt water load.
The simulated impedance results can be used to predict the efficiency of transferring
RF generator power to the load using (3.12) through (3.14). The simulated efficiency
characteristics are shown in Figure 3.16 and Figure 3.17. Analytic model results are included
for comparison and similar to impedance, there is close agreement between simulated and
analytic results.
57
100
98
Power transfer efficiency (%)
96
94
92
90
88
86
84 Analytical
Simulation
82
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.16 Power transfer efficiency of the RF heating system for a 0.03315 N salt water load.
100
98
Power transfer efficiency (%)
96
94
92
90
88
86
Analytical
Simulation
84
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 3.17 Power transfer efficiency of the RF heating system for a 0.069 N salt water load.
The COMSOL electromagnetic simulation results can also be used to predict the
uniformity of the electric field between the electrodes. The simulator calculates the electric
field in a three dimensional model and the field variation in a YZ plane that passes through
58
the center of the load is shown in Figure 3.18. The field intensity between the electrodes is
plotted as a function of distance from the top electrode. The input power is 1 kW and the
electric field is approximately 37 V/cm for the 0.03315 N load and 26 V/cm for the 0.069 N
load. A larger field is expected for the 0.03315 N load, since the electrodes are spaced closer
than in the 0.069 N design. In both designs, the field is uniform and uniform heating is
expected.
Figure 3.18 Electric field across the load as function of distance (arc length) for (a) 0.03315 N salt
59
Chapter 4: Experimental Results
A prototype RF heating applicator was built and tested. The RF heating system
includes the RF generator, the applicator, and software to measure load temperature and
control RF generator power. The applicator was tested with salt water loads and samples of
WAS. Loads were heated from room temperature to 120ºC and impedance measurements of
the applicator and load were made to verify the design. Excellent results were obtained and
the RF heating applicator shows significant promise as a method of heating liquid and sludge
loads.
As described in Chapter 3, the RF heating system consists are three main components:
1) the RF generator, 2) the RF heating applicator, and 3) a coaxial cable to connect the RF
generator to the applicator. A software control system is also required to control the load
temperature in the applicator. An experimental test bed for heating WAS with a 4.5% bio-
solid content was implemented. Pictures of the test bed are shown Figure 4.1 and annotations
are added to identify the primary components in the system. A detailed description of the
sections.
60
Figure 4.1 Pictures of the experimental test bed: (a) RF heating system and software control; (b)
61
4.1.1 RF Generator
The RF generator is the source in the heating system and delivers the required power to
heat the load in the applicator. As described in section 2.4, a frequency of 13.56 MHz was
selected for the design based on the dielectric properties of the load and the frequency falls
amplifier module is commercially available as an evaluation unit from Directed Energy, Inc,
Colorado, model number PRF-1150. The input signal to the amplifier is a sinewave source
power amplifier module requires three external power supplies. A 5 V power supply (Agilent
6824A) is required for DC logic and control circuits in the module, a 15 V power supply
(Agilent E3632A) is required for the driver stage, and the output power stage requires a high
voltage DC supply (Ametek XFR 300-4) that can range from 100 V to 300 V. The power
amplifier, waveform signal generator, and the DC power supplies are shown in Figure 4.2.
62
5V DC power supply
Signal generator Voltmeter
PA
15V DC
power supply
300V DC power supply
R s =50Ω
AC
63
The amplitude of the RF generator is dependent on the high voltage DC supply voltage
which powers the amplifier module. Power is proportional to the square of voltage; therefore
there is ideally a square law relationship between the DC supply voltage and the power
delivered to the load. The square law relationship is ideal and does not consider the losses in
the amplifier circuit. The implication of losses is that the power efficiency of the amplifier
varies with the supply voltage which leads to a deviation from the square law relation
A short length of coaxial cable (RG-58/U) is used to connect the RF generator to the
heating applicator. A photograph of the coaxial cable is shown in Figure 4.3. The coaxial
cable has a length of 30.5 cm with two male BNC connectors attached to each end of the
cable. The output connector on the power amplifier module is a female BNC connector. The
heating applicator has a female N-type coaxial connector and an adapter (BNC to N) is used
to connect the cable to the applicator. The cable and connectors have a characteristic
impedance of 50 Ω, and if the generator and load are perfectly matched, the cable does not
64
C CX LCX
Since the load is not a constant 50 Ω impedance, the impedance presented to the RF
generator interface is modified by the coaxial cable. In Chapter 3, the cable was modeled as a
LC lumped element circuit. The equivalent circuit component values can be determined
experimentally by measuring the input impedance of the cable for different termination
Impedance measurements are made with an Agilent model 5061A vector network
65
calibrated reference plane. With reference to Figure 4.3, the equivalent circuit model for the
coaxial cable consists of a series inductor series inductor LCX and a shunt capacitor CCX . The
model is valid at low frequencies and the cable length is much less than the wavelength at
13.56 MHz. The inductance and capacitance of the cable are found by measuring the input
impedance of the cable with a short circuit termination at the other end of the cable. With a
short circuit on the right-hand interface in Figure 4.3, the input impedance looking in from
the left-hand side is a parallel resonant circuit. At low frequencies, the impedance of the
parallel resonant circuit is dominated by the inductor, while at high frequencies, the circuit is
dominated by the capacitance. There is also a resonant frequency where the reactance of the
two elements completely cancels and the input impedance is nearly zero because of the short
and at resonance ( f r )
1
Zin 0 CCX (4.2)
(2 f r )2 LCX
Input impedance measurements were made over a frequency sweep range of 300 kHz
to 500 MHz with a short circuit termination. At a low frequency ( flow ) of 300 kHz, the
extracted value of the inductance is 75 nH. The input impedance had a parallel resonance at a
frequency of 339.8 MHz from which an equivalent cable capacitance of 29.2 pF is obtained.
In Chapter 3, the analytic and modeled values for the equivalent circuit were presented, and
the cable inductance was 72 nH and the cable capacitance was 28.7 pF. The analytic and
measured values are similar, and the close agreement supports the choice of the equivalent
66
4.1.3 Heating Applicator
The electrical design of the RF heating applicator was described in Chapter 3, and for a
4.5% WAS load, the dimensions of the applicator were given in Figure 3.6. From the
electrical design, the size and spacing of the electrodes were determined to provide a good
match between the load and the RF generator for a load volume of 400 ml. The electrical
design included modeling the coaxial connector interface to the electrodes and the coaxial
Although the electrical design determined the principal geometry of the applicator,
there are also very important mechanical design requirements for the heating applicator.
Since the apparatus is designed to heat liquid and sludge loads above 100°C, the applicator
must heat the load under pressure to maintain a constant volume. In other words, the RF
applicator operates under pressure, the design requires materials, seals, and safety valves that
are consistent with a maximum operating temperature range of 120°C. The mechanical
design of the heating system was done by another member of the research team, Ehssan
Koupaie. Ehssan is a PhD student in environmental engineering and he will use the RF
The heating applicator was built in the machine shop at UBC and a picture of the final
prototype is shown in Figure 4.1. The prototype consists of three main parts: 1) a top
electrode assembly which includes an aperture for a pressure relief valve and a temperature
sensor, 2) a thick walled Teflon pressure vessel with seals and a capacity of 400 ml, and 3) a
coaxial aluminum enclosure that has the RF connector which connects to the RF generator.
More detailed photographs of the top electrode assembly are shown in Figure 4.4. The
67
electrode is an aluminum disk cut from half inch aluminum plate. The electrode has blind
holes that are used to mount the electrode to a Teflon cap and aluminum plate. The Teflon
cap matches the outside diameter of the Teflon pressure vessel and the aluminum plate is
68
Figure 4.4 Photo of the top electrode assembly in the heating apparatus.
69
The Teflon pressure vessel is shown in Figure 4.5. At the bottom of the vessel is the
sandwiched between the electrode and the Teflon vessel. At the top of the Teflon vessel, a
second O-ring seal is used to seal the top electrode to the container walls.
70
Figure 4.5 Photos of the Teflon load vessel.
71
Photographs of the aluminum enclosure are shown in Figure 4.6. Looking in from the
top of the enclosure shows a hole where the RF connector on the Teflon pressure vessel
passes through to the outside of the enclosure. Since the bottom electrode is sealed inside the
Teflon pressure vessel, the port for the connector does not require additional seals. The
enclosure is made from 8’’ aluminum pipe with a 5/16’’ wall thickness. Two flanges are
welded to the enclosure. The top flange has holes to bolt the top plate of the applicator to the
72
Figure 4.6 Photos of the aluminum enclosure.
73
After fabricating the prototype, the electrical characteristics of the applicator were
measured. As described in Chapter 3, the load in the pressure vessel can be modeled as a
shunt resistor and capacitor. The resistance and capacitance are temperature dependent and
inductance ( LHA ) and capacitance ( CHA ), and the values are measured experimentally to
compare with values obtained from electromagnetic simulations. Figure 4.7 shows the
74
Figure 4.7 Photo of the heating apparatus and the corresponding equivalent circuit model.
75
A two-step process was used to measure and extract the equivalent circuit model for
the applicator. First, an impedance measurement is made without a load in the applicator.
The load capacitance is small in this case because the dielectric is air. From this
made with a calibrated salt water load. The load was a salt water solution of 0.03315 N at a
temperature of 22.8°C, and the equivalent circuit values for the load are derived from the
Starting with the no load measurement, the capacitance is measured over a frequency
range from 10 MHz to 15 MHz. In this case, the capacitance is the sum of the capacitance of
the enclosure, CHA , and the capacitance of the load cylinder, Cload , estimated to be 0.25 pF.
The capacitance of the enclosure is then determined by subtracting the load cylinder
capacitance from the measured capacitance. The inductance, LHA , does not contribute
1
Zin (Ω) (4.3)
j 2 f (CCX Cload )
The corresponding capacitance of the enclosure extracted using this method is shown in
Figure 4.8. At a frequency of 13.56 MHz the intrinsic capacitance of the enclosure is
76
13.98
13.96
13.94
Capacitance (pF)
13.91 pF
13.92
13.9
13.88
13.86
10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15
Frequency (MHz)
Figure 4.8 Measured intrinsic capacitance of the heating applicator as a function of frequency.
Once the capacitance of the heating applicator is known, the impedance is measured
with a calibrated salt water load. Since CHA , Rload and Cload are all known, the inductance
LHA is found by tuning the impedance of the analytic model until it matches the experimental
results. A Smith chart is used to compare the experimental and model values as shown in
Figure 4.9. Over a frequency range of 10-15 MHz, the inductance of the enclosure is
estimated to be 115 nH. Note that it is difficult to see the two traces in Figure 4.9 which
shows that the modeled and measured values match very well. The extracted value of 115 nH
is very close to 117 nH which was the value derived from COMSOL simulations shown in
Chapter 3. The experimental values of the enclosure capacitance and inductance are used
later to generate power transfer models for the experimental heating apparatus.
77
Figure 4.9 Measured and modeled impedance of the heating applicator with a 0.03315 N salt water
The purpose of the RF heating system is to heat the load in a predictable and controlled
way. Therefore, temperature measurements and temperature profiles across the load are very
important.
A K-type thermocouple, model GK11M from Test Product International, Inc. is used to
measure the temperature of the load. A small hole is drilled through the top electrode and the
tip of the thermocouple extends about 1 mm into the load. Because the vessel is under
pressure, the thermocouple is sealed into the top electrode using an adhesive called JB Weld.
The location of the temperature sensor on the top electrode is shown in Figure 4.4.
A thermocouple consists of two different metals, and at the tip, the metals are joined.
Through a diffusion process, a potential difference is developed across the two wires which
can be measured with a digital voltmeter. The digital voltage meter is shown in Figure 4.2.
The voltmeter is an Agilent 34401A unit and measurements of the thermocouple voltage are
78
converts the voltage measurements into temperature and displays the temperature of the load
on a graph. The temperature of the load is calculated using the following equation
T T0 K slopeVTC (4.4)
In this equation, T0 is room temperature, K slope is temperature constant for the K type
thermocouple has an operating temperature range of -40°C to +510°C. The room temperature
electromagnetic interference (EMI). It was found that the electric fields at the surface of the
thermocouple voltage. Experiments were made to shield the thermocouple wires but the
results always had significant error. An example of the interference problem is shown in
Table 4.1 where thermocouple temperature measurements are compared with a mercury
thermometer. The errors increase as the supply voltage (electric field) increases.
Tmt (°C)
79
As a way to mitigate the sensitivity of the thermocouple to electromagnetic
the power amplifier module. The DC supply is pulsed off for 2 seconds while a temperature
measurement is made and then turns on for 10 seconds. Using this method, accurate
measurements with and without the pulsing scheme is shown in Figure 4.10.
25
Temperature (oC)
15
10
0 10 20 30 40 50 60
Time (sec)
(a)
DC power supply voltage (V)
100
50
0
0 10 20 30 40 50 60
Time (sec)
(b)
Figure 4.10 RF generator voltage (a) and load temperature (b) as a function of time.
There are two modes that will be used with the heating apparatus. The first mode heats
the material under test at a specific thermal ramp rate specified in °C per minute. The second
mode heats the material at a target steady state temperature for a specific length of time. In
both heating modes, the heating rate is controlled by the DC voltage applied to the RF power
amplifier. The output power from the amplifier is proportional to the DC voltage and, by
80
measuring the temperature of the load, software implemented in LabView can control the
Although an open loop control method can be implemented using a look-up table, there
are limitations to this method. Ideally, the power delivered to the load has a square law
relation with the power supply voltage. However, the load impedance is nonlinear which
creates deviation from the square law relation. Other significant limitations of open loop
control include increasing thermal losses as the temperature gradient between the load and
ambient temperature increases, as well as the nonlinear relationship between the power
Given the limitations of open loop temperature control, a closed loop temperature
control system is implemented. An adaptive control loop is implemented using the least mean
square algorithm. The algorithm measures the error between the desired (reference)
temperature and the actual temperature measured using the thermocouple. The power supply
voltage is then periodically adjusted in a direction dependent on the magnitude and sign of
the error. At steady state, there is a small oscillation and the magnitude of the variance at
All the software control is implemented in LabView 2013. The software includes
graphs that show the load temperature as a function of time as well as power supply control
81
Figure 4.11 LabView front panel control interface: (a) temperature graph and other control
82
4.2 Thermal Profiles of the RF Heating System
Experiments were made to verify the operation of the RF heating system. The
experiments include heating salt water and WAS at different thermal ramp rates, as well as
heating loads at a constant temperature. Experiments were also made to measure the
temperature variation across the load volume since uniform heating is important.
The first test conducted to verify heating profiles was to measure the thermal ramp rate
for a salt water load. A salt water load of 0.03315 N is used and the DC voltage was set to
four different voltages: 100 V, 150 V, 200 V, and 250 V. The salt water was initially at room
temperature, approximately 22.5°C, and heated to a target temperature of 80°C. The time
required to reach the target temperature was recorded and the corresponding thermal ramp
rate was calculated from the measurements. The experimental results are summarized in
90
80
70
Temperature (oC)
60
50
40
30
20
0 10 20 30 40 50 60
Time (minute)
Figure 4.12 Ramp rate measurement for salt water load with 250 V (blue line), 200 V (magenta
line), 150 V (red line) and 100 V (yellow line) input to the PA.
83
Table 4.2 Ramp rate of the RF heating device with 0.03315 N salt water load.
(̊C) (̊C)
For 100 V, the thermal ramp rate is 1.5°C per minute and progressively higher ramp
rates are measured for higher voltages. The highest thermal ramp rate is 13.6°C per minute
for a voltage of 250 V. The data show that the thermal ramp rate and the corresponding
voltage are not linear. For example, the thermal ramp rate increases from 1.5°C to 5.7°C for a
50 V change from 100 V to 150 V, while the thermal rate changes from 8.7°C to 13.6°C for a
50 V step from 200 V to 250 V. The nonlinear relationship between the voltage step and the
thermal ramp rate step is related to the square law relationship between voltage and power as
voltage. The amplifier is most efficient at high power (high voltage) and power efficiency
Limited samples of WAS were available for testing and, at this point in the research
project, testing was conducted for 100 V only.2 The results for heating 4.5% WAS are shown
2
Testing is conducted in a conservative manner with the least aggressive ramp rate first. This
caution was exercised as the RF amplifier can be damaged if a test did not work as expected.
84
in Figure 4.13. For comparison, the heating for an equivalent salt water load is shown. The
sludge was heated from an initial temperature of 22.2°C to a final temperature of 80°C over a
time interval of 42.5 minutes. Therefore the average ramp rate for the sludge is 1.35°C per
minute compared to 1.5°C per minute for the equivalent salt water load (see Table 4.3).
Although salt water has similar electrical characteristics compared to sludge, the physical
characteristics of sludge and salt water are different. Sludge is heterogeneous and has 4.5%
bio-solid content from organic matter, while salt water is homogenous and the thermal
characteristics of salt water are primarily determined the properties of water. The
measurements show that a slightly lower thermal ramp rate can be expected for the sludge.
Figure 4.13 Ramp rate measurement for a 4.5% WAS load with 100 V input to the PA.
85
Table 4.3 Comparison of ramp rates for 0.03315 N salt water and 4.5% WAS.
(̊C) (̊C)
water
After ramping the temperature of the load to a target temperature, the load needs to be
maintained at a constant temperature. At a constant temperature, the load loses heat to the
ambient environment. The load is contained in a Teflon vessel and the enclosure is
aluminum. Each of these materials conducts heat from the load and power must be
periodically applied to maintain a constant temperature. For this project, a steady state
temperature is obtained when the load temperature is within ± 0.5°C of the target
temperature.
the DC power supply is pulsed on at a voltage of 75 V after reaching the target temperature.
Using this scheme, the settling times for a salt water load were measured and the results are
shown in Figure 4.14. For a low thermal ramp rate of 1.5°C per minute using a 100 V supply,
the settling time is negligible. The load has been ramped up to a temperature of 120°C and
the pulsing scheme maintains a steady state temperature of 120°C. A second experiment was
run for the most aggressive thermal ramp rate (13.6°C per minute) using a 250 V supply. For
the fast ramp rate, the load is heated quickly and once the target temperature is reached there
86
is an initial overshoot in temperature. For this experiment with salt water, it took 9.53
minutes for the load to settle to the target temperature of 120 ± 0.5°C.
140
settling time= 9.53 minutes settling time=0 minute
120
tolerance
limit
100
Temperature (oC)
20
0 10 20 30 40 50 60 70 80
Time (minute)
The thermal gradient between the load and the ambient environment is large at a target
temperature of 120°C. If the target temperature is reduced, then the thermal settling time is
Figure 4.15. Similar to the experiment with a target temperature of 120°C, a low ramp rate
using 100 V does not require any settling time. For an aggressive thermal ramp rate using a
250 V supply, the thermal settling time is 6.1 minutes, approximately 3.5 minutes less than
for a 120°C target temperature. Based on these experiments, a thermal settling time of 10
minutes is recommended to cover the fastest thermal ramp rate and highest target
87
32
30 tolerance limit
settling time=0 minutes
settling time= 6.1 minutes
Temperature (oC) 28
26
24
22
20
0 2 4 6 8 10 12 14
Time (minute)
Figure 4.15 Settling time for a target temperature of 30 ̊C (blue line: 250 V and red line: 100 V).
An important objective for the RF heating system was to design a heating system
which produces uniform heating across the load. The electrical design of the apparatus
consists of two electrodes, and providing the leakage fields outside the Teflon cylinder are
The electromagnetic simulations of the electric field profile across the load were shown
earlier in section 0. If there were no thermal losses in the system, a uniform electric field
would create uniform heating and the temperature would be constant across the load.
However, there are thermal losses in the RF heating system which creates a deviation from a
uniform thermal profile. Thermal conduction losses include heat loss through the aluminum
The thermal profile of the temperature between the electrodes was measured by
inserting a mercury thermometer into the load through a small hole in the electrode. The hole
88
in the top electrode was shown earlier in Figure 4.4 and the hole is normally used for a
connection to the pressure relief valve. An illustration of the measurement method is shown
in Figure 4.16. The thermometer is inserted into the load at five different depths. The five
depths are uniformly spaced across the load in 30.15 mm increments. In order to calibrate the
Figure 4.16 Thermometer tip positions for measuring temperature across load.
89
Figure 4.17 Marks on the mercury thermometer used to set insertion depth in the load.
A large number of experiments were run with the salt water load to measure the
temperature gradient across the load. Measurements were made at steady state temperatures
from 30°C to 90°C in 10°C steps. A thermal settling time of ten minutes was used before
each measurement was made. The upper limit of 90°C is the maximum temperature before
the volume of the load expands significantly and starts to bubble through the hole in the top
electrode. For each temperature step, the thermal gradient was measured for four different
initial ramp rates using 100 V, 150 V, 200 V and 250 V. The results for each steady state
90
50
Temperature (oC)
Top heating plate Bottom heating plate
40
30
20
10
0 20 40 60 80 100 120 140
Distance across load (mm)
60
Temperature (oC)
50
40
30
20
0 20 40 60 80 100 120 140
Distance across load (mm)
Figure 4.18 Thermal profile across 0.03315 N salt water load for a target temperature of 30°C and
40°C with 250 V (blue line), 200 V (red line), 150 V (cyan line), 100 V (blue circle).
70
Temperature (oC)
60
50
40
30
0 20 40 60 80 100 120 140
Distance across load (mm)
80
Temperature (oC)
70
60
50
40
0 20 40 60 80 100 120 140
Distance across load (mm)
Figure 4.19 Thermal profile across 0.03315 N salt water load for a target temperature of 50°C and
60°C with 250 V (blue line), 200 V (red line), 150 V (cyan line), 100 V (blue circle).
91
90
Temperature (oC)
80
70
60
50
0 20 40 60 80 100 120 140
Distance across load (mm)
100
Temperature (oC)
90 11.3
mm
80
70
region of -2oC lower than target temperature
60
0 20 40 60 80 100 120 140
Distance across load (mm)
Figure 4.20 Thermal profile across 0.03315 N salt water load for a target temperature of 70°C and
80°C with 250 V (blue line), 200 V (red line), 150 V (cyan line), 100 V (blue circle).
110
105
100
Temperature (oC)
95
90
85
80
75
70
0 20 40 60 80 100 120 140
Distance across load (mm)
Figure 4.21 Thermal profile across 0.03315 N salt water load for a target temperature of 90°C with
250 V (blue line), 200 V (red line), 150 V (cyan line), 100 V (blue circle).
92
From the experiments, the following observations are made:
1. The thermal gradients are independent of the initial ramp voltage. This provides
additional confirmation that ten minutes is an adequate settling time for the RF
heating apparatus.
2. The temperature near the top electrode where the temperature sensor is located is
3. The temperature in the middle of the load is slightly higher than the target
temperature.
4. The temperature near the bottom electrode is less than the target temperature and the
deviation from the target temperature increases as the temperature increases. The
maximum and minimum temperatures measured across the load are summarized in
Table 4.4. For a target temperature of 40°C the maximum temperature deviation is
+2/-3°C and for a target temperature of 90°C the maximum temperature deviation is
+2/-11°C.
5. The thermal gradient near the bottom electrode is confined to a small region of
temperature of 80°C and with a voltage ramp of 250 V, the temperature varies from
(Table 4.4) at the surface of the bottom electrode. Over the remainder of the load
93
Table 4.4 Summary of maximum and minimum temperatures across a 0.03315 N salt water load.
30 30.8 28
40 42 37
50 52.5 46
60 62.8 55.8
70 73 64.5
80 83.1 70.8
90 92 79
conduction through the aluminum electrodes. Thermal conduction through the electrodes was
considered in the design phase and thin posts are used to connect the electrodes the electrical
source. Since the mechanical design of the top and bottom electrodes is similar, a uniform
thermal conduction gradient was expected as shown in Figure 4.22. Obviously the
measurements show an asymmetric thermal gradient in the region of near the electrodes, and
upon further analysis, the asymmetric gradient can be explained by a change in density as the
load is heated. With reference to Figure 4.23, as the load is heated, the density decreases
because of the increase in kinetic energy of the molecules in the load. The less dense
molecules float to the top of the Teflon enclosure and gravity acts on the colder molecules
that are denser. The movement of mass is a thermal convection process, and in steady state,
the temperature at the bottom of the vessel is cooler than the top of the vessel. The same
observation can be made after heating a cup of water in a microwave oven. Therefore, the
94
primary reason for the steeper temperature gradient at the bottom electrode arises from the
orientation of the experimental apparatus where the electrodes are in a vertical profile.
Improvements in the apparatus could include mechanisms to rotate or stir the load. These
Figure 4.22 Temperature gradient across the load due to heat conduction loss.
95
Figure 4.23 Temperature gradient across the load due to heat conduction loss and water density
variation.
vector network analyzer. The impedance measurements were made at different temperatures,
and from the impedance data, the transfer efficiency of power from the RF generator to the
load can be calculated using equations shown in section 3.3. The experimental results for a
A series of impedance measurements were made for a 0.03315 N salt water load over a
temperature range of 30°C to 120°C. The load was heated using two different ramp rates and
then held at a steady state temperature for ten minutes before making an impedance
96
measurement with the network analyzer. The ramp rates correspond to 100 V and 250 V
power supply voltages, and for each ramp rate, the load temperature was increased in 10°C
steps from 30°C to 120°C. At each temperature step, a dwell time of ten minutes was used
such that the load reached a steady state temperature. After reaching a steady state
temperature, the RF heating applicator was disconnected from the RF generator and
measurements were made at a frequency of 13.56 MHz. The first measurement was made at
the RF generator interface to measure the total impedance of the overall system presented to
the RF generator and the second measurement was made at the connector interface to the
applicator (see Figure 3.8). The impedance measurements for all the experiments are
100
Resistance (100 V)
Resistance (250 V)
80 Reactance (100 V)
Reactance (250 V)
60
Impedance ()
40
20
-20
30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 4.24 Measured impedance of the RF heating applicator with a 0.03315 N salt water load for
97
100
Resistance (HA including load)
Reactance (overall system)
80 Resistance (overall system)
Reactance (HA including load)
60
Impedance ()
40
20
-20
30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 4.25 Impedances for the RF applicator and the RF applicator plus the coaxial cable (overall
1. The impedance measurements for the 100 V ramp rate and the 250 V ramp rate
are very similar (see Figure 4.24). This result is expected because the load is
held at a steady state temperature for each temperature step and the ramp rate
temperature of 70°C (see Figure 4.24). The 70 °C reference point for the best
match is consistent with the design objective where the match is optimum near
that the experimental apparatus meets the original design specifications for a
apparatus and the measurements with the addition of the coaxial cable (overall
98
system) are similar. This shows that when the applicator is designed to be
the RF generator.
The impedance measurements can be used to calculate the efficiency of power transfer
from the RF generator to the load using equations shown in section 3.3. The corresponding
power transfer characteristics are shown in Figure 4.26. The power transfer efficiency is
greater than 93% over the entire operating temperature range, and peak efficiency occurs
between 60°C and 70°C. The peak efficiency point is where the load impedance is very close
to 50 Ω.
100
99
Power transfer efficiency (%)
98
97
96
95
94
93
Heating applicator
Overall system
92
30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 4.26 Measured power transfer efficiency with a 0.03315 N salt water load.
99
4.3.2 Waste Activated Sludge
The input impedance of the RF heating apparatus was measured with WAS. The load
was heated using a 100 V ramp and then held at a steady state temperature for ten minutes
measurements were made in 10°C increments over a temperature range from 30°C to 120°C.
At each temperature point the impedance was measured at both the connector reference plane
and at the RF generator plane which includes the coaxial cable. The real (resistive) and
imaginary (reactive) components of the impedance measurements are shown in Figure 4.27.
80
Resistance (HA including load)
70 Reactance (overall system)
Resistance (overall system)
60 Reactance (HA including load)
50
Impedance ()
40
30
20
10
-10
-20
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 4.27 Measured impedance for the RF heating system with a 4.5% waste activated sludge
load.
From the impedance measurements, the real part of the impedance is approximately 50
Ω at a temperature of 40°C. The design objective was to match the real part for a temperature
range of 70°C which is the midpoint of the 30°C to 120°C operating temperature range. The
difference between the measured result and the design objective can be attributed to the
variation in the dielectric properties of the sludge. The design was based on a dielectric study
100
carried out several years ago and the samples tested for this experiment were recently
obtained from the Kelowna water treatment plant. Although both samples have a bio-solid
content of 4.5% the ionic conductivity of the samples is different. Many more samples would
have to be tested to determine the standard deviation of the dielectric characteristics and this
Despite the variance in the dielectric characteristics of WAS, sludge samples can still
be heated very efficiently using the RF heating system. This is illustrated by the power
transfer efficiency graphs shown in Figure 4.28. Power efficiency from the generator to the
load is greater than 86% over the entire operating temperature range. Efficiency peaks around
40°C which corresponds to the temperature where the real part of the load impedance is 50 Ω
as shown in Figure 4.28. The results demonstrate that the sludge can be heated very
100
Heating applicator
Overall system
98
Power transfer efficiency (%)
96
94
92
90
88
86
20 30 40 50 60 70 80 90 100 110 120
Temperature (oC)
Figure 4.28 Measured power transfer efficiency for the RF heating system with a 4.5% waste
101
4.4 Comparison of Analytic, Simulated and Measured Results
objective has been to implement a system that is matched to 50 Ω. Factors which affect the
match are the load impedance, the area of the electrodes, the distance between the electrodes,
and the parasitic inductance and capacitance of the enclosure and the coaxial cable. The load
impedance is the most significant variable in the design and the load impedance varies with
temperature. The operating temperature range of the heating system is 20°C to 120°C, and
the impedance match is designed for a temperature of 70°C such that the overall power
transfer to the load remains high over the entire operating temperature range.
A summary of the impedance characteristics of the overall heating system including the
coaxial cable is shown in Figure 4.29. For the salt water load with a concentration of 0.03315
N there is very close agreement between analytic, simulated and measured results. Salt water
is a homogenous load with well controlled electrical characteristics that make it ideal for
After verifying the RF heating system with a salt water load, the system was tested
with WAS. The sludge samples were obtained from effluent at the Kelowna wastewater
treatment plant and the electrical properties of sludge are much more variable than a prepared
salt water solution. The sludge is heterogenous and has 4.5% organic solid content. The
40°C compared to the salt water load which is matched at 70°C. Since the resistance of the
sludge is lower than the salt water load, the ionic conductivity of the sludge is higher than the
102
100
Resistance of salt water (measurement)
Resistance of salt water (simulation)
Resistance of salt water (analytical)
Reactance of salt water (simulation)
Reactance of salt water (analytical)
50 Resistance of WAS (measurement)
Reactance of WAS (measurement)
Impedance ()
-50
20 40 60 80 100 120
Temperature (oC)
Figure 4.29 Impedance comparison for the RF heating system between salt water and WAS load.
The impedance characteristics shown in Figure 4.29 can be used to calculate the power
transfer efficiency from the generator to the load (Figure 4.30). When the load is not equal to
the source impedance (50 Ω) there is a mismatch loss that reduces the efficiency of
delivering power to the load. The analytic, simulated and measured power transfer
characteristics for a salt water load are similar and peak efficiency is centered at 70°C. The
power transfer efficiency curve for WAS is also shown. The efficiency peaks at 40°C and
remains above 86% efficiency over the entire operating range. Therefore, despite the shift in
peak efficiency from a 70°C target, the load can still be heated very efficiently.
103
100
98
94
92
90
88
Figure 4.30 Power transfer efficiency comparison for the RF heating system between salt water
104
Chapter 5: Conclusion
The motivation for this research project was to design an electromagnetic heating
system for treating WAS. The heating process is used as a pretreatment method to
breakdown cell walls in organic material, and after pretreatment, the organic waste is
processed in an anaerobic digester to generate bio-gas. The heating methods used in research
projects to investigate the pretreatment of sludge have so far been limited to conventional
heating process using thermal conduction and microwave heating at a frequency of 2.45
GHz. These heating processes have been selected primarily based on the convenience of
accessing commercial equipment that is readily available for conventional and microwave
heating, and there appears to be no prior work that has investigated the design of more
5.1 Contributions
The dielectric properties of WAS were measured over a wide frequency, and based on
these measurements, a frequency of 13.56 MHz was selected for heating. The dielectric
properties show that sludge has significant ionic conductivity as well as β dispersion
associated with the cell membrane. At a frequency of 13.56 MHz, WAS has significant ionic
electrical properties that can lead to efficient ohmic heating. There is also dielectric heating
associated with β dispersion, although the heating associated with this mechanism is
The efficiency of ohmic heating could be increased by reducing the frequency below
13.56 MHz but other issues including electro-chemical reactions and electrode polarization
compromise.
105
Other advantages of heating at a frequency of 13.56 MHz include uniform volumetric
heating through the load rather than conventional heating processes that rely on thermal
through a liquid medium like water is 16.6 meters and therefore, the depth of penetration of
the electric field is large. Microwave heating on the other hand has a wavelength of .092
meters in water and has a much shallower penetration depth. RF heating at 13.56 MHz also
has other advantages over microwave heating in terms of the power efficiency of the
generator. A typical domestic microwave oven has a 1.2 kW microwave tube with a power
project delivers 1 kW with a power efficiency of 85% and the design is implemented with
the power efficiency of the RF generator and higher penetration depth compared to
Once the frequency of the RF heating apparatus was selected, the impedance
characteristics of the load were known, and an applicator to uniformly heat a load was
designed. Different design concepts were evaluated using electromagnetic simulation tools
and eventually a design based on a coaxial enclosure with two electrodes was selected. The
geometry of the design was optimized to intrinsically match the load to 50 Ω generator
impedance over an operating temperature range of 20°C to 120°C. The design included
important mechanical design requirements to operate under pressure and maintain a constant
load volume.
The design was fabricated and tested. A salt water load with similar electrical
characteristic to WAS was initially used to verify the design. Analytic, simulated and
106
experimental results using the salt water load were in close agreement. Thermal ramp rates
from 1.5°C to 13.5°C were measured and measurements were also made to evaluate the
temperature across the load. Uniform heating across the load was measured and the only
significant deviation occurred near the bottom electrode. A change in density with
temperature and convection flow in the load reduces the temperature near the bottom
electrode. Methods to improve uniform heating are discussed in the next section on future
work.
After salt water load testing, the RF heating apparatus was used to heat samples of
WAS. The dielectric characteristics of sludge are much more variable than salt water, and
despite the variation, sludge was efficiently heated to a temperature of 120°C. The RF
heating apparatus will now be used by the environmental engineering group at UBC to
evaluate pretreatment methods for WAS at a frequency of 13.56 MHz, and the heating
The RF heating system is very attractive for heating electrolytic liquids and sludges. In
addition to heating, the design can operate under pressure, and there are many other potential
applications for this technology; examples include heating processes for food and biofuels,
and heating to treat organic sludges from beer and wine making processes.
There are many design improvements which could be made to improve the RF heating
system. The temperature gradient near the bottom electrode could be reduced by changing
the orientation of the applicator or incorporating a way to stir or turn the load. Other ways to
improve uniform heating could be the implementation a tapered cylinder such that the
heating at the bottom of the pressure vessel is higher to compensate for the density and
107
convection flow in the load. Larger volumes could be heated with an array of electrodes.
Separate RF generators could be used to power different sets of electrodes to improve heating
uniformity across large volumes. A dynamic matching structure with tunable inductors or
capacitors could be used to track the load impedance as a function of temperature and
maintain very high power transfer efficiency to the load. Thermal conduction losses from the
pressure vessel could be reduced by changing the electrodes from aluminum to stainless
steel. Also, insulation could be added to the air cavity in the coaxial enclosure to reduce the
thermal gradient between the pressure vessel and the ambient environment. With many
possible improvements to this design, future research projects are possible and these may
108
Bibliography
[2] L. Appels, J. Baeyens, J. Degrève and R. Dewil, "Principles and potential of the
anaerobic digestion of waste-activated sludge," Progress in Energy and Combustion
Science, vol. 34, pp. 755-781, 2008.
[8] Y. Wang, Y. Li, S. Wang, L. Zhang, M. Gao and J. Tang, "Review of dielectric
drying of foods and agricultural products," International Journal of Agricultural and
Biological Engineering, vol. 4, pp. 1-19, 2011.
[10] J. Rice, "RF technology sharpens bakery's competitive edge," in Food Proc, 1993,
pp. 18-24.
109
[13] M. Demeczky, "Continuous pasteurization of bottled fruit juices by high
frequency energy," in Proceedings of IV International Congress on Food Science and
Technology, 1974, pp. 11-20.
[15] Q. Zhong, K. Sandeep and K. Swartzel, "Continuous flow radio frequency heating
of water and carboxymethylcellulose solutions," J. Food Sci., vol. 68, pp. 217-223, 2003.
[20] A. Stogryn, "Equations for calculating the dielectric constant of saline water
(correspondence)," Microwave Theory and Techniques, IEEE Transactions On, vol. 19, pp.
733-736, 1971.
110