; hystersia-controller based energy harvesting scheme for microbal fuel cells with parallel operation capability
Learning Center
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

hystersia-controller based energy harvesting scheme for microbal fuel cells with parallel operation capability


all types of wireless sensr network papers are avalaible in the site free downlaod conference and IEE opapaers 2012

More Info
  • pg 1
									            This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON ENERGY CONVERSION                                                                                                                         1

         Hysteresis-Controller-Based Energy Harvesting
         Scheme for Microbial Fuel Cells With Parallel
                     Operation Capability
                                                 Jae-Do Park, Member, IEEE, and Zhiyong Ren

   Abstract—Microbial fuel cell (MFC) is an emerging technology                      larger MFCs or simply connecting them in series or in parallel,
for sustainable energy production. An MFC employs indigenous                         because of the nonlinear nature of MFCs [3]–[5].
microorganisms as biocatalysts and can theoretically convert any                        The power density from MFCs has increased by orders of
biodegradable substrate into electricity, making the technology
a viable solution for sustainable waste treatment or autonomous                      magnitude in less than a decade of research. The reported maxi-
power supply. However, the electric energy currently generated                       mum power density from lab scale air-cathode MFCs increased
from MFCs is not directly usable due to the low voltage and cur-                     from less than 1 mW/m2 to 6.9 W/m2 [5]–[7]. This improvement
rent output. Moreover, the output power can fluctuate significantly                    can mainly be attributed to relieving physical and chemical con-
according to the operating conditions, which makes stable harvest                    straints through electrode material and reactor architecture im-
of energy difficult. This paper presents an MFC energy harvesting
scheme using a hysteresis controller and two layers of DC/DC con-                    provement, as well as optimization of operational conditions [2],
verters. The proposed energy harvesting system can capture the                       [8]. However, the reported power output from many MFC stud-
energy from multiple MFCs at individually controlled operating                       ies is based on the power dissipated on a static external resistance
point and at the same time form the energy into a usable shape.                      instead of the actual attainable power in a usable form [1]–[4],
   Index Terms—DC/DC converter, energy harvesting, microbial                         [7], [9]–[15], which indicates one crucial missing part before the
fuel cell (MFC).                                                                     technology can be commercialized—how to efficiently convert
                                                                                     the theoretical potential into a practically meaningful power
                                                                                     output. A few energy harvesting systems for sediment MFCs
                                                                                     have been reported [16], [17]: they can capture energy from the
                            I. INTRODUCTION
                                                                                     MFC and convert it into applicable voltage and current levels.
      HE finite resource of fossil fuels and environmental pol-
T     lution derived from their use are driving the search for
renewable and clean energy alternatives. This replacement of
                                                                                     However, a control scheme that actively harvests energy at an
                                                                                     optimal operating point especially from multiple MFCs has not
                                                                                     been researched extensively.
fossil fuels will require the utilization of many energy sources                        In this paper, an efficient MFC energy harvesting system using
suited to meet different end uses. Microbial fuel cell (MFC)                         two layers of dc/dc converters is presented [34]. The proposed
technology has been intensively researched in recent years as                        system can capture the energy from multiple MFCs at individu-
a novel technology, because it offers a solution for environ-                        ally controlled operating points and at the same time forms the
mentally sustainable energy by treating waste and recovering                         energy into a usable shape.
electricity simultaneously. MFCs use active bacteria to gener-
ate electrical energy from the environment electrochemically.
MFCs offer a simple, direct method for converting environ-                                                             II. MFCS
mentally available biomass into electricity and are very suit-                       A. Characterization of MFC
able for clean, distributed, and renewable energy source, for
                                                                                        MFCs use electrochemically active bacteria at the anode to
example, powering the remote sensors [1], [2]. However, like
                                                                                     catalyze the conversion of chemical energy stored in biodegrad-
other microenergy sources such as ambient heat, vibrations, and
                                                                                     able substrate into electricity. In a typical two-chamber system
lights, MFC reactors generate very low power and energy due to
                                                                                     in Fig. 1, the anode and cathode compartments separated by
thermodynamic limitations and it has been reported that larger
                                                                                     an ion-exchange membrane. Electrochemically active anaero-
power production cannot be easily achieved by just building
                                                                                     bic or facultative bacteria extracts electrons from the electron
                                                                                     donor and transfers them to the anode electrode. These elec-
                                                                                     trons flow from the anode through an external circuit to the
                                                                                     cathode, where they reduce an electron acceptor such as oxy-
   Manuscript received November 5, 2011; revised January 26, 2012; accepted          gen or ferricyanide. Protons are exchanged from the anode to
March 19, 2012. Paper no. TEC-00591-2011.                                            the cathode and participate in the oxidant reduction [2]. In a
   J.-D. Park is with the Department of Electrical Engineering, University of
Colorado Denver, Denver, CO 80204 USA (e-mail: jaedo.park@ucdenver.edu).             single-chamber air-cathode MFC, the ion exchange membrane
   Z. Ren is with the Department of Civil Engineering, University of Colorado        has been removed and constructs a cathode structure for open
Denver, Denver, CO 80204 USA (e-mail: Jason.Ren@ucdenver.edu).                       air diffusion [9]. Lab-scale MFC reactors are shown in Fig. 2.
   Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.                                                          An MFC can be treated as a weak voltage source because the
   Digital Object Identifier 10.1109/TEC.2012.2196044                                 output voltage does not remain constant as the current output
                                                                  0885-8969/$31.00 © 2012 IEEE
           This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2                                                                                                                  IEEE TRANSACTIONS ON ENERGY CONVERSION

Fig. 1.   Schematic of a two chamber MFC using O2 as electron acceptor.

                                                                                     Fig. 4. Polarization curves of MFCs used in the experiment. MFC#1 and
                                                                                     MFC#2 denote two-chamber ferricyanide cathode MFC and single-chamber air
                                                                                     cathode MFC, respectively. Points A and B are the operating points determined
                                                                                     by hysteresis controller. Double arrows denote MFC voltage, current, and power
                                                                                     in harvesting operation band.

                                                                                     is used as the terminal electron acceptor. The anode potential of
                                                                                     the two-chamber MFCs with ferricyanide cathodes is same as
                                                                                     that of single-chamber MFCs. Typically, the working potential
                                                                                     of ferricyanide cathode is +0.6 V and it is determined by the
Fig. 2. Lab scale two-chamber MFC using ferricyanide as electron acceptor            redox potential of ferricynaide. The thermodynamic limitation
(MFC#1, left) and single-chamber MFC using air as electron acceptor (MFC#2,          determines the voltage generally less than 0.8 V and the current
right).                                                                              output in the range of a few milliamperes, which cannot be used
                                                                                     directly in most real-world applications [10].
                                                                                        The typical method for static characterization of MFC power
                                                                                     production is operating the MFC with a series of external resistor
                                                                                     Rext between the anode and cathode, and monitoring voltage
                                                                                     across the resistor continuously to obtain a polarization data.
                                                                                     The interval to change the Rext is 5–30 min depending on
                                                                                     MFC condition. This could be done either manually or using a
                                                                                     potentiostat controller. The MFC reactors under test have shown
                                                                                     a clear activation characteristics but the voltage has not dropped
                                                                                     much in concentration region. It can be seen that the MFC output
                                                                                     voltage in the ohmic region is practically inversely proportional
Fig. 3.   Equivalent circuit of an MFC.
                                                                                     to the output current. The maximum power point (MPP) can be
                                                                                     defined by respective voltage and current, where the maximum
increases. It can be electrically modeled as a voltage source and a                  power is delivered by the MFC system. It can be shown that
resistance as can be seen in Fig. 3. The MFC internal resistance                     this operating point occurs when Rext equals Rint [12], [15].
Rint is the sum of system ohmic resistance, charge transfer                          Although the OCV of an MFC can reach as high as 0.8 V,
resistance, and activation resistance. It has been reported that the                 the actual voltage at MPP is much lower as can be seen in
internal resistance is reasonably constant in the ohmic region for                   Fig. 4, which makes the direct use of MFC voltage output more
given reactor parameters [6], [8]. The thermodynamic voltage                         difficult. Fig. 4 shows the polarization curves for the MFCs used
vint can vary nonlinearly as MFC condition changes. Possible                         for energy harvesting experiment in this paper.
causes include instantaneous output power level, accumulated                            The dynamic response of MFC reactor has also been tested
extracted energy, bacteria community and activity shifts, and                        with MFC#1 reactor. The output terminal of the reactor has been
environmental condition changes.                                                     short-circuited by a MOSFET switch with a constant 50% duty
   The potential difference between anode and cathode when the                       ratio. The MFC generates its maximum current that the reactor
circuit is open is called open-circuit voltage (OCV). The OCV                        condition permits. It turned out that the instantaneous voltage
of an air-cathode MFC is generally less than 0.8 V, because the                      and current response is quite fast, as can be seen in Fig. 5. The
MFC anode potential is around −0.3 V (versus normal hydrogen                         current output has about 2-μs delay, 4-μs rising and falling time
electrode), which is set by the respiratory enzymes of bacteria,                     for step-wise energy extraction and recovery, and 3 × 10−6 time
and the working cathode potential is around +0.5 V when oxygen                       constant τi for steady state. The test has been performed with
           This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PARK AND REN: HYSTERESIS-CONTROLLER-BASED ENERGY HARVESTING SCHEME FOR MICROBIAL FUEL CELLS                                                                     3

                                                                                                                 TABLE I
                                                                                     POLARIZATION TEST DATA OF MFC#1 AND ESTIMATED INTERNAL RESISTANCE
                                                                                              R int AT THERMODYNAMIC VOLTAGE v int AROUND MPP

                                                                                    recovered after energy extraction decreases as microbial system
                                                                                    continuously generates power, which in turn reduces the output
                                                                                    current. The time constant for this voltage decrease is consider-
                                                                                    ably slower compared to the current dynamics. For continuous
                                                                                    power extraction, the average thermodynamic voltage of MFC
                                                                                    can be given as follows, where Vint0 is the initial thermodynamic
Fig. 5. MFC#1 output voltage and current to step-wise energy extraction with        voltage and τv (t) is a time constant for decaying voltage:
50% duty ratio and 12.5-kHz switching. From top, MOSFET switch state (0:
OFF, 1: ON), MFC voltage, and MFC current.
                                                                                                            vint (t) = Vint0 (e−t/τ v (t) ).                  (2)

                                                                                    The time constant τv is a function of biochemical factors and
                                                                                    electrical load as well as the thermodynamic voltage vint .

                                                                                    B. Electricity Generation Using MFC
                                                                                       The instantaneous power output of an MFC reactor Po , which
                                                                                    can be measured across Rext , is inversely proportional to the
                                                                                    total system resistance squared as follows:

                                                                                                                         vint (t)2 Rext
                                                                                                             Po (t) =                                         (3)
                                                                                                                        (Rint + Rext )2
                                                                                    where vint is the thermodynamic voltage of the MFC. The output
                                                                                    power is also in proportion to the square of the thermodynamic
                                                                                    voltage, which is slowly varying according to load and reactor
                                                                                    conditions. The internal resistance Rint at the MPP can be esti-
                                                                                    mated from the polarization curve using the fact that for a given
                                                                                    voltage, the maximum power is generated when Rint and Rext
Fig. 6. MFC#1 output voltage and current to step-wise energy extraction
with 50% duty ratio with 10-kHz switching. From top, MFC voltage and MFC
                                                                                    is same. The thermodynamic voltage vint at the MPP can also
current.                                                                            be estimated using Rint and measured current Io . The internal
                                                                                    resistance in other operating points can be roughly estimated
                                                                                    using the slope on the polarization curve or measured accu-
different switching frequencies and slightly different reactor                      rately using electrochemical analysis, such as electrochemical
conditions, and results were similar. Because the time constant                     impedance spectroscopy [8], [18]. Estimated values of Rint and
is small enough, the output dynamics of MFC itself is negligible                    vint from the MFC#1 polarization test data are shown in Table I.
so that it can be modeled as resistive circuit. The instantaneous                      The power measurement on static Rext on MFC output can
MFC output current in generation mode can be given as follows:                      simulate the MFC power output to a load, but the generated
           iM FC (t) = IM FC (t)(1 − e−t/τ i ) ≈ IM FC (t)                  (1)     power is dissipated as heat instead of being utilized to support the
                                                                                    load. Although the resistors make it straightforward to measure
where IM FC is the current magnitude determined by the instanta-                    the MFC’s power generation, this scheme cannot be used for
neous thermodynamic voltage and internal/external impedance.                        practical purpose. For efficient harvesting and usage of the MFC
                                                                                    energy, a power conversion circuitry is indispensable to capture
   When the MOSFET is OFF, the MFC terminals are open                               the electrical energy from MFCs and shape it into a usable form.
and the voltage across terminals represents the OCV, which is                       DC/DC switching converters can be used and this can be defined
the thermodynamic voltage. It can be seen in Fig. 5 that the                        as “active” harvesting compared to the power dissipation on
voltage recovers instantaneously, but the amplitude is slightly                     resistance because the power converters actively extract energy
smaller than before extraction. Fig. 6 shows the MFC voltage                        from MFC by high-frequency switching action.
           This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4                                                                                                                  IEEE TRANSACTIONS ON ENERGY CONVERSION

Fig. 7.   Block diagram of the proposed MFC energy harvesting system.

   In order to make the MFC technology more applicable, fol-                         the external resistance should be removed in order to practically
lowing challenges need to be addressed:                                              utilize the generated energy.
   1) an efficient real-time control scheme without using static                         Energy management systems have been developed especially
      resistance to capture and provide the usable energy from                       for ocean sediment MFCs to power remote sensors and wire-
      MFCs using power electronics converters;                                       less transmitters for naval applications [17], [21], [22]. Given
   2) flexible parallel operation of harvesting systems for mul-                      the low energy output from sediment MFCs, intermittent opera-
      tiple MFCs to overcome the difficulty of increasing power                       tion rather than continuous harvesting has been suggested [17].
      and energy with stacked MFCs;                                                  Although the reported systems have supported actual electrical
   3) a real-time controller to maintain the operation of MFC at                     loads, mostly they used passive harvesting techniques to har-
      an optimal operating point.                                                    vest the energy from MFCs using capacitors or charge pumps.
   In this paper, the first two challenges have been addressed.                       Those passive devices do not have proper control over the op-
The proposed scheme can harvest energy from multiple MFCs                            erating point of MFCs. Hence, the system performance can be
at individually controlled operating point (e.g., MPP) and shape                     improved if MFCs can be controlled to operate at the most ef-
the harvesting energy to a usable form. The MPP tracking ca-                         ficient operating point. Moreover, an effective way to increase
pability has not been included in this paper, but can be readily                     the power/energy capacity of MFC system has not been inves-
integrated into the proposed harvesting controller [19].                             tigated.
                                                                                        Other proposed approaches include multiunit optimization
           III. MFC ENERGY HARVESTING TECHNIQUES                                     and detailed mathematical model-based approach [13], [14].

A. Current Techniques
   Research on MFC energy harvesting has been focused on the                         B. Proposed Harvesting System
following areas:                                                                        A harvesting system for maximizing energy recovery from
   1) increasing output voltage and power capacity;                                  multiple MFCs is proposed in this paper. The system consists
   2) optimizing external resistance Rext to find and keep the                        of two layers of DC/DC converters. The first converter harvests
       operating point at the MPP;                                                   the energy from an MFC and charges the storage capacitor, and
   3) development of energy management system to utilize en-                         the second layer converter boosts the voltage to an appropriate
       ergy harvested from MFCs for devices such as remote                           level for the connected load. Instead of a capacitor-based charge
       sensors and transmitters.                                                     pump that is not designed for linear voltage regulation [23],
   Some researchers tried to achieve a larger power from bigger                      [24], an inductor-based converter has been utilized for more
MFC or multiple interconnected MFCs. However, the amount                             controllability on MFC operation. Compared to a single-layer
of energy generated by MFCs is not a linear function of their                        system, which uses a single converter for capturing energy and
size; thus, the power density will not remain constant with the                      supporting the load, this double-layer configuration can achieve
increased electrode surface area. Stacks of MFCs are not op-                         better performance by doing energy harvesting and load support
erating as same as batteries, and the performance of the stack                       in separate subsystems. Fig. 7 shows the block diagram of the
is limited by the worst performing unit because of the voltage                       overall system.
reversal [3], [20].                                                                     The MFC operation voltage can be determined by the polar-
   Popular maximum power point tracking (MPPT) techniques                            ization curve assuming reasonably stable condition. Once the
such as perturbation and observation or gradient method for                          operating point is determined, the proposed real-time operating
photovoltaic systems and hydrogen fuel cells have been intro-                        point controller keeps the MFC output voltage in the determined
duced to MFC systems to find the optimal value of external                            operating voltage band. The MFC thermodynamic voltage that
resistance [11], [12]. Although it is important to identify the                      has been explained in Section II-A determines the dynamics of
steady-state operating condition for maximum power output,                           MFC output voltage and current, i.e., rate of change, with the
          This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PARK AND REN: HYSTERESIS-CONTROLLER-BASED ENERGY HARVESTING SCHEME FOR MICROBIAL FUEL CELLS                                                                              5

inductance in the power interface circuit. And, these dv/dt and                    point A. The MOSFET is ON and the MFC voltage decreases
di/dt determines the switching period and duty ratio.                              as current increases. As the MFC voltage decreases to the oper-
   The hysteresis-controller-based energy harvesting system                        ating point B, MOSFET turns OFF and the inductor energy is
controls the operating voltage by switching the harvesting con-                    discharged to capacitor. In this DISCHARGE mode, the voltage
verter MOSFET Q1 . Unlike the standard boost converter that                        increases to operating point A as current decrease. It can be
controls the output voltage, the proposed control scheme con-                      seen that the voltage is confined between points A and B and
trols the power extraction at input side, i.e., the Q1 switching                   the MFC can operate around the MPP as shown with double
frequency and duty ratio, according to the MFC’s condition to                      arrows.
maintain the MFC voltage at a predefined range and ensures                             During the CHARGE mode, the harvesting converter’s MOS-
enough recovery time of the MFC reactor. This scheme is ef-                        FET Q1 is ON, and the energy is extracted from the MFC and
fective especially when the MFC thermodynamic voltage drops                        stored in the inductor L1 . The MFC terminal voltage vM FC
significantly as output current increases. The second layer has a                   in this mode decreases due to the increasing current. Assum-
standard DC/DC boost converter that amplifies the output volt-                      ing negligible inductor resistance and constant thermodynamic
age to an appropriate level for powering the external electronic                   voltage Vint , the instantaneous MFC output voltage and current
device(s).                                                                         in a CHARGE period can be given as follows, where IoC is the
   The inductance can be determined with following equation                        initial inductor current when MOSFET Q1 is closed:
[25], assuming stiff response of MFC generation and edge of
                                                                                               vM FC (t) = vint (t) − Rint iM FC (t)                                (5)
continuous conduction:
                              VM FC Ts D                                                                       1
                       L1 =                                  (4)                               iM FC (t) =             vM FC (t)dt                                  (6)
                                2IM FC                                                                         L1
where VM FC is the average MFC output voltage in the hysteresis                                                           Vint          R int           Vint
                                                                                                           =     IoC −             e−    L1     t
                                                                                                                                                    +        .      (7)
voltage band, IM FC is the average output current, and Ts and                                                             Rint                          Rint
D are the average switching period and duty ratio, respectively.
                                                                                      It can be seen that the current would be increasing to a level
However, it should be noted that the switching period will vary
                                                                                   determined by the thermodynamic voltage and the resistance,
as the MFC condition changes as well as the duty ratio. For
                                                                                   which is the maximum generatable current in the polarization
example, the harvesting system does not switch to CHARGE
                                                                                   curve in Fig. 4. However, the energy harvesting controller keeps
mode from DISCHARGE if the MFC voltage does not recover
                                                                                   the current at the level of the specified operating point. The
to the upper threshold voltage due to reduced substrate concen-
                                                                                   external inductance and the internal resistance determine the
tration or microbial activity. This will decrease the switching
                                                                                   changing rate of voltage and current, which in turn determines
frequency and duty ratio. The diode is reverse biased and the
                                                                                   the energy extracting frequency.
converter operates in discontinuous conduction mode in this
                                                                                      During the DISCHARGE mode, the MOSFET switch Q1 is
case. Fast recovery by strong microbial activity will increase
                                                                                   OFF, and the energy stored in the external inductor L1 is dis-
switching frequency and duty ratio. The mode changes from
                                                                                   charged to the storage capacitor C1 . The MFC voltage increases
CHARGE to DISCHARGE similarly.
                                                                                   in this mode as current decreases. The MFC output current and
   The harvesting controller in the first layer captures the energy
                                                                                   the storage capacitor voltage in the DISCHARGE mode can be
from MFC in power mode; in other words, it injects the current to
                                                                                   given as follows:
the reservoir capacitor C1 ; hence the C1 voltage is not controlled
and it is a function of harvested power and load power. Larger                                      1
                                                                                    iM FC (t) =             (vint (t) − Rint i(t) − vo1 (t))dt                      (8)
capacitance can be used for longer power supply for the energy                                      L1
storage capacitor, but it takes longer to charge it up. On the other
                                                                                                               Vint − vo1 (t)           R int           Vint − vo1 (t)
hand, smaller capacitors are charged faster, but they cannot stand                              =     IoD −                        e−    L1     t
                                                                                                                    Rint                                     Rint
longer. The voltage on C1 affects the efficiency of first and
second layer converters because too high boost ratio reduces                                                                                                        (9)
the boost conversion efficiency [25]. Hence, the selection of                                        1
capacitor C1 should be application dependent and the tradeoff                          vo1 (t) =            (i1 (t) − i2 (t))dt.                                   (10)
between charge time, support time, and converter efficiency
needs to be considered.                                                               In (8) and (9), IoD and i2 is the inductor current when MOS-
   The design of the secondary boost controller can follow the                     FET Q1 is open and the load current drawn from the storage
standard design procedure [25].                                                    capacitor C1 by the second layer voltage boost converter, respec-
                                                                                   tively. The MFC output voltage in the DISCHARGE mode can
C. Operation of the Proposed Harvesting System
                                                                                   be given as same as (6). The instantaneous current is a function
   The operation of the energy harvester in the first layer con-                    of thermodynamic voltage, internal resistance, external induc-
sists of two modes, CHARGE and DISCHARGE, according to                             tance, external capacitance, and load current in DISCHARGE
the energy flow on the inductor connected to the MFC. The op-                       mode.
eration can be explained with the polarization curve in Fig. 4.                       The hysteresis controller turns OFF the MOSFET Q1 auto-
The CHARGE mode starts when the MFC reaches the operating                          matically when the MFC voltage reaches lower threshold in the
           This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6                                                                                                                  IEEE TRANSACTIONS ON ENERGY CONVERSION

                                                                                     D. Efficiency Calculation
                                                                                        The efficiency of the energy harvester can be given as
                                                                                                                    Po    Pin − Ploss
                                                                                                              η=        =             .                       (13)
                                                                                                                    Pin       Pin
                                                                                     From Fig. 7, the power loss in CHARGE and DISCHARGE
                                                                                     mode are as follows:
                                                                                            PlossC = (PlossL + PlossQ1 ) ×                                    (14)
                                                                                            PlossD = (PlossL + PlossD1 + PlossCap ) ×                       (15)
Fig. 8.   Schematic of the proposed harvesting converter controller.                                                                                  TSW
                                                                                     where PlossC and PlossD are loss in CHARGE and DISCHARGE
                                                                                     modes, respectively. And PlossL , PlossQ1 , PlossD1 , and PlossCap
                                                                                     are loss on inductors, Q1 and D1 . TCHG , TDISCHG , and TSW are
                                                                                     the time periods for CHARGE, DISCHARGE, and a switching
                                                                                     cycle, respectively.
                                                                                        Efficiency can be calculated using this loss model:
                                                                                                 VM PP IM PP − PlossC − PlossD − Punm o deled
                                                                                           η=                                                                 (16)
                                                                                                                VM PP IM PP
                                                                                     where VM PP and IM PP are average MFC voltage and current
                                                                                     at MPP, respectively, and Punm o deled is a unmodeled miscella-
                                                                                     neous losses including switching loss.

                                                                                     E. Parallel Operation
                                                                                        Parallel operation will be a viable option to increase the
                                                                                     capacity of MFC-based power system because the direct se-
                                                                                     ries/parallel connection has difficulties in increasing power and
                                                                                     energy capacity. Although there are number of MFC applica-
Fig. 9.   Simulation: MFC#1 operation at measured MPP.                               tions that can be readily operated in parallel for larger power and
                                                                                     energy capacity, harvesting systems to provide flexible control
                                                                                     over such systems have not been investigated. In this paper, a
CHARGE mode and turns it back ON when the MFC voltage                                multiple-input converter topology is used. The multiple-input
reaches the upper threshold in DISCHARGE mode. Because                               converters have recently been used for applications such as hy-
the MOSFET is ON when the output voltage is lower than the                           brid vehicles, photovoltaics, and wind power systems [26]–[28],
CHARGE threshold VthH and OFF when it is higher than the                             but it has not been applied to MFC energy harvesting systems.
DISCHARGE threshold VthL , a logic inverter is added to the                             The advantage of the proposed scheme is apparent with
comparator-based hysteresis controller. The threshold voltage                        multiple-reactor operation. The number of converters will be
toggles as operating mode changes. The upper and lower volt-                         N + 1 because N reactors can share one boost converter in the
age thresholds can be determined as follows and easily changed                       second layer. The proposed scheme can achieve better harvest-
using potentiometers:                                                                ing efficiency because it actively extracts the energy contin-
                                                                                     uously at individually controlled operating points, while it is
                                             R2                                      difficult to continuously harvest the energy and maintain output
                   VthH = Vcc ×                                             (11)     voltage at the same time with single converter configuration.
                                       R2 + (R1 //R3 )
                                                                                     Typically, a capacitor or a charge pump is used for single con-
                                          R2 //R3                                    verter system, but the operating point changes as capacitor volt-
                    VthL = Vcc ×                       .                    (12)
                                       R1 + (R2 //R3 )                               age changes because they just passively takes the power from
                                                                                     MFC, which leads to a low harvesting efficiency.
   The duty ratio and switching frequency can be controlled by                          It is straightforward to operate multiple MFCs in parallel us-
the hysteresis voltage band, VthH − VthL , and they will vary                        ing the proposed controller. As can be seen in the block diagram
depending on the generating capacity and recovery time of the                        in Fig. 10, only the harvesting controllers will be added for
operating MFC. The schematic of the proposed hysteresis har-                         multiple MFCs in parallel operation. Each harvesting controller
vesting controller is shown in Fig. 8. Simulated MFC output                          operates independently with separate voltage thresholds based
voltage and current using the proposed hysteresis controller can                     on its MFC’s maximum power operating point. The harvesting
be seen in Fig. 9.                                                                   controllers share a storage capacitor to put captured energy into
            This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PARK AND REN: HYSTERESIS-CONTROLLER-BASED ENERGY HARVESTING SCHEME FOR MICROBIAL FUEL CELLS                                                                     7

Fig. 10.   Parallel operation with multiple MFCs and a boost converter.

and a single boost converter generates increased voltage to sup-                     Fig. 11. Energy harvesting experiment setup for two MFCs. The experimen-
port the load. The output of the second layer boost converter can                    tal harvesting system consists of two hysteresis controller controlled energy
                                                                                     harvesters and a voltage boost converter.
be given as
                         vo2 = vo1 ×                                       (17)
                                           1 − DQ 2
where DQ 2 is the duty ratio of MOSFET Q2 .

                      IV. EXPERIMENTAL RESULT
A. Energy Harvesting From Multiple MFCs
   For the energy harvesting experiment, a prototype controller
has been implemented for two different MFCs in Fig. 2. It con-
sists of two harvesting controllers and one boost converter. The
controller uses Vishay MOSFET SI3460 and National Semi-
conductor’s comparator LMC7215 for low conduction resis-
tance and low power consumption, respectively. Also, a Schot-
tky diode 1N5711 has been used. For the energy harvesting
converter, a 14-mH inductor CST206-1A has been used for the
plots in this paper.
   The inductance is initially determined using (4) with 3 kHz
of switching and duty ratio 0.5 to match with the off-the-shelf                      Fig. 12. Charging/discharging cycles from a two-chamber ferricyanide-
                                                                                     cathode MFC#1 controlled by the harvesting system From top, switch state,
inductor available, but it should be noted that system opera-                        MFC output voltage, MFC output current.
tion and performance can be changing from calculation because
the operating parameters of MFC will vary according to factors
such as bacteria community and activity shifts, and environmen-                      connected MFC at its MPP independently and extract the power
tal condition changes. Although a smaller inductance makes the                       that is measured by the polarization test. However, the operating
switching frequency faster and voltage/current ripple smaller, it                    condition and generation capacity depends on microbial activity
could require too small hysteresis band and may not be able to                       and tends to vary especially in small reactors. The relation be-
trigger the switching action because of the small energy extrac-                     tween MFC bacteria activity and electrical energy extraction has
tion. Two 2.5-V 1-F supercapacitors have been used in parallel                       been researched and MPPT control can be one of the solutions to
for energy storage. The developed MFC energy harvesting sys-                         maintain the operating condition in changing environment [19].
tem is shown in Fig. 11.                                                                Fig. 14 shows the carrier wave and output of the second
   Figs. 12 and 13 show the typical operation cycles of the                          layer boost converter. A carrier wave and gating signal generator
harvesting system with MFC#1 and MFC#2, respectively. The                            circuitry for the second layer boost converter is shown in Fig. 15.
chambers of MFC#1 have a working volume of 48 mL, and                                The duty ratio of this boost converter is set manually using
ferricyanide was used as the electron acceptor to provide sta-                       potentiometer R6 in this experiment. The output was boosted to
ble cathode potential. MFC#2 is a single-chamber reactor with                        3.3 V at 1-kΩ load using the energy supplied by two MFCs. A
a working volume 250 mL and an air-cathode. For this ex-                             PI controller can be readily added for constant output voltage
periment, the MPP for each MFC has been obtained from the                            control. If the load demands more than supply power, the storage
polarization test, which are around 325 and 300 mV for MFC#1                         capacitor voltage decreases and in turn the duty ratio of the
and MFC#2, respectively. The hysteresis voltage band is set as                       second layer boost converter will increase to maintain a constant
20 mV. It can be seen that each harvesting controller operates                       output voltage. The operation of the second layer converter can
          This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8                                                                                                                 IEEE TRANSACTIONS ON ENERGY CONVERSION

                                                                                    Fig. 15. Carrier wave and reference signal generator for second layer DC/DC

                                                                                                                   TABLE II
                                                                                               EFFICIENCY CALCULATION DATA AND LOSS BREAKDOWN

Fig. 13. Charging/discharging cycles from a single-chamber air-cathode
MFC#2 controlled by the harvesting system. From top, switch state, MFC
output voltage, MFC output current.

                                                                                    Apparently, the diode drop is too high for such a low-voltage,
                                                                                    low-power application. However, the synchronous rectification
                                                                                    technique can improve the efficiency by replacing the diode with
                                                                                    a low on-loss MOSFET, although the control circuitry has to be
                                                                                    carefully designed for the issues such as reverse power flow pre-
                                                                                    vention and floating gate drive [29]–[31]. Also selecting lower
Fig. 14. Secondary layer DC/DC operation. From top, carrier wave, switching         loss components, e.g., ones with lower RDS(on) , dc resistance,
state, input and output voltage of the converter.
                                                                                    and equivalent series resistance, and implementing circuitry in
                                                                                    low-power application-specified integrated circuit (ASIC) [32],
be paused in this case because the efficiency drops significantly                     [33] would be feasible ways to improve efficiency.
with high boost ratio.
                                                                                                                   V. DISCUSSION
B. Efficiency                                                                           The result of the experiments in this paper have confirmed
                                                                                    the following:
   The energy harvesting controller has been tested for efficiency
                                                                                      1) MFC operating point control capability for energy capture
by charging a 3-F capacitor from 0 to 500 mV. It took 36 min
                                                                                          using a simple hysteresis controller;
with one MFC#1 type two-chamber reactor. The average switch-
                                                                                      2) parallel operation of multiple MFC reactors in power
ing frequency and duty ratio of the switch Q1 depends on the
MFC’s generation response and voltage recovery in conjunction
                                                                                      3) output voltage boost to a practically usable level using the
with the external power interface. The measured efficiency of
                                                                                          harvested energy.
the harvesting system is 45.21% and breakdown of the loss cal-
culated using the data from the polarization test and datasheet
                                                                                    A. Controllability
of the parts used in the control system are shown in Table II.
It can be seen that the power loss due to diode forward drop                           The control approach proposed in this paper enables the op-
is dominant (93%) in overall loss and causes low efficiency.                         erating point control. MFCs lost more than 50% of produced
           This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PARK AND REN: HYSTERESIS-CONTROLLER-BASED ENERGY HARVESTING SCHEME FOR MICROBIAL FUEL CELLS                                                                         9

power across the internal resistance if the operating voltage is                     [2] B. Logan and J. Regan, “Electricity-producing bacterial communities in
higher MPP voltage [11]. Other energy extracting techniques                              microbial fuel cells,” Trends Microbiol., vol. 14, no. 12, pp. 512–518,
                                                                                         Dec. 2006.
such as resistors or charge pumps cannot sustain the operating                       [3] P. Aelterman, R. Korneel, H. Pham, N. Boon, and W. Verstraete, “Con-
point at a desirable voltage level, even though the persistant                           tinuous electricity generation at high voltages and currents using stacked
operation at the MPP can deliver stable performance [13]. The                            microbial fuel cells,” Environ. Sci. Technol., vol. 40, no. 10, pp. 3388–
                                                                                         3394, 2006.
proposed control approach enables flexible energy capture based                       [4] A. Dewan, H. Beyenal, and Z. Lewandowski, “Scaling up microbial fuel
on the MFC condition at a controlled operating point without                             cells,” Environ. Sci. Technol., vol. 42, pp. 7643–7648, 2008.
complex control circuitry.                                                           [5] B. Logan, “Scaling up microbial fuel cells and other bioelectrochemical
                                                                                         systems,” Appl. Microbiol. Biotechnol., vol. 85, pp. 1665–1671, 2010.
                                                                                     [6] Y. Fan, E. Sharbrough, and H. Liu, “Quantification of the internal resis-
B. Scalability                                                                           tance distribution of microbial fuel cells,” Environ. Sci. Technol., vol. 42,
                                                                                         no. 21, pp. 304–310, 2008.
   The proposed control scheme can flexibly integrate multiple                        [7] B. Min, S. Cheng, and B. Logan, “Electricity generation using membrane
reactors into a single generation system and increase genera-                            and salt bridge microbial fuel cells,” Water Res., vol. 39, no. 9, pp. 1675–
                                                                                         1686, 2005.
tion capacity without affecting overall configuration because                         [8] Z. He, N. Wagner, S. Minteer, and L. Angenent, “An upflow microbial
each reactor is controlled independently as a current source.                            fuel cell with an interior cathode: Assessment of the internal resistance by
Even if some units are disconnected, the final output voltage                             impedance spectroscopy,” Environ. Sci. Technol., vol. 40, pp. 512–518,
can be maintained at a predefined level by the secondary boost                        [9] H. Liu and B. Logan, “Electricity generation using an air-cathode single
controller and the output setting is not affected by number of                           chamber microbial fuel cell in the presence and absence of a proton
reactors or their operating conditions. If the load can be operated                      exchange membrane,” Environ. Sci. Technol., vol. 38, pp. 4040–4046,
with a burst-type power [16], it can be supported without the                                                                              o
                                                                                    [10] B. Logan, B. Hamelers, U. Rozendal, R.and Schr¨ der, J. Keller, S. Freguia,
secondary voltage booster to improve overall efficiency.                                  P. Aelterman, W. Verstraete, and K. Rabaey, “Microbial fuel cells: method-
                                                                                         ology and technology,” Environ. Sci. Technol., vol. 40, no. 17, pp. 5181–
                                                                                         5192, 2006.
C. Self-Sustainability                                                              [11] L. Woodward, M. Perrier, and B. Srinivasan, “Comparison of real-time
                                                                                         methods for maximizing power output in microbial fuel cells,” Amer. Inst.
   One of the major challenges with use of MFCs is that MFCs                             Chem. Eng. J., vol. 56, no. 10, pp. 2742–2750, Oct. 2010.
produce very low voltage, current, power, and energy to directly                    [12] R. Pinto, B. Srinivasan, S. Guiot, and B. Tartakovsky, “The effect of real-
                                                                                         time external resistance optimization on microbial fuel cell performance,”
support load and its control circuits. It is especially true when                        Water Res., vol. 45, pp. 1571–1578, 2011.
it comes to lab-scale reactors. The implementation of minimal                       [13] G. Premier, J. Kim, I. Michie, R. Dinsdale, and A. Guwy, “Automatic
power consuming control circuitry has not been investigated in                           control of load increases power and efficiency in a microbial fuel cell,” J.
                                                                                         Power Sources, vol. 196, pp. 2013–2019, 2011.
this paper because this is the proof-of-concept study that shows                    [14] R. Pinto, B. Srinivasan, M. Manuel, and B. Tartakovsky, “A two-
the feasibility of the proposed scheme to harvest actually usable                        population bio-electrochemical model of a microbial fuel cell,” Biore-
energy from MFCs. Hence, an external power supply is used                                source Technol., vol. 101, pp. 5256–5265, 2010.
                                                                                    [15] Z. Ren, H. Yan, W. Wang, M. Mench, and J. Regan, “Characterization
to power the prototype control circuits, but the control power                           of microbial fuel cells at microbially and electrochemically meaningful
needs to be minimized to make a feasible solution for practical                          timescales,” Environ. Sci. Technol., vol. 45, pp. 2435–2441, 2011.
use of MFCs. The external power supply can be eliminated                            [16] C. Donovan, A. Dewan, H. Deukhyoun, and H. Beyenal, “Batteryless,
                                                                                         wireless sensor powered by a sediment microbial fuel cell,” Environ. Sci.
if the MFC scale is large enough such as the one in waste                                Technol., vol. 42, no. 22, pp. 8591–8596, 2008.
water treatment plants, so that the harvesting system can be                        [17] C. Donovan, A. Dewan, H. Peng, D. Heo, and H. Beyenal, “Power manage-
more self-efficient and provide more energy. Implementation of                            ment system for a 2.5 W remote sensor powered by a sediment microbial
                                                                                         fuel cell,” J. Power Sources, vol. 196, pp. 1171–1177, 2011.
low-power-consuming customized control circuits in an ASIC                                                                                            o
                                                                                    [18] B. Logan, P. Aelterman, B. Hamelers, R. Rozendal, U. Schr¨ eder, J. Keller,
chip [32], [33] would be a valid approach for more energy                                S. Freguiac, W. Verstraete, and K. Rabaey, “Microbial fuel cells: Method-
efficient system. Also, a self-start circuitry is required for black                      ology and technology,” Environ. Sci. Technol., vol. 40, no. 17, pp. 5181–
                                                                                         5192, 2006.
start capability.                                                                   [19] J. Park and Z. Ren, “Hysteresis controller based maximum power point
                                                                                         tracking energy harvesting system for microbial fuel cells,” J. Power
                            VI. CONCLUSION                                               Sources, vol. 205, pp. 151–156, 2012.
                                                                                    [20] S. Oh and B. Logan, “Voltage reversal during microbial fuel cell stack
   In this paper, an efficient MFC energy harvesting system us-                           operation,” J. Power Sources, vol. 167, pp. 11–17, 2007.
ing DC/DC converters has been presented. The proposed energy                        [21] A. Shantaram, H. Beyenal, R. Raajan, A. Veluchamy, and
harvesters capture the energy from multiple MFCs at individu-                            Z. Lewandowski, “Wireless sensors powered by microbial fuel cells,”
                                                                                         Environ. Sci. Technol., vol. 39, pp. 5037–5042, 2005.
ally controlled operating points and at the same time forms the                     [22] A. Meehan, G. Hongwei, and Z. Lewandowski, “Energy harvesting with
energy into a usable shape. The proposed parallel operation sys-                         microbial fuel cell and power management system,” IEEE Trans. Power
tem consists of multiple harvesting converters for each MFC and                          Electron., vol. 26, no. 1, pp. 176–181, Jan. 2011.
                                                                                    [23] J. Starzyk, Y.-W. Jan, and F. Qiu, “A dc-dc charge pump design based
a single voltage boost converter. The proposed control scheme                            on voltage doublers,” IEEE Trans. Circuits Syst. I: Fundamental Theory
has been validated experimentally and a successful result has                            Appl., vol. 48, no. 3, pp. 350–359, Mar. 2001.
been shown.                                                                         [24] Maxim Integrated Products. (2009) AN725: DC-DC conversion
                                                                                         without inductors. [Online]. Available: http://www.maxim-ic.com/app-
                               REFERENCES                                           [25] N. Mohan, T. Undeland, and W. Robbins, Power Electronics: Converters,
                                                                                         Applications, and Design. New York: Wiley, 2002.
 [1] Z. Ren, T. Ward, and J. Regan, “Electricity production from cellulose in a     [26] L. Solero, A. Lidozzi, and J. Pomilio, “Design of multiple-input power
     microbial fuel cell using a defined binary culture,” Environ. Sci. Technol.,         converter for hybrid vehicles,” IEEE Trans. Power Electron., vol. 20,
     vol. 41, no. 13, pp. 4781–4786, 2007.                                               no. 5, pp. 1007–1016, Sep. 2005.
          This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10                                                                                                                IEEE TRANSACTIONS ON ENERGY CONVERSION

[27] F. Boico and B. Lehman, “Mutliple-input maximum power point tracking                                     Zhiyong Ren received the Ph.D. degree from the
     algorithm for solar panels with reduced sensing circuitry for portable                                   Pennsylvania State University, University Park, in
     applications,” Solar Energy, vol. 86, pp. 463–475, 2012.                                                 2008.
[28] Y.-C. Liu and Y.-M. Chen, “A systematic approach to synthesizing multi-                                     He is currently an Assistant Professor of civil en-
     input DC-DC converters,” IEEE Trans. Power Electron., vol. 24, no. 1,                                    gineering at the University of Colorado Denver, Den-
     pp. 116–127, Jan. 2009.                                                                                  ver, where he also directs the Environmental Biotech-
[29] E. Carlson, K. Strunz, and B. Otis, “A 20 mV input boost converter with                                  nology Laboratory. His lab uses bioelectrochemical
     efficient digital control for thermoelectric energy harvesting,” IEEE J.                                  systems and fermentation technology to directly con-
     Solid-State Circuits, vol. 45, no. 4, pp. 741–750, Apr. 2010.                                            vert cellulosic biomass and waste water into H 2 and
[30] B. Acker, C. Sullivan, and S. Sanders, “Synchronous rectification with                                    electricity, and he uses molecular microbiology tools
     adaptive timing control,” in Proc. 26th IEEE Power Electron. Spec. Conf.,                                and electrochemical analyses to understand the fun-
     Jun, 1995, vol. 1, pp. 88–95.                                                   damental determinant factors of those systems so as to enhance design, op-
[31] J. Park and Z. Ren, “High efficiency energy harvesting from microbial fuel       eration, and monitoring in concert with traditional approaches. His research
     cells using a synchronous boost converter,” J. Power Sources, vol. 208,         focuses on bioenergy production during waste treatment processes, with the
     pp. 322–327, 2012.                                                              goal of expanding environmental engineering from pollution clean-up to sus-
[32] H. Lhermet, C. Condemine, M. Plissonnier, R. Salot, P. Audebert, and            tainable development of energy and environmental systems.
     M. Rosset, “Efficient power management circuit: From thermal energy
     harvesting to above-ic microbattery energy storage,” IEEE J. Solid-State
     Circuits, vol. 43, no. 1, pp. 246–255, Jan. 2008.
[33] A. Richelli, L. Colalongo, S. Tonoli, and Z. Kovacs-Vajna, “A 0.2V
     DC/DC boost converter for power harvesting applications,” IEEE Trans.
     Power Electron., vol. 24, no. 6, pp. 1541–1546, Jun 2009.
[34] J. Park and Z. Ren, “Efficient energy harvester for microbial fuel cells using
     DC/DC converters,” in Proc. IEEE Energy Convers. Congress Expo., Sep.
     2011, pp. 3852–3858.

                          Jae-Do Park (M’07) received the Ph.D. degree from
                          the Pennsylvania State University, University Park,
                          in 2007.
                              He is currently an Assistant Professor of electri-
                          cal engineering at the University of Colorado Denver,
                          Denver. Prior to his arrival at the University of Col-
                          orado Denver, he was with Pentadyne Power Corpo-
                          ration, Chatsworth, CA, as a Manager of Software and
                          Controls, where he took charge of control algorithm
                          design and software development for the high-speed
                          flywheel energy storage system. His research inter-
ests include various energy and power system research and education including
electric machines and drives, energy storage and harvesting systems, renewable
energy sources, grid-interactive distributed generation, microturbine control,
and microgrid systems.

To top