e94-e104

Upload: alex-samuel-silva

Post on 03-Jun-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/12/2019 e94-e104

    1/11

    education for chemical engineers 8 ( 2 0 1 3 ) e94e104

    Contents lists available at ScienceDirect

    Education for Chemical Engineers

    j ournal homepage: www.elsevier .com/ locate /ece

    Gas solute movementin packed columnsA remote

    control experiment

    Hugo Silvaa, Sandra Sa, Lcia Brandoa, J.M. Loureirob, Joaquim Gabriel c,Adlio Mendesa,

    a LEPAE-Departamento de Engenharia Qumica, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias,

    4200-465 Porto, Portugalb LSRE-Departamento de Engenharia Qumica, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias,

    4200-465 Porto, Portugalc IDMEC Plo FEUP, Departamento de Engenharia Mecnica, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias,

    4200-465 Porto, Portugal

    a b s t r a c t

    A novel teaching remotecontrolled experiment is describedconcerning thesolutemovement through an adsorption

    column. A packed-bed column filled with 5A zeolite was inserted in a thermostatic oven and connected to a TCD

    detector. The complete set-up can be remotely controlled and visualized alive using an internet connected camera

    to improve the sense of reality. The experiment purpose was to assist and motivate students regarding a topic that

    usually they demonstrate difficulty to assimilate, which is the prediction of concentration fronts behavior by using

    the solutemovement theory(SMT). Theset-upis versatileto studybreakthroughcurves andfeedpulses.Two loopsof

    2 cm3 and5cm3 allowthe injectionof O2or N2adsorbatespecies thathave different isothermstypeandconsequentlyconcentration fronts history. Interaction between shock and diffuse waves is addressed for the narrow pulse case

    of2 cm3. Also, students are able to obtain the nitrogen and oxygen isotherms at different temperatures for the 5A

    zeolite (chromatographicmethod) andcompare those isothermswith those obtained by thevolumetric method; the

    latter is used as reference method.

    2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

    Keywords: Gas adsorption; Solutemovement theory; Remote control experiment

    1. Introduction

    Educational laboratories are an important part of theundergraduate chemical engineering study plan. With the

    development of computers and the worldwide internet com-

    munications network, the laboratory pedagogical approach

    has been changing. Since the late 1990s, a new concept

    is emerging in engineering education institutions: remotely

    controlled laboratories (Selmer et al., 2007). Remote labs

    are systems based on real equipment, which allow stu-

    dents to perform experimental work through an internet

    connection (Coito and Palma, 2008). Although remotely con-

    trolled laboratories are in a maturation process, there are

    some advantages pointed in the literature when compared

    Corresponding author. Tel.: +351 22 508 1695; fax: +351 22 508 1449.E-mail address: [email protected] (A. Mendes).Received23 July 2012;Receivedin revisedform25 June 2013;Accepted30 June2013

    to traditional hands-on laboratories. Engineering education

    institutions face budget restrictions to create and maintain

    traditional hands-on laboratories, due to an increasing needof space,staff,availabilityand safety (Ogotet al.,2003;Wiesner

    andLan,2004;Abu-ElHumos etal., 2005;Azad,2007; Jing etal.,

    2007; Rafael et al., 2007; Selmer et al., 2007; Murrayet al., 2008;

    Wiseman et al., 2008; Gravier et al., 2009). On the other hand,

    remotelycontrolled experiments areavailable fromanywhere

    at anytime which facilitates the scheduling process. In addi-

    tion, students safety is guaranteed, minimum supervision

    is involved and more users can perform the experiment in

    comparison with hands-on. Finally, it opens the possibility of

    sharing experiments among different institutions, increasing

    the number and variety available (Bourne et al., 2005; Azad,

    1749-7728/$ see front matter 2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.ece.2013.06.002

    http://www.sciencedirect.com/science/journal/17497728http://www.elsevier.com/locate/ecemailto:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ece.2013.06.002http://localhost/var/www/apps/conversion/tmp/scratch_9/dx.doi.org/10.1016/j.ece.2013.06.002mailto:[email protected]://www.elsevier.com/locate/ecehttp://www.sciencedirect.com/science/journal/17497728
  • 8/12/2019 e94-e104

    2/11

  • 8/12/2019 e94-e104

    3/11

    e96 education for chemicalengineers 8 ( 2 0 1 3 ) e94e104

    Fig. 1 Experimental setup for the gas adsorption experiment on a porous adsorbent.

    Gas feed system two gases can be supplied to the GC,

    nitrogen and oxygen, following two feeding modes, step per-

    turbationand conventional injection (pulse perturbation). For

    the conventional injection, an automatic injection valve (A,

    VICI E36) is used. Three-way electric valve (B), Fig. 1, is used

    to select the gas and three-way electric valve (C) is used to

    select theinjection volume loop (either 2 cm3 or5cm3).A third

    three way electric valve (D) allows the selection of the exit

    position (from2cm3 to5cm3 loops) of theautomatic injection

    valve to the packed bed column. The carrier gas, helium, is

    supplied to the GC by a mass flow controller (E, Bronkhorst,

    100cm3min1). For step perturbations, electric on/off vales F,G and H are used. The mass flow controller is then used to

    control the flow rate. At the adsorption columntop,a pressure

    sensor (I) reads the feeding pressure.

    Powersupply andacquisition board a standard 12VDCpower

    supply is used to power the mass flow controller, pressure

    sensor and electric valves. The valves are controlled using

    two in-house built relay boards (with 4 relays each). The relay

    boards, mass flow controller, pressure sensor and GC oven

    temperature controller (West 8100 PID) are connected to an

    acquisition board (Advantech, PLCD-8710).

    Web camera an internet surveillance IP camera (PIV-6732,

    IP PZT network camera) was connected to a switch (Asus

    GX-100SB) together with the computer and the switch wasconnected to an internet socket. Computer software allows

    using thecamera that is capable of zoom anddirectional con-

    trol, giving a sharp viewof the experimental set-up.

    Controlling software a comprehensive controlling program

    was developed in LabVIEW7.1 (National Instruments) Fig. 2.

    It allows setting (a) the carrier gas flow rate, (b) the solute

    species, (c) the pulse injection volume (selectable 2 cm3 or

    5cm3), (d) the perturbation type, (e) the adsorption temper-

    ature (temperature of the oven) and (f) to trigger the injection

    valve. This interface plots on-line the TCD response (related

    to the exiting stream concentration), shows the status of all

    setting variables and displays the pressure at the columns

    top. The variables carrier gas flow rate, TCD response, pres-sure entrance and retention time can be saved in a text file.

    The LabVIEW software has a built-in web server facility to

    allow easy internet access, thus providing the option of pub-

    lishing the graphical user interface (GUI). This GUI is available

    at http://elabs.fe.up.pt and students can access and control

    the experiment by using a standard web browser, preferably

    Internet Explorer6. Fig. 3 shows a schematic representationof

    the installation set-up.

    The experimentalset-upwas developedmostlyby a gradu-

    atestudent, during hismaster thesis, andtook advantage of a

    very old GC obtained at a low price. The whole set-up is quite

    inexpensive and it was a great opportunity of learning some

    different and relevant competences.

    3. Solute movementtheory

    This section presents a brief description of the underlying

    theory to help the student in understanding and analysing

    the experiment. Reference books (Wankat, 1990; Seader and

    Henley, 1998) could be used to help in a deeper understanding

    of the solutemovement theory (SMT).

    When a pulse gas enters the packed-bed column, it either

    moves at the interstitial velocity defined by the carrier gas

    flow rate, or is immobilized inside the adsorbent (adsorbedor

    in the fluid stagnant phase). Accordingly, the average veloc-

    ity depends on the time that the solute is retained inside the

    adsorbent. This is quite effortless to understand and explain.

    However, complexity is present when it is necessary to inter-

    pret the formationof different concentration-frontwaves and

    how they interact. This is when the solute movement theory

    is helpful since, despite being a mathematical model based

    on simple physical phenomena, it can be used to analyze and

    understand rather complex processes.

    The SMTestablishesthat an infinitesimalelement of a sin-

    gle gas solute, with partial pressurepA, will migrate through

    the packed-bed column at a velocity us, which inversely

    depends on the slope of the adsorption isotherm (dqA/dpA),

    us(pA) =

    z

    t

    pA

    =ui

    1+ ((1 )/)pKd + ((1 )/)T(dqA/dpA)

    (1)

    where ui is the interstitial velocity of the carrier gas in the

    column, pA is the solute partial pressure in the fluid phase,

    qA is the solute concentration in the solid phase, is the

    http://elabs.fe.up.pt/http://elabs.fe.up.pt/
  • 8/12/2019 e94-e104

    4/11

    education for chemicalengineers 8 ( 2 0 1 3 ) e94e104 e97

    Fig. 2 Experiment user interface developed in LabVIEW 7.1, with a nitrogen chromatogram response.

    Fig. 3 Schematic representation of the experiment.

    solid apparent density, is the gas constant, T is the abso-lute temperature, z is the column axial position, t is time,

    is the packing (interparticle) porosity, p is the intraparticle

    porosity (within the particle) and Kd is the fraction of intra-

    particle volume the species can penetrate; Kd = 1 for small

    solute molecules (present case). As it can be seen from Eq.

    (1), the solute movement is retarded since the solute is in

    the adsorbed phase, ([(1 )/]T(dqA/dpA)) or in gas phase

    inside the porosity of the adsorbent ([(1 )/]pKd).

    The DeVault (SMT) equation can be derived from a solute

    mass balance to the absorbent column:

    uipA

    z

    +pA

    t

    +1

    pKdpA

    t

    +1

    TqA

    t

    = 0 (2)

    considering themathematical relations (3) and (4):

    dpAdt

    pA

    = 0 =

    pAz

    t

    z

    t

    pA

    +

    pAt

    z

    (3)

    zt

    pA

    = (pA/t)z(pA/z)t

    (4)

    sincepA =f(z, t), one obtains Eq. (1).

    The solute mass balance is based on the following main

    assumptions (Wankat, 1990):

    (i) Instantaneous equilibrium between the gas and solid phase;

    (ii) Plug flow of the gas phase;

    (iii) No pressure drop along the column;

    (iv) Isothermal operation;

    (v) Ideal gas behavior.

    Among these assumptions, the least likely to be valid

    concerns the local equilibrium. This assumption states that

    q relates to p through the adsorption isotherm, neglect-

    ing intraparticle mass transport limitations. The model also

    assumes plug flowwithnoaxial dispersion,whichisnotlikely

  • 8/12/2019 e94-e104

    5/11

    e98 education for chemicalengineers 8 ( 2 0 1 3 ) e94e104

    q1

    q2

    p1 p2

    Fig. 4 Two-segment discretized favorable adsorption

    isotherm.

    to happen. However, for favorable isotherms (with negative

    secondderivative Fig. 4) and when equilibrium is the domi-

    nant phenomenon, normally the SMT approximates well the

    adsorption experimental data.

    3.1. SMTapplied to Langmuir type isotherms

    Eq. (3) shows that in order to correctly predict the velocity

    of the solute inside the column, the adsorption equilibrium

    isotherm for the solute in the adsorbent should be known.

    Following, the formation of shock waves and concentration

    frontswhen the solute adsorbs in the adsorbent phase with aLangmuir type isotherm (Fig. 4) will be considered. This type

    of isothermis usuallycalled favorable isotherm(withnegative

    second derivative).

    The adsorption solute movement of species with a favor-

    able isotherm can bemore easily conceptualized discretizing

    the isotherm into two segments Fig. 4.

    For a feed concentration p1 the solute velocity would be

    u1; on the other hand, for an adsorption column initially in

    equilibriumwith a feed streamof concentrationp1, when thefeed concentration is changed top2, the solute velocitywould

    be u2and u2 >u1, since theadsorption columnretains (delays)

    more the soluteup to concentrationp1. When feeding a clean

    column with a stream of concentration p2, the solute front

    wouldbeas sketchedin Fig. 5. Thehigher concentrations (that

    could travel at higher velocities, according to Eq. (3)) cannot

    move faster than the lower concentrations like when a car at

    ahigherspeed cannotovercomea lower speedcar infrontof it.

    The following link shows thedescribedanalogy, http://chelta.

    fe.up.pt/chelta/student/animations/road.swf. Because of this

    limitation in the velocity of the higherconcentrations, a mass

    balance is established at the solute front and a shock wave is

    formed thataverages thesolute velocityforall concentrations,as it is presented in Eq. (5):

    ushock =ui

    1+ ((1 )/)pKd + ((1 )/)T(qA/pA) (5)

    For the desorption of a column initially saturated in equi-

    libriumwithp2, a dispersivewave front is formed; as is shown

    in Fig. 6, the lower concentrationmoves more slowly.

    The solutemovement is usually explainedbased on solute

    movement diagrams, graphical expression of the mathe-

    matical method of characteristics used in the solution of

    (hyperbolic) partial differential equations Fig. 7. In Fig. 7, it is

    represented the shock wave velocity slope ushock, and the gen-eration of the spreading zone, for the response to a feed pulse

    p2

    p1

    z

    Shock

    Wave

    Fig. 5 Conceptual adsorption solute front movementfor the isotherm depicted in Fig. 4; a shock wave front is formed.

    p2

    p1

    z

    Fig. 6 Conceptual desorption solute front movementfor the isotherm depicted in Fig. 4; a dispersivewave is formed.

    http://chelta.fe.up.pt/chelta/student/animations/road.swfhttp://chelta.fe.up.pt/chelta/student/animations/road.swfhttp://chelta.fe.up.pt/chelta/student/animations/road.swfhttp://chelta.fe.up.pt/chelta/student/animations/road.swf
  • 8/12/2019 e94-e104

    6/11

    education for chemicalengineers 8 ( 2 0 1 3 ) e94e104 e99

    ushock(pH)

    L

    Z

    pout

    pin

    Time

    pH

    Time

    Time

    0

    0

    0

    pH

    Fig. 7 Feed pulse diagrams according to the solute

    movement theory: (a) inlet concentration; (b) shock and

    diffuse waves formation and; (c) outlet concentration.

    of a solutewith favorable isotherm. In this case, the shockand

    diffuse waves for the highest concentrations exit the column

    at the same time, without intersection during their migra-

    tion. Fig. 8 sketches thepulse shape as it progresses along the

    adsorption column.

    For favorable isotherms the following relation between

    velocity and concentration is of great importance: us(Chigh) >

    ushock> us(Clow). Therefore, for a feed pulse perturbation theleading shock wave is intersected by the high concentration

    solute movement if the adsorption column is long enough, as

    it can be deduced fromFigs. 7 and8. Consequently, the solute

    elutesat lower concentrationswhen compared to theoriginal

    feedpulse.Regardingtheconcentrationcurves, students often

    have difficulty to mentally visualize that the profile shown

    in Fig. 8 (t=5), where concentration is represented as a func-

    tion of the axial position inside the column, is symmetrical to

    the representation of the concentration-fronts as a function

    of time (histories), shown in Fig. 7.

    For the case of linear isotherms, the solutewaves travel all

    at the samevelocity and the waves formedareusually named

    simple. The student should refer to known reference bookson SMT to a better understanding of these concepts (Wankat,

    1990; Seader andHenley, 1998).

    3.2. Adsorption isotherms determined by the SMT

    Theadsorption isothermof thesolute species in theadsorbent

    packed in the column can be obtained if the concentration

    history to a feed pulse is obtained at the exit of the column.

    Fig. 8 Sketch of the solute pulse progress along an adsorption column.

  • 8/12/2019 e94-e104

    7/11

    e100 education for chemicalengineers 8 ( 2 0 1 3 ) e94e104

    Fig. 9 Chromatograph response to a solute feed pulse

    with favorable isotherm.

    Theadsorption isothermcanbe obtained from the desorp-

    tion response to a feed perturbation (diffuse wave). This

    happens because the shape of the diffuse wave is related to

    solutevelocityas a functionof thesoluteconcentration(Fig. 9):

    uiu(pA)

    =trt0

    (6)

    where t0is thecarriergas retentiontimeand tr(pA)isthesolute

    retention time, which is a function of solute partial pressure.

    Replacing Eq. (1) in (6), it becomes:

    trt0

    = 1+1

    Tf(pA)+

    1

    p (7)

    and expressing the isotherm derivative it becomes:

    dqAdpA

    =1

    (1 )T

    trt0

    [(1 )p + ]

    (8)

    Integrating nowEq. (8) one obtains:

    qA =1

    (1 )T

    pA00

    trt0

    [(1 )p + ]dpA (9)

    Eq. (9) can be solved based on the TCD response obtained

    for the solute concentration as a function of the retentiontime, pA =f((tr/t0) [(1 )p + ]). However, the TCD used

    is old and has a large dead volume, acting as a stirring

    tank at the end of the adsorption column that originates the

    peak to exhibit a shape similar to the case of a favorable

    isotherm (Wankat, 1990). This is a common source of devi-

    ations between experimental andmodel results.

    Thepartialpressurevalues ontheordinate axis of the chro-

    matogram, h, do not represent the solute partial pressure, but

    the h values of ordinate axes are directly proportional to the

    solute partial pressure, according to Eq. (10):

    pA = k h (10)

    where h is thearbitraryscale chosen for theordinateaxis (e.g.

    the electrical signal from the GC in mV) and k the propor-

    tionality factor. This way, the solute partial pressure can be

    obtained since the product between the total pressure (pA0 )

    Time / s

    0 50 100 150 200 250 300

    NormalizedAmplitude

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    N2

    -5 cm3

    Fig. 10 Response to a nitrogen feed pulse, at 60C and

    20cm3Nmin1 helium flow rate for the injection loops of

    5 cm3.

    and the injection time tinjis equal to the chromatogram total

    area, according to:

    pA0 tinj =

    0

    pAdt = k

    0

    h dt (11)

    where:

    tinj =Vinj

    U (12)

    Vinj is the loop injected volume and U is the carrier gas flowrate.

    The proportionality factor is obtained from:

    k =pA0 tinj

    0 hdt

    (13)

    After obtaining thepartial pressurepA value, the next step

    is the integration of Eq. (9) to obtain the adsorption isotherm,

    qA =f(pA).

    4. Observations and discussion

    4.1. Determination ofadsorption isotherms

    In this experimental section, students can determine the

    adsorption isotherms of nitrogen and oxygen using the chro-

    matographic method. Fig. 10 shows the chromatographic

    experimental response to a nitrogen feed pulseof 5 cm3 in the

    packed bed column, at 60C and 20cm3Nmin1 helium flow

    rate. Students candetermine the nitrogen adsorption equilib-

    rium isotherm for these conditions, as described in thesolute

    movement theory section.

    Fig. 11 reveals that nitrogen follows a Langmuir type

    isotherm on the 5A zeolite; also it compares the adsorption

    isotherm obtained from the volumetric method (reference

    method) with the chromatographic response for a 5cm3

    feed pulse of nitrogen at 60 C and for a carrier flow rate

    of 20cm3Nmin1. Students can conclude that the nitrogen

    equilibrium isotherms determined by both methods are very

    similar. Other main conclusion that can be taken in this

  • 8/12/2019 e94-e104

    8/11

    education for chemicalengineers 8 ( 2 0 1 3 ) e94e104 e101

    Pressure / bar

    0.0 0.1 0.2 0.3 0.4 0.5 0.6

    q/m

    ol.kg

    -1

    0.00

    0.05

    0.10

    0.15

    0.20Chromatographic method

    Volumetric method

    Fig. 11 Comparison of the adsorption equilibrium

    isotherms for nitrogen in the 5A zeolite obtained by the

    volumetric method and from applying the SMT for a

    nitrogen feed pulse, at 60 C.

    procedure is that temperaturestrongly affects theequilibrium

    and consequently the determined isotherms.

    In order to apply the SMT, Eq. (1), in the following sections,

    students can use the adsorption equilibrium isotherms for

    oxygenandnitrogeninthe5Azeolite obtainedby theso-called

    volumetricmethod at 40 C and 60 C. These adsorption equi-

    librium isotherms are illustrated in Fig. 12. Nitrogen shows a

    typical favorableshape isotherm Langmuirtype (represented

    by Eq. (15)) as observed above, while oxygen displays a linear

    behavior (represented by Eq. (14)).

    qO2 = HPO2 (14)

    qN2 =QKLPN21+ KLPN2

    (15)

    Students should use the adsorption isotherms parame-

    ters represented in Eq. (14) or (15) obtained by the volumetric

    method (Table 1) when they areapplying theSMTmathemat-

    icalmodel.

    Table 1 Nitrogen and oxygen equilibriumparametersisotherms on the 5A zeolite at 40C and 60 C.

    T(C) N2 O2

    Q(molkg1) KL(bar1) H (molkg1 bar1)

    40 2.42 0.117 0.094

    60 2.27 0.077 0.069

    Time / s

    0 50 100 150 200 250 300

    NormalizedAmplitude

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    Step at 60 CStep at 40 C

    Fig. 13 Response to a nitrogen feed step at 40C and 60 C

    column temperature and heliumflow rate of 20cm3Nmin1.

    4.2. Determination ofconcentration fronts by the SMT

    With the equilibrium isotherms, it is now possible for stu-

    dents to use the SMT and compare the model predictions

    with the results obtained experimentally. With this set of

    experiments, students canevaluate which variables affect thevelocityof a concentration-front.Theexperimental procedure

    allows to analyze the influenceof thefollowing variables: feed

    perturbations, injection volumes (for feed pulse), adsorption

    temperature (T) and interstitial velocity (ui).

    4.2.1. Feed step

    Fig. 13 shows the normalized experimental response for a

    nitrogen positive feed step into the packed bed column, with

    pressure / bar

    0.0 0.5 1.0 1.5 2.0

    q/mol.kg-1

    0.0

    0.1

    0.2

    0.3

    0.4

    0.5

    O2at 40 C

    O2at 60 C

    pressure / bar

    0.0 0.5 1.0 1.5 2.0

    q/mol.kg

    -1

    0.0

    0.1

    0.2

    0.3

    0.4

    0.5

    N2at 40 C

    N2at 60 C

    Fig. 12 Adsorption equilibrium isotherms on zeolite 5A of (a) oxygen and (b) nitrogen at 40C and 60 C, determined by the

    volumetric method.

  • 8/12/2019 e94-e104

    9/11

    e102 education for chemicalengineers 8 ( 2 0 1 3 ) e94e104

    a feedflowrate of 20cm3Nmin1 (average column pressure of

    1.17bar) at 40C and 60 C. According to the SMT, this pertur-

    bation leads exclusively to the formation of a shock wave; it

    is expected that nitrogen exits the column at tshock = L/ushock,

    where L is the length of the adsorption column and Eq. (5)

    is used to obtain the velocity of the shock wave. Consider-

    ing the column temperature at 60C, the exit time of the

    shock wave for the data presented in Fig. 13 is ca. 163s

    and the SMT predicts an exit time of 156s . The SMT was

    expected to approximate much more closely the experimen-

    tal data since the dominant phenomenon during nitrogen

    migration through thepacked-bed column should be adsorp-

    tion. Actually, for favorable isothermsandwhenequilibriumis

    the dominant phenomenon the SMT normally approximates

    well the adsorption experimental data (Wankat, 1990). As the

    temperature increases the adsorption isothermbecomes less

    favorable and the concentration front becomes more disper-

    sive (Wankat, 1990) (see Fig. 13). Finally, it was assumed that

    the column was operating at an average value between the

    outlet and inlet pressures (for this particular case, an average

    column pressure of 1.17barwas used in the calculations).

    4.2.1.1. Effect oftemperature. Inorderto illustratethe temper-

    ature influence in theadsorption process, Fig. 13 presents the

    experimental results fornitrogen feed stepsperformedat two

    different temperatures, 40 C and 60 C, and for a carrier gas

    and nitrogen feed flow rate of 20cm3Nmin1.

    Nitrogen retention time is higher at 40 C, since the

    adsorptioncapacityof theadsorbentdecreases forhighertem-

    peratures. This happens because adsorption is an exothermic

    process and the equilibrium constants follow the Vant Hoff

    equation. In this way, students can verify that adsorption

    equilibrium is strongly affected by temperature. Additionally,

    the dispersion phenomenon can be compared for both step

    results. Thefront shapeof a breakthroughcurve is theresultof

    dispersion and compression actions. Thedispersion ismostly

    related to the axial dispersion and to the intraparticle mass

    transport while thecompressionresults fromthenonlinearity

    of the adsorption isotherm as well as the adsorption capac-

    ity. Both dispersion andcompression forcesdecreasewith the

    temperature but the compression forces decrease normally

    more. Fig. 13 indicates that in the present case the concen-

    tration front at 60C shows indeed a higher dispersion.

    4.2.2. Feedpulse

    Students can realize with this set of experiments that com-

    plexity of the SMT is increased for a feed pulse mainly when

    the shock wave, originated due to the positive concentration

    step of the pulse, interacts with the diffuse wave, originated

    due to the negative concentration step of the pulse Fig. 8.

    4.2.2.1. Influence of solute species and concentration.

    Figs. 14 and 15 show the response of the chromatographic

    system at different feed pulses. Two volumes of nitrogen

    and oxygen were injected: 5cm3 and 2cm3. The carrier gas

    velocity and temperature were 20cm3Nmin1 and 60 C,

    respectively.

    Students can take several conclusions from these exper-

    imental observations. The injection of different volumes

    of nitrogen (different inlet concentrations) into the column

    results in different retention times (Fig. 14). According to the

    SMT predictions, the retention time of a shock wave for both

    volumes injectedin thenitrogen feed pulseis 156s (see above).

    Fig. 14 shows that the 5 cm3 pulse retention timecorresponds

    Time / s

    0 50 100 150 200 250 300

    NormalizedAmplitude

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    N2pulse of 2 cm3

    N2pulse of 5 cm3

    Fig. 14 Response to nitrogen feed pulse, at 60C and

    20cm3Nmin1 helium flow rate and for two injection loops:

    5 cm3 and 2cm3.

    Time / s

    0 50 100 150 200 250 300

    NormalizedAmplitude

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    O2pulse of 2 cm3

    O2pulse of 5 cm3

    Fig. 15 Response to oxygen feed pulse, at 60C and

    20cm3 Nmin1 helium flow rate and for two injection

    loops: 5 cm3 and 2cm3.

    to this value. However, the retention time for the 2cm3 feed

    pulse is higher, ca. 170 s, suggesting that the shock wavewas

    intercepted by the higher concentrations of the diffuse wave.

    To confirm this hypothesis, the expected intersection time

    (t1) of the concentration-waves was determined. The shock

    wave and thediffuse wave at thehighest concentration inter-

    sect atz = z1when, ushockt1 = us(t1 tF)where tF = Vinj/U isthe

    injection pulse time and us is the diffuse wave velocity at the

    highest concentration. Students should obtain usby using Eq.

    (3) where dqA/dpAshould be evaluated at the average column

    pressure (in this case ca. 1.17bar, see above). The computed

    time for the nitrogen 5cm3 injection volume is 234 s, which

    corresponds to the column axial position (z1 = ushock t1) of

    172cm. Since the packed-bed column is only 115cm long, the

    shock and diffuse waves for the 5cm3 nitrogen pulse never

    intercept. In the 2 cm3 nitrogen feed pulse, where the pulse

    is narrower, the interception time is at t=94 s, which corre-

    sponds tothecolumnaxialpositionof 69cm. Therefore, in this

    case the shock wave is intersected by the higher concentra-

    tions of the nitrogen diffuse wave before exiting the column.

    Taking in to account this interception, the SMT predicts a

    retention time of 164s for the 2cm3 feed pulse, comparing

    quitewell with the experimental result.

    Concerning the oxygen feed pulse, experimental observa-

    tions show that the retention time is similar for both injected

  • 8/12/2019 e94-e104

    10/11

    education for chemicalengineers 8 ( 2 0 1 3 ) e94e104 e103

    Time / s

    0 20 40 60 80 100 120 140 160 180

    NormalizedAmplitude

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    Fig. 16 Response to a feed pulse of 5cm3 of nitrogen at

    60 C and 40cm3Nmin1 helium flow rate.

    volumes (Fig. 15). Students can realize this as a consequence

    of the oxygen linear type isotherm. In this case, all the oxygenconcentrations travel through the column at same velocity,

    forming a simple concentration front wave. The experimen-

    tal retention time is around 90s, while SMT prediction is 83s.

    However, the asymmetric shape of the peaks might indicate

    a favorable isotherm, forming a shock front and a dispersive

    wave on the backward. This is not the case and the shape

    is partially related to the dead volume of the TCD, ca. 3 cm3,

    which behaves like a perfectlymixed tank.

    Figs. 13 and 14 shownormalized responses to respectively

    step and pulse nitrogen perturbations. The adsorption dis-

    persion to the step perturbation is larger than to the pulse

    perturbation but it should be noted that the maximum pres-

    sure for the step perturbation is ca. 1bar and for the pulseperturbation is just ca. 0.35bar.

    4.2.2.2. Influence of the carrier flow rate. The influence of the

    carrier gas flow rate was observed by performing a 5cm3

    nitrogen feed pulse at 40cm3Nmin1 and 60C. The obtained

    chromatogramis presented in Fig. 16. Students canshow that

    theSMT predictsa retention time of 91s, considering an aver-

    age column pressure of 1.33bar (higher flow rate than before)

    and compare with the experimental retention time. Fig. 16

    shows that the retention time is approximately 81s. This 10s

    difference between the SMT prediction and the experimental

    result canbe related to thehigher pressure drop in thepacked

    bed column. In the literature, it is reported that prematurebreakthrough curves can be obtained due to velocity devia-

    tions, which result from high pressure drop effects (Zwiebel,

    1969). Finally, by comparing these experimental results with

    the ones for the carrier flow rate at 20cm3Nmin1 (Fig. 14),

    students can conclude that higher carrier flow rates simply

    decrease the exit time of the concentration front.

    5. Students opinion

    Thisexperiment is available to students thatenroll Separation

    Processes II subject, third year of the Chemical Engineering

    program at Faculty of Engineering of Porto University (FEUP),

    when the traveling concentration wave topic is taught. For

    performing an experimental run, students must book a time

    window. The booking process is currently available by e-mail

    but soon will also be possible to book directly from the web-

    site of the experiment. 23 students that contacted with this

    Table 2 Responses of the students to the inquire.

    # Question Fraction of positiveanswers (%)

    1 Since remote controlled experiments are

    available at anytime, in your opinion would

    they be suitable for learning consolidationthrough repetition?

    82.6

    2 Would you preferto perform the experiments

    during the class time or at anytime using the

    remote control facility?

    73.9

    3 Duringtutorialclassesdo youprefer touse

    experimental data acquired using a remote

    controlled experiment or to use assumed

    data?

    91.3

    4 Doyou consider that thelivecamera

    improves the really sensewhen using the

    remote controlled experiment?

    100

    5 Are youfamiliar withshock waves and

    dispersive front?

    82.6

    experiment were inquired and questions and answers orga-

    nized in Table 2.

    Students generally prefer to use this remotely controlled

    experiment to generate the data needed for solving train-

    ing problems than to use assumed data. Though they think

    this experiment can be used to consolidate their knowledge

    because,theyprefer toperformitduringthelabclasses,proba-

    blydue tosomeindependent learning insecurity. Thepresence

    of the live camera is unanimously viewed as very relevant to

    give a sense of reality to the experiment. Finally, the students

    state that the shock wave and dispersive fronts are familiar

    concepts, showinga relevantassimilation of critical concepts.

    6. Conclusions

    A remotely controlled educational experiment was developed

    to supportstudents acknowledgmentofa quite complextopic

    in chemical engineering: gasadsorption throughapacked-bed

    column. From anywhere, at any time, students can comple-

    ment the theoretical knowledge gathered about this topic,

    withreal experimentsperformedonline(http://elabs.fe.up.pt).

    The solute movement theory, as oversimplified model, is

    the ideal tool for students first contact with gas adsorp-

    tionphenomena.Whileperforming theexperiments,students

    should critically discuss the results with the model predic-

    tions. This way, it is expected that students consolidate their

    knowledge on adsorption and on solute movement in an

    adsorption columnandunderstand themodel limitations and

    all the influence factors inherent to this kind of separation

    processes. Regarding students opinion, it was asked if they

    preferred to use the remote controlled experiment with a

    virtual interface or a real set-up during class time. A sig-

    nificant percentage (73.9%) stated a preference for making

    experiments in the laboratory class, using real equipment.

    Theexperiment targets both laboratory classes as well as lec-

    tures/tutorials classes.The greatestnovelty is,however,its use

    to assist teaching lecture/tutorial classes and namely Sepa-

    ration Processes courses. This experiment can be used, for

    example, to obtain the real solution for a problem ques-

    tioned to a group of students in a tutorial class. Comparing

    their result with real result, students can learn more and

    deeper since they are also involving more their emotion.

    Finally, students seem to be motivated with this learning

    http://elabs.fe.up.pt/http://elabs.fe.up.pt/
  • 8/12/2019 e94-e104

    11/11

    e104 education for chemicalengineers 8 ( 2 0 1 3 ) e94e104

    strategy, since they expressed their preference of using real

    experimental data instead of assumed data.

    Acknowledgments

    The authors would like to acknowledge the Chemical Engi-

    neering Department of the Faculty of Engineering at the

    University of Porto and paint company CIN SA for supportingthis work. The authors are also grateful to Eng. LuisMatos for

    collaborating assembling the experimental set-up.

    References

    Abu-El Humos, A., Alhalabi, B., et al., 2005. Remote labsenvironments (RLE): a constructivist online experimentationin science, engineering, and information technology. In: 31stAnnual Conference of IEEE, Industrial Electronics Society2005, IECON 2005.

    Azad, A.K.M., 2007. Delivering a remote laboratory course withinan undergraduate program. Int. J. Online Eng. 3, 7.

    Becnel, J.M., Holland, C.E., et al., 2002. Fundamentals of Fixed BedAdsorption Processes: Analysis of Adsorption Breakthroughand desorption Elution Curves. American Society forEngineering Education,Montreal, Quebec, Canada.

    Bourne, J., Harris, D., et al., 2005. Online engineering education:learning anywhere, anytime. J. Eng. Educ. 94,131145.

    Coito, F., Palma, L.B., 2008. A remote laboratory environment forblended learning. In: First International Conference onPervasive Technologies Related to Assistive Environments,PETRA 2008, July 16, 2008July 18, 2008. Athens, Greece,Association for ComputingMachinery.

    Cruz, P., Mendes, A., et al., 2001. Using in-bed temperatureprofiles for visualizing the concentration-front movement.Chem. Eng. Educ. 35, 122127.

    Fernandes, D.L.A., Xavier, A.M.R.B., et al., 2005. Dynamic andequilibrium adsorption experiments. J. Chem. Educ. 82,919923.

    Gravier, C., Fayolle, J., et al., 2009. State of the art about remotelaboratories paradigms: foundations of ongoing mutations.Int. J. Online Eng. 4, 1925 (Copyright 2009, The Institution ofEngineering and Technology).

    Jing, G., Kettler, D., et al., 2007. A chemical engineering laboratoryover distributed control andmeasurement systems. Comput.Appl. Eng. Educ. 15, 174184 (Copyright 2007, The Institutionof Engineering and Technology).

    Lang, D., Mengelkamp, C., et al., 2007. Pedagogical evaluation ofremote laboratories in eMerge project. Eur. J. Eng. Educ. 32,5772.

    Murray, S., Lowe, D., et al., 2008. Experiences with a hybridarchitecture for remote laboratories. In: 38th ASEE/IEEEFrontiers in Education Conference, FIE 2008, October 22,2008October 25, 2008. Institute of Electrical and ElectronicsEngineers Inc., Saratoga Springs, NY, United States.

    Ogot, M., Elliott, G., et al., 2003. An assessment of in-person andremotely operated laboratories. J. Eng. Educ. 92, 5764.

    Rafael, A.C., Bernardo, F., et al., 2007. Virtual applications using aweb platform to teach chemical engineering: the distillationcase. Educ. Chem. Eng. 2, 2028.

    Rojko, A., Debevc, M., et al., 2009. Implementation, effectivenessand experience with remote laboratory in engineering

    education. Organizacija 42 (1), 2333.Seader, J.D., Henley, E.J., 1998. Separation Process Principles.

    Wiley, NewYork.Selmer, A., Kraft, M., et al., 2007. Weblabs in chemical

    engineering education. Educ. Chem. Eng. 2, 3845.Wankat, P.C., 1990. Rate-Controlled Separations. Elsevier Science

    Publishers, NewYork.Wiesner, T.F., Lan,W., 2004. Comparison of student learning in

    physical and simulated unit operations experiments. J. Eng.Educ. 93, 195204.

    Wiseman, C., Wong, K., et al., 2008. Operational experience with avirtual networking laboratory. In: 39th ACMTechnicalSymposiumon Computer Science Education, SIGCSE 2008,March 12, 2008March 15, 2008. Association for ComputingMachinery, Portland, OR, United States.

    Zwiebel, I., 1969. Fixed bed adsorption with variable gas velocitydue to pressure drop. Ind. Eng. Chem. Fundam. 8, 803807.

    http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0100http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0095http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0090http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0085http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0080http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0075http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0070http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0065http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0060http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0055http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0050http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0045http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0040http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0035http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0030http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0025http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0020http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0015http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0010http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005http://refhub.elsevier.com/S1749-7728(13)00010-9/sbref0005