006-1Kinetic study of scrap tyre pyrolysis and combustion

Applied Catalysis A:General197(2000)165–173Kinetic of liquid-phase reactions catalyzed by acidic resins:the formation of peracetic acid for vegetable oil epoxidationR.L.Musante1,R.J.Grau2,M.A.Ba1tanás∗Instituto de Desarrollo Tecnológico para la Industria Qu´ımica(INTEC)Güemes3450,3000Santa Fe,ArgentinaAbstractA heterogeneous kinetic model which takes into account the complete physicochemical interaction of reactive species in a polar liquid phase with an ion-exchange resin acting both as selective sorbent and heterogeneous catalyst has been employed to analyze the peracetic acid synthesis from acetic acid and hydrogen peroxide in an aqueous solution.Model parameters were estimated using uncoupled data from phase equilibria,polymer sorption,chemical equilibrium and reaction kinetics.Activities rather than molar concentrations in the polymer phase and specific(dry weight of catalyst basis)rather than volume-based expressions were found to give the best constitutive equations for the heterogeneous reaction rate.©2000Elsevier Science B.V.All rights reserved.Keywords:Vegetable oil epoxidation;Heterogeneous models;Acidic resins1.IntroductionThe epoxidation of unsaturated triglycerides with percarboxylic acids is a common practice for obtain-ing low cost plasticisers of good performance from natural and renewable sources such as vegetable oils. Cost constrictions dictate the use of inexpensive per-acetic acid(PAA)as the active reagent,while safety concerns demand an in situ preparation of the reagent, to avoid the handling of a pre-formed concentrated peracid.The in situ process involves a heterogeneous system,as epoxidation reaction occurs in the organic phase whereas the formation of PAA takes place ∗Corresponding author.Tel.:+54-342-4559175;fax:+54-342-4550944.E-mail address:tderliq@.ar(M.A.Ba1tan´a s)1Research Assistant of U.N.L.2Professor at U.N.L.and member of CONICET research staff.in an aqueous medium.The latter step is slow and controls the overall reaction rate;then,high conver-sions are generally achieved after several hours of reaction.Traditionally,homogeneous acidic catalysts(e.g. sulfuric acid)have been used to facilitate the forma-tion of peracetic acid by reacting acetic acid(AA) and hydrogen peroxide(HP):CH3COOH(AA)+H2O2(HP)H+←−−→CH3COOOH(PAA)+H2O(W) Hydrogen peroxide(an oxygen source)reacts with acetic acid(an oxygen carrier)in the aqueous phase in the presence of the acidic catalyst(solvated pro-tons)to give peracetic acid(PAA).The latter,in turn, transfers to the organic phase and quickly attacks the double bonds of the unsaturated vegetable oil(VO)to form the oxirane ring in the homogeneous epoxidation0926-860X/00/$–see front matter©2000Elsevier Science B.V.All rights reserved. PII:S0926-860X(99)00547-5166R.L.Musante et al./Applied Catalysis A:General197(2000)165–173 reaction:R1CH=CHR2 (VO)+CH3COOOH (PAA)→R1CHOCHR2(EVO)+CH3COOH (AA)The co-produced AA returns to the aqueous phase to close the sequence and re-start the production cycle. Mass and heat transfer limitations may impose severe process constraints(epoxidations are highly exothermic)or may lead to undesirable side reac-tions.So,numerous workers have studied the epoxi-dation reaction pathway and a qualitative picture of it was agreed upon in the past decades[1–3].The rate-limiting step of the in situ process is the forma-tion of PAA in the aqueous phase.However,quan-titative determination of the kinetic parameters was lacking,and only for two-phase epoxidations using homogeneous mineral acids more rigorous kinetic models had begun to emerge in the recent past[4,5]. As mentioned,secondary(acid-catalyzed)side re-actions do appear.Invariably,they involve an opening of the oxirane ring and,consequently,lower yields of epoxidized vegetable oil[5]:R1CHOCHR2+CH3COOHH+→R1CH(OH)CH(OOCH3)R2R1CHOCHR2+CH3COOOHH+→R1CH(OH)CH(OOOCCH3)R2R1CHOCHR2H +→R1COCH2R2R1CHOCHR2+H2O H+→R1CH(OH)CH(OH)R2R1CHOCHR2+H2O2H+→R1CH(OH)CH(OOH)R2 Heterogeneous catalysts such as functionalized micro-reticular ion-exchange resins(IER)can be advanta-geously used instead,since only the small carboxylic acid molecules can enter into their gel-like structure, while the bulky epoxidized triglyceride molecules cannot.The oxirane ring can thence be protected from the attack of the protons which are confined inside the gel matrix and,as a result,further ring decomposition is prevented.Also,catalyst recovery and/or regeneration is much easier then.Several such strongly acidic sulfonic IER(e.g.Dowex50;Am-berlite IR120,Chempro C20)have been reported to contribute to minimizing oxirane ring opening[6–9]. Three phases are present in the in situ processes cat-alyzed by IER:(1)a polymer phase,whose behavior (notably its swelling properties)is highly dependent on the physicochemical properties of the system,to-gether with(2)an aqueous phase,immiscible with(3) the organic phase.A heterogeneous kinetic model of this three-phase system using IER has not been pre-sented so far.We are presently developing a complete study of the epoxidation process,uncoupling the reac-tion and transport system into sub-systems of increas-ing complexity to allow a better quantification of the relevant process parameters.This piece of work presents the modeling of one of these sub-systems whose understanding is crucial: the kinetics of formation of PAA in the heteroge-neous aqueous phase-polymer(acidic catalyst)phase system.For this,the partition of each component between the two phases,as well as the swelling ra-tio(relative increase of the volume of the resin)are first quantified.Selective sorption/swelling leads to values of the relative compositions in the reaction locus(i.e.the polymer resin)and of the reactants’polymer-phase concentration which are different from those in the liquid phase.Both aspects are taken into account in the modeling.Then,the chemical equilibrium constant and kinetic parameters are esti-mated from the equilibrium and kinetics experimental data,respectively,using operating conditions typical of the industrial stly,the activation en-ergy of the acid-catalyzed reaction is estimated from non-isothermal experimental data.The appropriate-ness of managing the kinetic equations in terms of activities,rather than concentrations,and mass of dry resin,rather than its volume,are discussed.2.Experimental2.1.Activation and conditioning of the ion-exchange resinRohm&Haas Amberlite IR-120microreticular gel-type resin,DVB-styrene matrix,8%cross-linking, d p(dry)=215–775␮m(54.5%<530␮m;75.3%<R.L.Musante et al./Applied Catalysis A:General197(2000)165–173167600␮m;94.8%<670␮m),functionalized withsulfonic groups,was used throughout this work.AsIER is commercially available in sodium form,hy-drochloric acid(10%w/w)was used to fully activateit in successive ion-exchange steps(7in total),withfurther washing(1:10w/w)using distilled demineral-ized water until complete elimination of the residualstly,glacial acetic acid(purity:>99.7%w/w)was used to replace water inside the ion ex-change resin.The exchange capacity of the resin,as determinedby titration using conventional volumetric tech-niques,was:[H+]o=4.507meq/g(dry basis).The dry polymer density,as measured by pycnometry usingn-heptane[10],was1.437kg/m3.A portion of the dryresin was crushed and sieved;successive washing anddecantation in distilled demineralized water allowedthe removal of undesiredfines adhered to the crushedparticles.Two fractions,<53and>595␮m,wereused to evaluate the possible impact of mass-transferresistances on the process rate.2.2.Determination of phase equilibrium partitioningof reactants between the aqueous and polymericphasesDifferent dilutions of either acetic acid or hydro-gen peroxide in water(40g)were added to the dryresin(15g).The system was kept at333K with occa-sional stirring until physicochemical equilibrium wasachieved in about4h.The equilibrium compositionof the liquid phase was measured by volumetric tech-niques(see below).A portion of the swollen resin wascentrifuged at2000g for5min to eliminate any resid-ual interstitial liquid[10].The amount of sorbed liq-uid retained by the resin was found by weighting thecentrifuged resin before and after drying in a stove at378K for12h[11].2.3.Reactor and experimental proceduresThe rate of formation of PAA was studied in exper-imental runs conducted in a1000cm3closed cylin-drical Pyrex reactor,furnished with a variable speed(0–1000rpm)mechanical stirrer.The device had fourteflon baffles to eliminate vorticity,an overhead re-flux condenser and a fast-sampling device for collect-ing representative liquid-resin samples of the reacting system.An internal cooling coil(OD:1/4 ;316SS), combined with an external cylindrical heating mantle (400W)linked to a PID temperature controller,en-sured operating within±0.1K.In a typical experiment,glacial AA and pre-activated IER were added to the empty reactor and slowly heated to the reaction temperature under stirring.Si-multaneously,the desired volume of a solution of HP (30%w/w)was also heated to that temperature,under reflux,and then added,at once,to the reactor(zero time of the reaction).Samples were periodically taken and immediatelyfiltered to separate the resin and stop the reaction.Next,aliquots of thefiltrates(1cm3) were diluted in ethanol(100cm3)and refrigerated for further analysis by GLC.2.4.Analytical techniquesThe concentration of AA and HP were measured by volumetric titration,using NaOH(0.l N)and KMnO4 (0.1N),respectively.The contents of PAA acid in the reaction samples was determined by GLC after a previous derivatization with methyl p-tolyl sulphide; n-octadecane was used as internal standard[12].A glass column(500×2mm)packed with3%FFAP on Chromosorb W AW DMCS(80–100mesh)was em-ployed.3.Results and discussion3.1.Intra-and extra-particle mass-transfer resistancesThe absence of intra-particle mass-transfer resis-tances to any significant extent was experimentally corroborated using two widely differing particle sizes of the resin:>595and<53␮m,under typical reaction conditions of the well-stirred mixture(375rpm).Both runs gave identical results as is shown in Fig.1.Also, the ratio of initial rates was close to unity even though particle diameters differed by more than10-fold. Hence,extra-particle mass-transfer resistances are of no concern either and,thus,the commercial un-crushed lER beads were used to perform the kinetic study.168R.L.Musante et al./Applied Catalysis A:General 197(2000)165–173Fig.1.Influence of particle size of the ion-exchange resin (Amberlite IR-120)on the reaction rate and final concentration of PAA at 333K (reactants molar ratio:H 2O 2/HOAc =1.1/0.5).The full line corresponds to model predictions.3.2.Kinetic modelAs we have pointed out in Section 1,the selec-tive sorption and swelling of the resin can have a strong influence on the observable reaction rate of this two-phase system,even though no change in the se-quence of involved reaction steps (as compared to the homogeneous reaction)is to be expected for a strongly acidic material being the active catalyst.Then,a pseu-dohomogeneous kinetic model such as the following:dd t C PAA =k 1C IER C HP C AA −1K C PAA C W(1)which considers IER merely as a source of protons in the multicomponent liquid mixture would be certainly insufficient.In fact,for the set of experimental runs summarized in Table 2,every attempt we made to get satisfactory data fittings by means of a non-linear re-gression algorithm was unacceptable.So,a new model was developed to include explicitly the presence of the polymer phase.Two approaches can be followed to solve this hur-dle.The first one focuses exclusively on the catalytic reaction pathway at the reaction locus,where two ex-treme situations are recognizable according to whether slightly dissociated sulphonic groups or free solvated protons are the catalytic agents [13–15].The first case applies whenever the resin is imbedded into a slightly polar medium;the classic LHHW formalism can then be applied to model the reaction kinetics [16,17].Here,the inclusion of adsorption parameters is both justifi-able and flexible enough so as to fit up any observ-able changes in catalytic activity due to the presenceof polar components in minor amounts.If the resin operates into a strongly polar solvent such as water,the sulphonic groups are fully dissociated and the free solvated protons catalyze the reaction through ionic mechanisms of protonation.In many such situations,specially if just small amounts of resin are used,a pseudohomogenous model describing the reaction rate in terms of power-law kinetics suffices,as it does in homogeneous catalysis.However,none of these models takes into account the selective sorption and swelling of the resin,which leads to values of the reactants’polymer-phase con-centration and of the relative composition of both reactants and products in the polymer phase (i.e.the reaction locus)that are different from those in the liquid phase.A second approach is then needed,to ex-plicitly differentiate the compositions of the aqueous and polymer phases whenever a significant amount of resin is used as a catalyst.For such purposes,the overall system can be considered as a two-phase sys-tem composed of a highly viscous multicomponent fluid phase containing N +l species (with the swollen polymer as (N +1)th species,)in physicochemical equilibrium with the N -component liquid phase,since the characteristic time for reaching phase-equilibrium conditions between them is usually of the order of a few minutes [18].3.3.Activities of the species in the liquid phase A predictive model is needed to compute the ac-tivity of the four species in the aqueous phase,rather than a correlative one,due to the scant informationR.L.Musante et al./Applied Catalysis A:General197(2000)165–173169 available in the open literature about this reactingsystem.For these components,just a pair of binarydata is available:water–hydrogen peroxide[19]andwater–acetic acid[20].Then,the UNIFAC group con-tribution method[21]is a suitable tool,as it does notinvolve any adjustable parameter.The central con-cept of the method rests on considering any mixtureas a solution of functional groups interacting amongthemselves.Most of the extensions and refinementof the original concept have rested onfinding newprocedures for calculating the molecular interactionparameters between components,which initially camefrom vapor–liquid equilibrium data(UNIFAC VLE).Later,a new set of interaction parameters has beenderived from the liquid–liquid equilibrium data(UNI-FAC LLE)for making predictions related to thesesystems[22].More recently,the dependence of theinteraction parameters with temperature has also beenincluded,in the Modified UNIFAC method[23].We used each of these methods to predict the activ-ities in the aqueous phase.The UNIFAC LLE methodwas found to give the closestfit when a comparison ofmodel predictions with experimental data taken fromthe available vapor–liquid equilibria was made.It wasthen adopted for further use in our calculations.3.4.Activities of the species in the polymer phaseThe activity a P i of the i th species of a multicompo-nent polymeric solution can be evaluated in the frame-work of the extended Flory–Huggins model[25]:ln a P i=1+ln v i−N+1j=1m ij v j+N+1j=1χij v j−N+1j=1j−1k=1m ik v j v kχkj+ηV i53v1/3P−76v P(2)where N is the number of components excluding thepolymer,which is the(N+l)th species;νandνP are thevolume fractions of the i th species and of the polymerin the polymer phase,respectively;m ij is the ratio ofmolar volumes of the i th and j th species(m iP=0); V i is the molar volume of i th species;ηrepresentsthe number of moles of active elastic chains per unitvolume andχij represents the molecular interaction between components i and j.The latter parameters are known to be temperature-dependent[24]. According to Eq.(2),11adjustable parameters would have to be determined for calculating the set of a P i of the four species in the polymer phase:η,and 10molecular interaction parameters,four of them corresponding to interactions between the polymer and the other species(χiP)and six binary interaction parameters among the liquid phase components(χij), sinceχij=χji andχii=0(see Ref.[18]for a full discussion on the subject).The elasticity parameter andfive of the binary interaction parameters were found independently (i.e.uncoupled),from the sorption equilibria of the water–acetic acid and water–hydrogen peroxide bi-nary mixtures in contact with the resin.For this purpose,the activities of each pair of components in the aqueous phase were calculated using the UNIFAC LLE routine,which does not involve any adjustable parameter.Next,as in thermodynamic equilibriuma L i=a P i(3) Eq.(2)was used for each binary mixture,using the Levemberg–Marquardt algorithm,to estimate the cor-responding interaction parameters.Fig.2shows representative sets of tie lines,in adsorbent-free mass coordinates(N=g dry resin/g ad-sorbate),obtained experimentally at333K.These data clearly indicate that water is more strongly sorbed than either acetic acid or hydrogen peroxide,and that the resin swelling is much higher in water than in acetic acid.It is also apparent that the polymer beads can expand even more(albeit slightly)with higher concentration of hydrogen peroxide.Fig.3re-plots the experimental data(tie lines) shown in Fig.2,as molar fractions in the polymer and liquid phases,for the water–acetic acid and water–hydrogen peroxide binary pairs,together with model predictions using the calculated parameters (the latter are included in Table1).The agreement is satisfactory.The experimental determination of the partition of PAA was not made owing to the obvious experi-mental difficulties and hazards involved in handling a highly concentrated peracid.Instead,the heuristic assumption was made that in the polymer phase the analogs—carboxylic acids(acetic and peracetic)—can be assumed to behave as identical molecules,as170R.L.Musante et al./Applied Catalysis A:General 197(2000)165–173Fig.2.Binary sorption and phase partition equilibria on Amberlite IR-120of (a)water–acetic acid and (b)water–hydrogen peroxide binary mixtures,obtained experimentally at 333K,in adsorbent-free mass coordinates (N =g dry resin/gadsorbate).parison of experimental data and model predictions for the water–acetic acid (a)and water–hydrogen (b)binary pairs shown in Fig.2(333K).far as their molecular interactions are concerned,and so χAA −PAA =0.Likewise,an uncoupled estimation of the interaction parameters of the reactive binary pairs,PAA–water and acetic acid–hydrogen peroxide cannot be made,for obvious reasons.Nevertheless,under the previous hypothesis χW −PAA is identical to χW −AA .Lastly,the remaining interaction parameter can be estimated from chemical equilibrium data us-Table 1Estimated values of the interaction parameters of the extended Flory–Huggins model [Eq.(2)],at 333K.χij j i AA HP Water PAA Resin AA 0−0.00880.339000.1592HP −0.00360−1.9670−0.00360.8809Water 0.1039−1.488600.1039−0.6673PAA−0.00880.33900.1592ing asymptotic compositions (i.e.after long enough contact times)of the experimental runs.The proce-dure is described below,in Section 3.5.As is shown in Table 1,the complete set of χij inter-action parameters indicates relatively low interaction between acetic acid,PAA and hydrogen peroxide,and moderate interactions of these with water and of the four liquid components with resin.The value obtained for the elasticity parameter was:η=0.022mol/cm 3,which is about one order of magnitude higher than the theoretical one.However,by imposing progres-sively lower values to this parameter there is worse agreement between model predictions and experimen-tal data results,owing to the tight correlation among the parameters in Eq.(2).This problem has been dis-cussed in a recent work on the kinetics of liquid-phase esterification of acetic acid with ethanol using Am-berlyst 15,where about two orders of magnitude dif-ferences were encountered [18].Despite these shortcomings of presently available phase partition equilibrium models,they allow one to reproduce rather satisfactorily the experimental data and seem sufficient to help in describing the kinetic behavior of these systems under reaction conditions.3.5.Chemical equilibriumBy solving the set of multicomponent sorption equi-librium equations,Eq.(3),together with the mass bal-ances of each of the i th species:n L i +n P i =n 0i +νi ξ(4)R.L.Musante et al./Applied Catalysis A:General 197(2000)165–173171Table 2Initial loading and operating conditions of the batch experimental runs Run Temp (K)AA (moles)HP (moles)Water (moles)Resin mass (g)13331.329 2.73411.4707.44723331.345 3.48015.2058.66733332.783 2.73411.945.22643335.497 2.73411.9474.820533310.945 2.73411.9475.61663330.768 2.73411.94711.99373330.574 2.73411.94721.37583231.329 2.73411.9477.44793431.3292.73411.9477.447and the condition of chemical equilibrium:K =(a P PAAa P W /a P AA a PHP )eq (5)it is possible to jointly estimate the interaction param-eters for the reactive couple acetic acid–hydrogen per-oxide and the equilibrium constant,K .Indeed,owing to a lack of reliable data,it was impossible to estimate K from G 0data,as the standard free energy of the formation of PAA is known within broad uncertainty limits.The calculated value of χHP −AA is given in Table 1.At 333K the equilibrium constant was found to be K =2.18.Values ranging from 0.7to 5,which were found to be dependent both on the initial molar ratio of reac-tants and on the catalyst concentration,calculated from the equilibrium concentrations,have been reported by Rangarajan et al.[5].With our approach,for the broad set of experimental conditions given in Table 2,a good fit of the data could be achieved using a single value of K ,as is shown in Fig.4.The values of K at 323and 343K,obtained exper-imentally,were 1.911and 2.778,respectively.Fig.4.Experimental and calculated concentrations of PAA in liquid phase at equilibrium conditions (333K).3.6.Kinetic equationFrom Eq.(4)it is straightforward to recognize thatin the well-mixed isothermal batch reactor the time rate of change of the observable degree of advance-ment of the reaction (ξ)is sufficient to fully describe the reaction process whenever equilibrium conditions between the bulk liquid and the polymer phase hold.Also,because the catalyzed reaction proceeds only in-side the polymer phase:dd tξ=W P R P (6)where W P indicates mass of dry resin placed in the system (an invariable property)and R P is the specific reaction rate (dry weight basis).Another choice,which is to write Eq.(6)in terms of a volumetric reaction rate,leads to awkward rate ‘constants’,as the volume of the polymer phase continuously changes during the reaction owing to the swelling properties of the resin.In addition,as the system is highly non-ideal,the kinetic equation describing the reaction rate has to be written in terms of activities [26],accounting for the172R.L.Musante et al./Applied Catalysis A:General 197(2000)165–173Fig.5.Concentration of PAA in the liquid phase as a function of time for various values of the initial composition of the reacting mixture:(a)Run 7;(b)Run 6;(c)Run 3.Full lines represent model predictions.chemical equilibrium as well (Eq.(5)).The simplest expression which satisfies these requirements is the following:R P =ka P AA a P HP [1−K −1a P PAA a P W /a P AA a PHP ](7)where k [mol s −1(g dry resin)−1]=k 0[H +]0.Its value was estimated by means of the Marquardt–Levemberg algorithm,solving for Eqs.(2),(3),(5)-(7)and using the experimental data from the runs indicated in Table 2,all of them obtained in the absence of mass-transfer limitations.The previous calculation of the interaction parameters and of the chemical equilibrium constants at each temperature allowed the uncoupled estimation of k .The values of the interaction parameters at 323and 343K were obtained from those at 333K using the well-tested derivation of Flory [24]:χ(T )T =χ(T )T .The estimated values of the pre-exponential factor and activation energy of k are (8.48±l.l6)×10mol s −1(g dry resin)−1and 48.4±0.47kJ mol −1,respectively,for a 95%confidence level.Given that the absence of mass transfer constraints was corroborated,this some-what low E act value,as compared with the onereportedFig.6.Concentration of PAA in the liquid phase as a function of time for various values of the reaction temperature:(a)Run 8;(b)Run 1;(c)Run 9.Full lines represent model predictions.in homogeneous systems,might be due to the steric constraints imposed on the acid-catalyzed bimolecu-lar rds of the reaction [3,4]by the microreticular resin rather than an incomplete degree of solvation [27].Figs.5and 6compare experimental results and model predictions which are fair.Additional fittings were made using empirical kinetic expressions for:(a)the specific reaction rate (dry weight basis)in terms of molar concentrations instead of activities of the re-actants,and (b)the volumetric reaction rate in terms of activities of the components,factorized by the resin volume.In both such cases the new predictions were poorer than the one using the more sound Eq.(7),as the residual sum of squares were 58and 238%higher,respectively.4.ConclusionsA two-phase model has been proposed to describe the catalyzed reaction of the formation of PAA from acetic acid and hydrogen peroxide under a broadR.L.Musante et al./Applied Catalysis A:General197(2000)165–173173range of conditions,using a sulphonated ion-exchange resin acting as both a sorbent and a heterogeneous catalyst.The developed model incorporates relevant aspects with regard to the different affinities of the reactive species toward the liquid and resin phases. The selective partitioning of each component be-tween the two phases and the relative increase of the volume of the resin(i.e.its swelling ratio)with vary-ing composition were corroborated,quantified and taken into account in kinetic modelling.These fea-tures,which have adequate literature support,had not been previously considered for this particular reactive system.A progressive,uncoupled estimation of the model parameters was made using:(a)phase equilib-rium/sorption data of unreactive pairs of components, to obtain the binary interaction parameters;(b)chem-ical equilibrium data,to estimate the thermodynamic equilibrium constant and binary interaction parame-ters of the reactive pairs,and(c)reaction rate data in absence of mass-transfer resistances to estimate the specific kinetic rate constant.Activities rather than molar concentrations and specific(dry weight of cat-alyst based)rather than volume-based reaction rates were used throughout the work.For processing purposes the preferential partition of water inside the catalytic polymer phase(the reaction locus)is inconvenient because,being both a diluent and a reaction product,water lowers the rate of for-mation of PAA from the reversible reaction which is involved.Yet,the use of microreticular ion-exchange resins as heterogeneous catalysts for epoxidizing un-saturated triglycerides is desirable.Protons are then confined inside the polymer phase,which prevents their further attack on the oxirane ring and,so,higher oxirane indexes than those achievable in homogeneous catalytic processes can be realised. AcknowledgementsThanks are given to Leonardo Machaca González and Adolfo Larese for their dedicated experimen-tal work and to Dante L.Chiavassa for his selfless help in handling the computer programs.Thefi-nancial help of Universidad Nacional del Litoral,CONICET,ANPCyT(PD No.019)and JICA is grate-fully acknowledged.References[1]F.Greenspan,R.Gall,JAOCS33(1956)391.[2]H.Wohlers,M.Sack,H.LeVan,Ind.Eng.Chem.50(11)(1958)1685.[3]M.Abraham,R.Benenati,AIChE J.18(4)(1972)807.[4]T.Chou T,J.Chang,mun.41(1986)253.[5]B.Rangarajan,A.Havey,E.Grulke,P.D.Culnan,JAOCS72(10)(1995)1161.[6]R.Gall,F.Greenspan,JAOCS34(1957)161.[7]V.Nagiah,H.Dakshinamurthy,J.Aggarwal,Indian J.Technol.4(1966)280.[8]J.Wisniak,E.Navarrete,Ind.Eng.Chem.Prod.Res.Dev.9(1970)33.[9]B.M.Badran,F.M.El-Mehelmy,N.A.Ghanem,J.Oil ColourChem.Assoc.59(8)(1976)291.[10]B.M.P.Cornel,F.M.H.Sontheimer,Chem.Eng.Sci.41(7)(1986)1791.[11]M.Iborra,C.Fité,J.Tejero,F.Cunill,J.Izquierdo,ReactivePolymers21(1993)65.[12]F.Di Furia,M.Prato,U.Quintily,S.Salvagno,G.Scorrano,Analyst109(1984)985.[13]B.Gates,W.Rodr´ıguez,J.Catal.31(1973)27.[14]J.Tejero,F.Cunill,M.Iborra,J.Mol.Catal.42(1987)257.[15]J.Aragón,J.Vegas,L.Jodra,Ind.Eng.Chem.Res.32(1993)2555.[16]A.AI-Jarallah,A.Siddiqui,K.Lee,Can.J.Chem.Eng.66(1988)802.[17]A.Rehfinger,U.Hoffmann,Chem.Eng.Sci.456(1990)1605.[18]M.Mazzotti,B.Neri,D.Gelosa,A.Kruglov,M.Morbidelli,Ind.Eng.Chem.Res.36(1997)3.[19]G.Scatchard,G.Kavanagh,L.Ticknor,J.Am.Chem.Soc.74(15)(1952)3715.[20]J.Gmehling,U.Onken,Vapor–Liquid Equilibria DataCollection V ol I,Part I,DECHEMA,Frankfurt,1977. [21]A.Fredenslund,R.Jones,J.Prausnitz,AIChE J.21(6)(1975)1086.[22]T.Magnussen,P.Rasmussen, A.Fredenslund,Ind.Eng.Chem.Proc.Des.Dev.20(1981)331.[23]rsen,A.Fredenslund,Ind.Eng.Chem.Res.26(1987)2274.[24]P.J.Flory,Principles of Polymer Science,Cornell UniversityPress,Ithaca,NY,1953.[25]M.Mazzotti,V.Kruglov,B.Neri,D.Gelosa,M.Morbidelli,Chem.Eng.Sci.51(10)(1996)1827.[26]G.Froment,K.B.Bischoff,Chemical Reactor Analysis andDesign,2nd Edition,Wiley,New York,1990,p.39. [27]F.Ancillotti,M.Massi Mauri,E.Pescarollo,L.Romagnoni,J.Mol.Catal.4(1978)37.。

Detoxification ofandA,two abundantmycotoxins,by lactic acid bacteriaS.Fuchsa ,G.Sontag a ,R.Stidl a ,V.Ehrlichb,*,M.Kundi c ,S.Knasmu ¨llerbaDepartment of Analytical and Food Chemistry,Faculty of Chemistry,University of Vienna,Vienna,AustriabInstitute of Cancer Research,Department of Medicine I,Medical University of Vienna,Vienna,Austria cInstitute of Environmental Center for Public Health,Medical University of Vienna,Vienna,AustriaReceived 9July 2007;accepted 10October 2007AbstractAim of the present study was to investigate the detoxification of two abundant mycotoxins,namely ochratoxin A (OTA)and patulin (PAT)which are frequently found in human foods,by lactic acid bacteria.The removal of the two mycotoxins from liquid medium by thirty different LAB strains was analyzed in a screening trial by the use of HPLC coupled with UV-or fluorescence detection.Two highly effective strains were identified;Lactobacillus acidophilus VM 20caused a decrease of OTA by P 95%and Bifidobacterium animalis VM 12reduced PAT levels by 80%.Subsequently experiments showed that the binding of these compounds depends on different parameters,i.e.the concentration of toxins,the cell density,the pH-value and on the viability of the bacteria.To proof that the decrease of the toxins LAB from liquid medium results in a reduction of their toxic properties,micronucleus (MCN)assays were conducted with a human hepatoma cell line (HepG2).Indeed,a substantial decrease (39–59%)of OTA and PAT induced MCN formation was observed the most effective strains detected in the chemical analyses.Furthermore,also the inhibition of the cell division rates by the toxins was significantly reduced.These findings indicate that certain LAB strains are able to detoxify the two toxins and may be useful to pro-tect humans and/or animals against the adverse health effects of these compounds.Ó2007Elsevier Ltd.All rights reserved.Keywords:Patulin;Ochratoxin A;Detoxification;Lactic acid bacteria;Micronuclei;HepG21.IntroductionLactic acid bacteria (LAB)are widely used for the pro-duction of fermented foods and are also part of the intesti-nal microflora.Numerous investigations indicate that LAB have beneficial health effects in humans (Ouwehand et al.,2002;Saxelin et al.,2005).One of the effects identified is the protection against toxins contained in foods such as heterocyclic aromatic amines,polycyclic aromatic hydrocarbons,mycotoxins and reactive oxygen species (Knasmu ¨ller et al.,2001;Stidl et al.,2007a ).The Food and Agriculture Organization esti-mated that about 25%of human foods and animal feed are contaminated with mycotoxins and strong efforts have been made to decontaminate them by the use of physical and chemical adsorbents but the success made so far is limited (Huwig et al.,2001;Shetty and Jespersen,2006).It has been shown earlier by El-Nezami et al.(2000)that LAB detoxify aflatoxin B 1(AFB 1),which is the most potent known human carcinogen and contributes to the high prevalence of liver cancer in regions such as Central Africa and China (IARC,2002).Also with other fungal toxins such as zearalenone,trichothecenes and fumonisins,binding effects have been observed in chemical analytical0278-6915/$-see front matter Ó2007Elsevier Ltd.All rights reserved.doi:10.1016/j.fct.2007.10.008Abbreviations:AFB 1,aflatoxin B 1;BNC,binucleated cells;cfu,colony forming units;DMEM,Dulbecco’s minimal essential medium;DMSO,dimethylsulfoxide;FCS,foetal calf serum;FLD,fluorescence detection;HPLC,high performance liquid chromatography;LAB,lactic acid bact-eria;MCN,micronuclei;MRS,de Man-Rogosa-Sharpe;OTA,ochratoxin A;PAT,patulin;PBS,phosphate buffered saline;UV–VIS,ultraviolet–visible.*Corresponding author.Tel.:+431427765147;fax:+43142779651.E-mail address:veronika.ehrlich@meduniwien.ac.at (V.Ehrlich)/locate/foodchemtoxAvailable online at Food and Chemical Toxicology 46(2008)1398–1407investigations(Kabak et al.,2006;El-Nezami et al.,2002b, 2002a;Shetty and Jespersen,2006;Niderkorn et al.,2006).In the present study,the removal of patulin(PAT)from liquid medium by a variety of LAB strains was investigated for thefirst time.PAT is found sometimes in higher concen-trations as other mycotoxins in foods as it is produced by a large number of fungi within several genera such as Bys-sochlamys,Eupenicillium,Penicillium,Aspergillus and Paeci-lomyces(Speijers,2004)and is predominantly found in vegetables and fruits(in particular in apples and apple derived products)(Thurm et al.,1979;Pohland and Allen, 1970;Moureau,2002;Paster et al.,1995).Its acute toxic effects in mammals are primarily due to impairment of kid-ney functions(Speijers et al.,1988).It is also well docu-mented that PAT causes oxidative damage(Liu et al., 2007),has a negative impact on reproduction in males via interaction with hormone production(Selmanoglu and Koc-kaya,2004)and affects the immune system(Llewellyn et al., 1998;Sharma,1992).PAT caused potent genotoxic effects in experiments with mammalian cells(Alves et al.,2000; Schumacher et al.,2005;Schumacher et al.,2006)but is devoid of activity in bacterial mutagenicity assays(Wu¨rgler et al.,1991;Wouters,1996;Ueno et al.,1978).Since DNA-damage is causally related to cancer induction and a number of other diseases(Fenech,2002b;2002a),it was suspected that PAT may possess carcinogenic properties,but in older studies(which do not meet the current standards)negative results were obtained and newer investigations are not avail-able at present(Speijers,2004).In addition,also experiments with ochratoxin A(OTA), which is produced by several species of Aspergillus and Pen-icillium(Aish et al.,2004),were included in the present study. This compound is present in human foods like beer,cereals, wine,cocoa,dried vine fruits,coffee and spices(JECFA, 2001;Birzele et al.,2000;Fazekas et al.,2002).OTA causes severe nephrotoxic effects in animals and humans and its genotoxic and carcinogenic effects are well documented (Aish et al.,2004).It has been shown in a few earlier investi-gations that LAB cause removal of OTA from liquid med-ium(Piotrowska and Zakowska,2000;Piotrowska and Zakowska,2005;Turbic et al.,2002;Skrinjar et al.,1996).In the present investigation,the protective properties of LAB towards PAT and OTA were studied in co-incubation experiments with chemical analytical methods.In total,30 different LAB strains were included in a screening trial. Subsequently,the dose dependency,as well as the impact of the viability of the bacteria on the binding of the toxins was tested with selected strains.Since chemical analytical studies do not provide afirm proof for the detoxification of mycotoxins,as active metab-olites may be formed,additional experiments were carried out,in which the influence of the incubation of bacteria with toxins was studied in human derived liver cells(HepG2).We investigated if the treatment of the toxins with the bacteria has an effect on the inhibition of cell division and on the formation of micronuclei(MCN)which are formed as a consequence of chromosome breakage(clastogenicity)and aneuploidy(Fenech,2000).The cell line we used has retained the activities of a broad variety of drug metabolising enzymes which are normally lost during cultivation and it was shown earlier that it detects the effects of genotoxic car-cinogens,which give negative results in conventional in vitro test systems(Knasmu¨ller et al.,1998,2004b).Also mycotox-ins such as citrinin,fumonisin B1and OTA,which can not be detected in conventional in vitro assays,caused positive results in this experimental model(Ehrlich et al.,2002a, 2002b;Knasmu¨ller et al.,2004a).2.Materials and methods2.1.Chemicals and mediaDe Man-Rogosa-Sharpe(MRS)broth,was obtained from Oxoid (Bassingstoke,England),Elliker Bouillon was purchased from Difco (Sparks,USA).Patulin(4-hydroxyl-4H-furo(3,2c)-pyran-2(6H)-on,CAS No.149-29-1,>98%)and ochratoxin A((-)-N-[(5-chloro-8-hydroxy-3-methyl-1-oxo-7-isochromanyl)carbonyl]-3-phenyl-alanine,CAS No.303-47-9,>98%)as well as cytochalasin B and L-cysteine,which were used for the cultivation of Bifidobacteria,were purchased from Sigma–Aldrich (Vienna,Austria).Dulbecco’s minimal essential medium(DMEM),anti-biotics and trypsin were obtained from PAA Laboratoires GmbH(Linz, Austria).Foetal calf serum(FCS)was obtained from Gibco(Paisley, Scotland).Inorganic salts(P99%)and DMSO(P99.9%)came from Merck(Darmstadt,Germany).Acetonitrile and methanol which were used for the HPLC analyses were purchased from Fluka(Hannover, Germany).Ammonia solution(33%)was obtained from Riedel-de Haen (Seelze,Germany).Stock solutions of PAT and OTA for HPLC analyses were prepared in MeOH(0.1mg/ml),for MCN-assays the two toxins were dissolved in sterilefiltered DMSO(1.0mg/ml).Sodium phosphate buffer (50mM,pH5.0)was prepared by mixing a50mM Na2HPO4-solution with a50mM NaH2PO4-solution.2.2.Storage and cultivation of LAB strainsLAB strains were either obtained from Lactosan(Kapfenberg,Aus-tria)or provided by the Institute of Milk Hygiene,Milk Technology and Food Science of the University of Veterinary Medicine(Vienna,Austria). The strains were coded(LA–Lactosan;VM–Veterinary Medicine)and stored deep frozen atÀ80°C.For preparation of overnight cultures, Lactobacilli were grown in MRS(pH6.2,37°C),Bifidobacteria were cultivated in MRS(pH6.2,37°C)with addition of0.03%of L-cysteine and Streptococci were grown in Elliker Bouillon(pH6.8,37°C).The Bifidobacteria were cultivated under anaerobic conditions,i.e.in anaerobic jars obtained from Oxoid(Basingstoke,England)with Anaerocult A (Merck,Darmstadt,Germany).The colony forming units(cfu)in the overnight cultures were determined by turbidimetry at600nm(Niderkorn et al.,2006)with a spectrophotometer(DU640,Beckman Industries Inc, Fullerton,CA,USA).2.3.Storage and cultivation of HepG2cellsThe cells were kindly provided by F.Darroudi(University of Leiden, The Netherlands),stored in deep frozen portions atÀ20°C and cultivated in DMEM,supplemented with15%FCS and addition of1.0%penicillin/ streptomycin in250cm2cultureflasks(Kremsmu¨nster,Austria).2.4.Preparation of LAB-strains for HPLC-analysis andMCN-assaysBacteria cultivated in overnight cultures were harvested by centrifu-gation(4000rpm,4°C,10min).The supernatants were discarded and the pellets resuspended in10mlThe washing step wasS.Fuchs et al./Food and Chemical Toxicology46(2008)1398–14071399repeated three times.Subsequently,the pellets were resuspended in physiological saline to obtain thefinal concentrations.The adsorption assays were performed in Eppendorf(1.5ml,safe lock)vials.For HPLC-analyses,50l l of stock solutions of the mycotoxins,250l l of the bacterial suspensions(final concentration109cfu/ml)and700l l sodium phosphate buffer(50mM;pH5.0)were used.To obtain heat inactivated cells,bac-teria were autoclaved for20min at121°C.The vials were vortexed (Vortex Genie2,Scientific Industries,Bohemia,NY,USA)shortly(3s) and incubated(37°C,4h)on a horizontal shaker table(IKA,Staufen, Germany)with soft agitation(500rpm).Subsequently,the incubation was terminated by centrifugation(10000rpm,4°C,10min)and700l l of the supernatants were transferred to Eppendorf vials and stored at4°C.In a set of preliminary experiments it was excluded that the preincubation of bacteria with the liquid medium had any effect on the outcome of the chemical analyses.For MCN-assays,the initial incubation mixtures consisted of50l l of the mycotoxin solution,500l l bacterial suspension(final concentration of 5·109cfu/ml)and450l l sodium phosphate buffer(50mM;pH5.0).For PAT,the concentrations were25,50and100l g/ml.In the case of OTA, concentrations in the range between100and300l g/ml were used.After the incubation with the bacteria,the mixtures were centrifuged,and ali-quots of the supernatants(50l l in the case of PAT-mixtures and250l l of OTA-mixtures)added to HepG2-cultures grown in DMEM(final volume 5ml).2.5.Detection of patulin by HPLC–UVFive hundred microliter of the samples(supernatants,control samples or negative samples-which contained only bacteria and no toxin)were diluted with500l l mobile phase.The Merck Hitachi(Tokyo,Japan) chromatographic system consisted of a high pressure pump(L-6200A)and an auto sampler(AS-2000A).The column(BischoffChromatography SC-02-100Prontosil C18,100·2.1mm,3l m particles;Leonberg,Germany) was equilibrated with acetonitrile–water(1:99v/v)as mobile phase (Shephard and Legott,2000),40l l of the samples were injected and eluted with aflow rate of0.3ml/min and detected by UV–VIS spectrophotom-eter(L-4250,Merck Hitachi,Tokyo,Japan)at276nm.The calibration curves for PAT were obtained by daily analyses of eight standard solutions(range between25and1000ng/ml).The intra-day reproducibility for1000ng/ml was1.6%whereas the inter-day reproduc-ibility(measured over a period of5days)was1.75%.To determine the sample concentrations,the peak areas were calculated over the calibration curve.By subtracting the sample values from the values of the control samples,the decrease of PAT by LAB(in%)was obtained.2.6.Detection of ochratoxin A by HPLC-FLDTwo hundred microliter of the samples(supernatants,control samples or negative samples-which contained only bacteria and no toxin)were diluted with800l l mobile phase.The Merck Hitachi(Tokyo,Japan) HPLC-System contained a pump system(L-6200A)and an auto sampler (AS-2000A).An ACT column(Advanced Chromatography Technologies ACE3C18,15cm·3.0mm;3l m particles,Aberdeen,Scotland)with a guard column of the same material was used for performing the mea-surements.According to the protocol of Dall’Asta et al.(2004),an ace-tonitrile–ammonia buffer(NH3/NH4Cl,100mM,pH9.8)eluent(20:80v/ v),was used for the measurements.Samples(40l l)were injected and eluted with aflow rate of0.45ml/min and detected with afluorescence detector(F-1000,Merck Hitachi,Tokyo,Japan)at k ex=380and k em=440nm.The calibration curve was obtained by analysing the peak area of OTA standard solutions of six concentrations in the range between50and 1000ng/ml and was measured every day.The relative standard deviation for100ng/ml was1.49%within one day and1.75%over a period of5 days.To determine the concentration of the samples,the peak area was evaluated over the calibration curve.The decrease of OTA by LAB(in%)was determined by subtracting the sample value from the value of the control sample.2.7.MCN assaysThe MCN experiments were carried out according to the protocol of Darroudi and Natarajan(1991).Briefly,the cells were cultivated infilter topflasks(25cm2,Greiner,Kremsmu¨nster,Austria)for2days in CO2 atmosphere(8.0%)at37°C and95.0%relative humidity.Subsequently, the medium was changed and the cells were exposed to different concen-trations of the test compounds or the solvent(DMSO)for24h.Thefinal concentration of DMSO did not exceed1.0%(v/v).To evaluate the fre-quency of the MCN in binucleated cells(BNC),cytochalasin B(final concentration of 3.0l g/ml)was added to the growth medium after treatment and washing.Forfixation,the cells were trypsinized after28h, treated with cold hypertonic KCl solution(5.6g/l)and subsequently air dried preparations were made(Darroudi and Natarajan,1991).For the detection of the MCN in the binucleated cells,the slides were stained with Diff-Quik(Dade Behring Austria GmbH,Vienna,Austria).For each experimental point,three cultures were treated in parallel and1000cells were evaluated for MCN induction.To study the effects of the mycotoxins on cell division,the fractions of binucleated cells(in%)in relation to the number of mono-,tri-and tetranucleated cells were determined.2.8.StatisticsReduction of the mycotoxin concentrations after incubation in liquid medium with the different strains was calculated by Student’s t-tests,by comparison of the results of the individual experiments with the corre-sponding control samples.Correlation of the removal of the two toxins across strains and two concentrations were assessed by Spearmans rank correlation.Statistical analyses of the genotoxicity tests were performed by repe-ated measurements with one-way ANOVA in Prism version5.0(Graph-Pad Software,Inc.,Sandiago,CA;USA),followed by Bonferroni’s multiple comparison test.P-values with P60.05,were considered statis-tically significant.3.Results3.1.Chemical analytical measurementsThe results of the screening trial are summarized in Table1.The strongest decline of OTA(97%)was detected with a Lactobacillus acidophilus(VM20)strain,also two Bifidobacterium longum(LA02,VM14)strains were highly effective and caused a decrease by approximately50%. With PAT,the strongest effect(ca.80%decrease)was seen with a Bifidobacterium animalis(VM12)strain.Interest-ingly,no correlation between the detoxification capacities of the individual strains and the two mycotoxins was found i.e.the correlation coefficient was0.2and the P-value>0.05.In other words,strains which removed OTA were not effective against PAT and vice versa.However, it is notable that two strains,namely L.plantarum(VM 37)and L.curvatus(LA42),caused similar reduction of both toxins.In order to establish optimal conditions for the screen-ing trials,we conducted several experiments in which the pH dependency of the elimination of the mycotoxins was studied.The results of a representative experiment with PAT and OTA are shown in Fig.1.It can be seen that1400S.Fuchs et al./Food and Chemical Toxicology46(2008)1398–1407maximum removal was observed at pH 5.0,which was used in all subsequent experiments;no increase was observed in preliminary experiments with lower pH values (data not shown).It can be seen in Fig.2that the adsorption of the myco-toxins depends strongly on the amount of bacteria in the incubation mixtures.With both toxins,significant effects were seen when the cfu/ml was P 108/ml.The elimination from liquid medium depends also on the concentrations of the mycotoxins (Fig.3a and b).While OTA was removed from the medium even when high doses (up to 500l g/ml)were added,clear effects were seen with PAT only with substantially lower levels (6100l g/ml).To find out if,and to which extent,the viability of the bacteria affects their detoxification properties,we con-ducted comparative experiments with viable and heat inac-tivated cells.We found that the viability of the cells plays an important role as with heat inactivated cells only a mod-erate reduction of PAT (16%)and OTA (11%)was observed (data not shown).Table 1Removal of PAT and OTA from liquid medium by different LAB-strains a LAB-speciesStrainReduction (%)Patulin Ochratoxin A500ng1000ng 500ng 1000ng B.adolescentis LA187±1.095±1.39**7±1.3610±2.02**B.animalis VM1282±3.07**78±1.65**22±0.23**15±1.34**B.animalis LA1711±5.56*15±3.32**8±3.18*6±1.02B.bifidum VM1311±3.61**16±1.43**28±1.68**25±0.90**B.breve LA146±2.433±2.2516±1.25**17±2.64**B.longum VM147±2:914±1.1257±4.60**58±3.39**B.longum LA024±5.377±2.7752±4.61**54±1.32**L.acidophilus VM194±2.596±2.4218±1.71**12±0.37**L.acidophilus VM2021±3.07**18±1.65**97±2.11**96±1.91**L.acidophilus VM0519±4.15**19±2.65**32±2.03**31±1.30**L.brevis VM2116±3.67**16±2.40**4±1.934±1.58L.brevis VM223±4.199±1.63**11±1.49**5±1.82L.caseiVM0111±1.96**14±2.01**31±1.72**28±1.09**L.casei casei VM253±1.623±1.4112±1.90**15±0.44**L.casei casei LA095±1.10**4±2.015±3.172±1.27L.curvatus LA437±1.946±2.8315±1.72**16±2.42**L.curvatusLA4218±3.11*17±1.70**20±2.17**14±1.44**L.delbrueckii ssp.bulg.VM264±3.287±4.2329±3.40**27±2.38**L.delbrueckii ssp.bulg.VM2713±4.41**10±3.10**18±3.25**13±1.46**L.helveticusVM3410±2.49**8±2.40**24±2.18**25±2.80**L.delbrueckii ctis VM327±2.974±1.9434±1.21**34±0.84**L.paraplantarum VM3526±3.45**24±2.00**12±3.23**18±2.55**L.paraplantarum LA0720±3.13**16±2.11**3±4.394±1.15L.pentosus VM3631±3.26**30±2.98**22±7.45**18±7.46**L.plantarum VM3734±1.34**39±3.23**43±2.84**44±1.27**L.plantarum VM028±2.268±2.78**36±2.24**33±1.53**L.rhamnosus VM406±1.966±.3613±3.89**12±2.16**L.rhamnosus VM044±3.245±1.643±1.082±1.27L.slivarius LA016±3.807±1.835±2.239±1.75**S.thermophilusVM4221±2.78**19±1.43**13±.089**12±0.84**aThe incubation mixture contained 109cfu/ml and either 500ng/ml or 1000ng/ml of PAT or OTA,respectively;the mixtures were incubated at 37°C for 4h.Values indicate means ±SD obtained in three independent measurements.Stars indicate statistical significance to results obtained with corre-sponding control samples which contained only the toxins without bacteria (Student’s t -test,*P 60.05,**P 60.01).Fig.1.pH-dependent removal of PAT (white bars)and OTA (grey bars)from liquid medium by Bifidobacterium animalis VM12(in the case of PAT)and Lactobacillus acidophilus VM20(in the case of OTA).The incubation mixtures contained 109cfu/ml and either 1000ng/ml of PAT or OTA respectively;and were incubated at 37°C for 4h.Bars indicate results obtained in three independent measurements.Stars indicate statistical significance in comparison to the results obtained with pH 5.0(Student’s t -test,**P 60.01).S.Fuchs et al./Food and Chemical Toxicology 46(2008)1398–140714013.2.Impact of LAB on mycotoxin induced MCN formation In order to proof that LAB are able to protect human cells against the toxic effects of the two mycotoxins,the impact of the bacteria on MCN induction and on the inhi-bition of cell division by the mycotoxins in human derived hepatoma (HepG2)cells was investigated in a series ofexperiments.As negative control supernatants of incuba-tion mixtures without toxins were used.While for OTA data on acute toxic effects on HepG2cells were available (Ehrlich,2002),this was not the case for PAT.In the case of OTA a significant reduction of BNC was seen already with a concentration of 5l g/ml;with 50l g/ml the division rate declined by approximately 50%(Ehrlich et al.,2002a ).The acute toxicity of PAT was monitored over a broad dose range.The results indi-cate that the critical concentration of PAT lies between 1.0and 2.0l g/ml.After 24h exposure with the former dose,the viability of the cells was not significantly affected,while the higher dose caused a decline by approximately 60%.The results of MCN experiments with different dose lev-els of PAT are depicted in Fig.4a and b.It can be seen that the toxin caused a significant induction of MCN in the range between 0.25and 1.5l g/ml.PAT/ml (0.5l g)induced a 2.8-fold increase of MCN formation over the background level.Results of experiments with OTAareFig.2.Impact of the cell density of the removal of PAT (white bars)and OTA (grey bars)by Bifidobacterium animalis VM12(in the case of PAT)and Lactobacillus acidophilus VM20(in the case of OTA)from liquid medium.The incubation mixtures contained 109cfu/ml and either 1000ng/ml of PAT or OTA,respectively,and were incubated at 37°C for 4h.Bars indicate results obtained in three independent measurements.Stars indicate statistical significance in comparison to the results obtained with a cell density of 109cfu/ml (Student’s t -test,**P 60.01).Fig.3.(a and b)Concentration dependent removal of the two mycotoxins from liquid medium by LAB.The incubation mixtures contained 109cfu/ml and different concentrations of PAT or OTA,respectively,and were incubated at 37°C for 4h.Bars indicate results obtained in three independent measurements.Stars indicate statistical significance in com-parison to the results obtained with 1l g/ml (Student’s t -test,*P 60.05,**P 60.01).Fig.4.(a and b)Impact of PAT on MCN formation (a)and on cell division (b)in HepG2cells.The cells were incubated with different concentrations of PAT for 24h;subsequently the MCN-frequencies were determined after cytochalasin B treatment as described in Section 2.Bars indicate means ±SD of three incubation experiments.In total,1000cells were evaluated per experimental point.Stars indicate statistical signifi-cance in comparison to the results obtained without toxin (ANOVA,***P 60.001).1402S.Fuchs et al./Food and Chemical Toxicology 46(2008)1398–1407not shown as they were similar to the findings reported in previous articles (Ehrlich et al.,2002a;Knasmu ¨ller et al.,2004a ).The impact of pretreatment of the mycotoxins with LAB on MCN-frequencies and cell division rates in HepG2cells is shown in Figs.5and 6.It can be seen that the LAB caused a significant reduction of MCN induction by both toxins.In the case of PAT (Fig.5a),MCN frequencies were decreased by 39%,the effect caused by OTA (Fig.6a)was reduced by 59%at the highest concentrations.Also the inhibition of cell division (i.e.the decline of BNC forma-tion)by the two toxins was reduced significantly (Fig.5and 6b).4.DiscussionIn the present study,it was shown for the first time that LAB are able to remove PAT from liquid medium.Fur-thermore,the findings of earlier investigations were con-firmed,which indicate that the bacteria are able to eliminate OTA (Turbic et al.,2002;Piotrowska and Zak-owska,2005;Skrinjar et al.,1996)which is a potential genotoxic carcinogen (Ehrlich et al.,2002a;Cavin et al.,2007).It was also demonstrated that preincubation of the toxins with LAB reduces the toxic effects of these com-pounds in human derived liver cells (HepG2).As mentioned above,it is known that certain LAB strains are able to detoxify other mycotoxins such as AFB 1(for review see Kabak et al.,2006),fusarium toxins such as deoxynivalenol,nivalenol,T-2toxin,HT-2toxin (El-Nezami et al.,2002c;Niderkorn et al.,2006)and zea-ralenone (El-Nezami et al.,2002b,2004).Also genotoxic carcinogens contained in fried meat such as heterocyclic aromatic amines and polycyclic aromatic hydrocarbons are inactivated by LAB strains (Stidl et al.,2007a;Knasmu ¨ller et al.,2001).Fig.5.(a and b)Impact of selected LAB strains on MCN-formation (a)and inhibition of cell division (b)by PAT.5·109cfu/ml were incubated for 4h with PAT (25,50,or 100l g/ml);subsequently,the HepG2cells were exposed to 50l l of the supernatants for 24h.MCN frequencies and the rates of BNC were determined as described in Section 2.Bars represent means ±SD of three independent experiments.White bars indicate measurements with LAB;grey bars show results of the experiments without LAB treatment.In total,1000cells were evaluated per experi-mental point.Stars indicate statistical significance in comparison to the results obtained with control samples without toxins (ANOVA,**P 60.01,***P 60.001).Fig.6.(a and b)Impact of selected LAB strains on MCN-formation (6a)and inhibition of cell division (b)by OTA.5·109cfu/ml were incubated for 4h with OTA (100,200or 300l g/ml);subsequently,the HepG2cells were exposed to 250l l of the supernatants for 24h.MCN frequencies and the rates of BNC were determined as described in Section 2.Bars represent means ±SD of three independent experiments (one culture/experiment).White bars indicate results of measurements with LAB;grey bars show results obtained without LAB treatment.In total,1000cells were evaluated per experimental point.Stars indicate statistical significance in comparison to the results obtained with control samples without toxins (ANOVA,*P 60.05,**P 60.01,***P 60.001).S.Fuchs et al./Food and Chemical Toxicology 46(2008)1398–14071403。

JMEPEG (2001)10:567–575᭧ASM InternationalKinetics of Strain Aging in Bake Hardening Ultra Low Carbon Steel—a Comparison with Low Carbon SteelA.K.De,S.Vandeputte,andB.C.De Cooman(Submitted 22February 2000)The kinetics of the static strain aging process have been analyzed in a vacuum-degassed ultra low carbon bake hardenable (ULC BH)steel with a total carbon content of 20wt.ppm through measurement of the strength properties.The influence of prestrain and free interstitial carbon content has been studied.The kinetic results were compared with those of a BH low carbon (LC)steel.In the derivation of the time exponent and the activation energy,only the first stage of aging was considered.It was observed that,at all prestrain levels and matrix solute carbon contents,the initial aging process in the ULC steel obeyed the t 2/3kinetic law and the kinetics were not influenced by the changes in dislocation structure due to prestrain and the dissolved carbon content.In comparison,the aging process and the kinetics in the LC steel were found to be significantly influenced by the amount of prestrain.The presence of carbide particles in LC steels can modify the aging kinetics.atoms n t segregating to per unit length of dislocation in time t Keywords internal friction,kinetics,LC BH steel,prestrain,is given bystrain aging,ULC BH steel1.Introductionn t ϭn 03΂␲2΃13΂ADt k T΃23(Eq 1)Bake hardenable (BH)vacuum-degassed ultra low carbon where n 0is the number of solute atoms per unit volume,A is (ULC)steels (C Ͻ50wt.ppm)have recently received increased the interaction energy between a dislocation and solute atom,attention for autobody applications in the automotive industry.and D is the diffusion coefficient of the segregating solute at Compared to low carbon (LC)BH steels (C Ͼ100wt.ppm),the absolute temperature T .With the advance of the aging ULC BH steels have excellent forming properties and an process,Eq 1fails,however,to describe the kinetics,as it does increased strength that is achieved due to the aging during the not consider the solute depletion surrounding the dislocations,paint baking of the final product.These steels can be processed a fact which was also recognized by Cottrell.on hot dip galvanizing/galvannealing lines without an overaging In order to extend the applicability of the model to systems section,which is necessary for LC steels to allow cementite of supersaturated Fe-C solid solutions,Harper modified the precipitation.Bake hardening is essentially a strain aging proc-above equation to allow for the lowering of solute concentration ess resulting from the interaction between interstitial carbon in the matrix surrounding the dislocations as the aging pro-atoms dissolved in the matrix and the dislocations generated ceeds.[6]He assumed that the rate of segregation at any time during forming operation.The kinetics of the process is con-would be proportional to the solute concentration remaining in trolled by the long-range diffusion of interstitial atoms to the solution and obtainedstrain fields of dislocations.Atmospheres of interstitial atoms are formed in the vicinity of the dislocation cores.Further segregation of interstitials to the dislocations results in carbide W ϭn t n 0ϭ1Ϫexp ͫϪ3L ΂␲2΃13΂ADt k T΃23ͬ(Eq 2)precipitation.The most obvious manifestation of the strain aging process is the increase in the yield stress of the material at all solute levels and aging times.[1]Earlier investigations of the where W is the fraction of solute atoms already segregated in strain aging process in Fe-C alloys have established distinctly time t and L is the total length of dislocations per unit volume.the mechanisms and stages of the process.[2–4]With regard to Harper found a good agreement with his experiments for frac-the kinetics of the aging process,it is likely that the entire tions of solute depletion up to 0.90.The limitations to the aging process cannot be described by a single model.The initial Harper model have been reviewed by Baird.[7]As the Harper stage of the aging process was originally described by Cottrell model does not allow for the back diffusion from the core of and Bilby’s kinetic model.[5]According to Cottrell and Bilby,dislocations,it can only be valid for steels with low atmosphere during the atmosphere formation,the total number of solutedensities.In many strain aging studies in the Fe-C system,a generalized form of Harper’s equation has been used to derive the kinetics of the aging process.Equation 2can be rewritten asA.K.De andB.C.De Cooman,Laboratory for Iron and Steelmaking,Ghent University,9052Ghent,Belgium;and S.Vandeputte,OCAS N.V .,Research Centre of the SIDMAR Group,9060Zelzate,Belgium.ln (1ϪW )ϭϪ΂t␶΃nϭϪ(k t )n(Eq 3)Contact e-mail:bruno.decooman@rug.ac.be.where ␶is a temperature-dependent relaxation constant obeying an Arrhenius-type relation from which the activation enthalpy of the aging process can be derived.The kinetic parameters n and ⌬H derived from fitting experimental data to Eq 3have often been used to interpret precipitation mechanisms or to take into account dislocation inhomogeneities.Any deviation in the value of n from 2/3is generally regarded as being associated with a change in the precipitation mechanism of carbon on dislocations or in the matrix.[8–12]In BH ULC steel,a very low amount of carbon is retained in solid solution at the end of processing and the precipitation of iron carbides is unlikely to take place in this type of steel.[13]Moreover,because of the ultra low level of solute content,the C backdiffusion is expected to be insignificant.Therefore,the application of the Harper derivation should give an accurate description of the aging kinetics until the completion of the Fig.1Measurement of increase in yield stress ⌬␴due to strain agingatmosphere formation in ULC steels.Strain aging in ULC BH steel is technologically very important.Presently,attempts are being made to increase the bake hardenability of these steels through retention of more The parameter W has often been equated to the fractional solute carbon in the matrix by increasing the cooling rates after increase in yield stress,⌬␴/⌬␴max ,during the aging process,soaking in a continuous annealing cycle.In this work,the strain where ⌬␴is the increase in yield stress after aging for time t aging results of a BH ULC steel have been examined with and ⌬␴max is the maximum stress increment due to prolonged respect to changes in prestrain and solute carbon content aging.[1]The proportionality between W and ⌬␴/⌬␴max may be resulting from the application of different cooling rates after questioned when the aging process occurs in supersaturated annealing.The kinetic parameters n and ⌬H of the aging process solid solutions,where the strengthening may result from differ-were determined with the analytical models describing atmos-ent concurrent mechanisms,but it probably is a fair approxima-phere formation.The influence of the prestrain and the solute tion in the case of the ULC steels.carbon content on the time exponent n was evaluated.In the derivation of the kinetics,the increase in upper yield strength 2.2Hartley Modeldue to the aging process was taken into account rather than the solute segregation,as the former is the most consistent This is the only model available so far that allows the kinetics manifestation of the atmosphere formation process.This consid-of strain aging to be derived by measuring the changes in yield eration also stems from the fact that a very small amount of stress.Hartley described the increase in yield stress during carbon is required for atmosphere formation even for a highly aging as due only to the reduction of mobile dislocation length,deformed material.Considering the occupancy of one carbon which is proportional to the linear concentration of carbon on atom per atomic plane threaded by the dislocation at atmosphere the dislocations.Hartley proposed the following aging saturation,the amount of carbon required to saturate a disloca-kinetic equation:[14]tion density of ␳(m Ϫ2)in bcc ferrite can be calculated as(␴y Ϫ␴f )1/2(␴y ϩ␴f )ϭ⌬␴␴ϭK 1ϩK 2΂DtT΃2/3(Eq 5)[C]ppm ϭ8.9и10Ϫ15и␳(Eq 4)So,even for a large dislocation density of 1014/m 2,onlywhere ␴y is the upper yield stress after prestraining and aging,about 1ppm carbon is needed to saturate all the dislocations.␴f is the flow stress at the end of prestraining (Fig.1),t is Whereas in the past the C aging has been successfully analyzed the aging time,T is the aging temperature,D is the diffusion by means of internal friction (IF)or resistivity measurements,coefficient,[14]and K 1and K 2are constants for constant test no diagnostic tool is presently available to monitor accurately conditions,The slope S of the ⌬␴/␴versus t 2/3plot is given such extremely low levels of carbon segregation during atmos-by S ϭK 2(D /T )2/3.With D ϭD 0exp [Ϫ(⌬H /R T )],the activa-phere formation.tion energy ⌬H for carbon diffusion therefore can be easily obtained from the plot of ln (ST 2/3)versus 1/T .However,apart from dimensionality consideration,the phys-2.Application of Kinetic Modelsical interpretation of the use of the term 1/2(␴y ϩ␴f )is not clear in Hartley’s derivation.[14]Since in the derivation of Eq 2.1Harper Model5the degree of atmosphere saturation has been taken into account instead of the total fraction of solute segregating to From Eq 3,a plot of ln [Ϫln (1ϪW )]against ln t will result in a straight line with a slope n and a y intersect proportional to the dislocations,it was considered more appropriate to use the term ⌬␴/⌬␴atm for the degree of saturation in the present work,the diffusivity of the interstitial solute.Since k is expressed as k ϭk 0exp [Ϫ(⌬H /R T )],a plot of ln k versus 1/T will give where ⌬␴atm is the maximum increase in yield stress at atmos-phere saturation.It has been observed that the maximumthe activation energy of the aging process.Table1Chemical composition of the steels(in wt.%)during the fast heating(100ЊC/min)of the specimen in aninfrared radiator furnace over a temperature range of20to300 Steel C Mn P S Al Ti NЊC.[16]At40kHz,the Snoek peak occurs at around192ЊC.The inherent advantage associated with this technique is that ULC0.00200.090.0450.00300.04900.00700.0016it has a very high signal-to-noise ratio compared to the conven-LC0.0390.180.0350.00700.0540…0.0044tional torsion pendulum instrument.Hence,a very low amountof interstitials can be traced accurately with this instrument.Anadditional advantage of this technique is that,as the interstitial increase in yield stress⌬␴atm at atmosphere saturation is con-carbon segregates to the dislocations during heating to the peak stant for a prestrain level up to10%and aging temperatures temperature,one measures the actual interstitial C content in up to170ЊC for the ULC steel investigated.[13]the matrix at the paint baking temperature.Therefore,from Eq5,if the time exponent is set as n,thenthe slope of the ln⌬␴/⌬␴atm versus ln t plot will give the valueof n. 4.Results and Discussion4.1Aging Behavior—ULC and LC Steel and Effect of3.Experimental Procedure and MaterialPrestrain3.1Material and Processing Figures2(a)and(c)describe the aging behavior in theprestrained ULC steel with respect to time and temperature.It The material used for the present study was a vacuum-is observed that,at all prestrain levels,the increase in strength degassed ULC BH steel with the composition as given in Tablereaches a distinct saturation plateau after a time characteristic 1.The aging results for a LC BH steel with the compositionof the aging temperature.The strength then remains almost given in Table1were used to compare the kinetics of aging.[15]constant during further aging.Except for a marginal increase The hot-rolled sheets of both the ULC and LC steel werein strength in specimens prestrained2%at higher aging temper-given75%cold reduction in a laboratory cold rolling mill.atures,no second stage of hardening,i.e.,carbide precipitation, After cold rolling,the sheets were annealed in a(Carl-Wezel,can be seen.This suggests that the solute carbon available in Germany)continuous annealing simulator at850ЊC for60sthe matrix of the ULC steel(SC sheets)is sufficient only to with an overaging cycle of180s at400ЊC.The cooling ratecomplete atmosphere formation.The attainment of the satura-from the annealing temperature was10ЊC/s.The annealedtion plateau marks the end of Cottrell atmosphere formation,as sheets were further given a temper rolling reduction of1.3%.studied previously through the changes in yield point elongation Tensile specimens of80mm gage length were prepared(YPE)behavior.[13]Second,at all prestrain levels and tempera-from these sheets and were prestrained2,5,and10%at a straintures of aging,the maximum strength increase at the end of rate of4ϫ10Ϫ4sϪ1and then aged at temperatures betweenthe atmosphere formation,⌬␴atm,was found to beϷ30MPa. 50and170ЊC for different times in a silicone oil bath with aIn comparison,the aging in the LC steel has two distinct temperature control ofϮ1.5ЊC.stages(Fig.2d and e).A significant strength increase is observed For varying the amount of solute carbon content in the matrixin the second stage or the precipitation stage in the LC steel. of the ULC steel,the cold-rolled sheets were cooled from theThe maximum increase in strength decreases with the increase annealing temperature of850ЊC at three different cooling rates:in prestrain.(1)sheets cooled at the rate of10ЊCиsϪ1,(2)sheets cooled atthe rate of50ЊCиsϪ1to room temperature,and(3)sheets waterquenched from the annealing temperature at the rate of 4.2Effect of Cooling Rate550ЊCиsϪ1to room temperature.In the review of the agingFigure3shows the IF spectra observed for the SC,MC, results,these samples are designated as(a)SC(slow cooling,and FC specimens due to different rates of cooling.The carbon 10ЊCиsϪ1),(b)MC(medium cooling,50ЊCиsϪ1),and(c)FCcontent in the matrix increases with the increase in cooling (fast cooling,550ЊCиsϪ1).Tensile specimens prepared fromrate,as reflected in the IF spectra.The solute carbon content these sheets were prestrained5%and then aged at50ЊC forin the rapidly cooled specimens will be slightly higher than different times.what is measured considering the2min heating time neededto reach the peak temperature of measurement.Rapid cooling 3.2Mechanical Testing introduces some dislocations or vacancies in the material,and,hence,there is a possibility of losing some interstitial carbon The increase in yield stress⌬␴was determined as the differ-to the dislocations during the measurement.This effect is ence between the upper yield stress,␴y,after aging for time texpected to be very limited in slowly cooled specimens(Eq4). and the flow stress,␴f,at the end of prestraining based on theThe strain aging results of these specimens for an aging original specimen dimensions(Fig.1).temperature of50ЊC are shown in Fig.4(a)and(b)with The solute carbon content in the SC,MC,and FC specimensrespect to changes in yield stress and YPE.The distinct features was determined by IF measurements using a high frequencyrevealed in the aging results are as follows.piezoelectric ultrasonic composite oscillator.In this technique,the specimen is set to vibrate longitudinally over a piezoelectricoscillator at40kHz at a vibration strain amplitude of10Ϫ7.•With increasing interstitial carbon content,the aging stage The IF due to stress-induced ordering of interstitials is recordednow gradually advances to the second stage of aging andFig.2Increase in the yield stress with time for different aging temperatures for prestrained(a)to(c)ULC and(d)and(e)LC steelsa significant second stage hardening is observed in the FC saturation(as indicated by the maximum in the YPE values)is again30MPa in all the SC,MC,and FC specimens.and MC specimens.•The completion of the first stage of aging or the atmosphere4.3Kinetics of Agingsaturation occurs faster with the increase in carbon content,as revealed in the YPE results(Fig.4b).The aging results of Fig.2(a)to(c)were replotted in termsof Eq3and5and are shown in Fig.5(a)and(b),respectively.•The maximum increase in yield stress⌬␴atm at atmospherein Table2.It is clear from the figures that within the prestrainand temperature range studied,no change in the slopes isobserved.The values of n found through Harrier’s analysisfall within0.65to0.80,which are quite close to that derivedby Cottrell and Bilby for the interaction of the dislocationand the interstitial carbon for which nХ0.66.Values of nfound through Hartley analysis are also close to the valueof0.66.However,relatively higher values obtained throughHarper’s model were due to the neglect of saturation effectsin this model.The analysis of the kinetics through these models suggestsa normal strain aging kinetics(t2/3law),i.e.,carbon segrega-tion to dislocations alone,and that the kinetics is not alteredby the changes in the dislocation density in the ULC steelswithin the range studied.This is important since amplitude-Fig.3IF spectra in annealed ULC specimens cooled at different ratesdependent IF measurements on prestrained specimens demon-strated that prestraining the ULC steel in excess of7.5%results in dislocation structure changes.[17]The TEM observa-presence of cellular dislocation network formation.While thekinetics changes due to such dislocation structure changeshave not been reported so far,but based on Bullough andNewman’s analysis,a time exponent value of0.77had beenreported earlier[18]considering inhomogeneities in dislocationdistribution(in the region of clusters,cell walls,and carbides).The aging results of the LC steel were also analyzedthrough the Harper and Hartley models for comparison ofaging kinetics with those of ULC steel,and the results areshown in Fig.6(a)and(b),respectively.The n values calcu-lated from the data are given in Table2.It is clear that inthis case the amount of prestrain influences the values of nwithin the temperature range examined.The values of n forspecimens prestrained2%derived through the Harper(0.54to0.55)and Hartley(0.46to0.48)models are both lower (a)than those observed for the ULC specimens.The results pointmore toward a t1/2kinetic law.In specimens prestrained5%,words,at higher prestrain,the Cottrell atmosphere formationprocess dominates,whereas at lower prestrain,the n valuesare suggestive of a mechanism of carbon diffusion toward agrowing cementite particle.In a recent work by Kozeschnikand Buchmayr,[19]it was shown that within a prestrain rangeof0to5%there is a change in the precipitation mechanism.They indicated that,at low dislocation density,the precipita-tion of carbide is favored and,at about5%prestrain andmore,the ferrite matrix is depleted by Cottrell atmosphereformation and no carbide particles can form until at least themajority of carbon atoms have diffused to the dislocations.In earlier works,[11]a t1/2kinetic law had been found to beassociated with dislocation locking by carbon atoms at ferrite-cementite interfaces.The LC steel contains many cementite (b)particles[15]and also a higher amount of manganese than the Fig.4Aging behavior in ULC specimens with different cooling rates ULC steel.Therefore,it is likely that the carbide particles with respect to(a)increase in yield stress and(b)YPE can grow during aging and lead to an additional increase inyield stress.[8,20]Leslie[8]demonstrated that,during aging ofan Fe-Mn-C alloy at temperatures between60and100ЊC,precipitation of carbides both in the matrix and the disloca-Both the equations describe the data quite well up to thecompletion of atmosphere saturation at all the prestrain and tions could be observed as Mn shortens the incubation timeFig.5Kinetic analysis of the aging data of prestrained ULC specimens using (a )Harper and (b )Hartley modelsfor the formation of critical nuclei and affects the activity the solute carbon content (within 20wt.ppm),the aging mech-anism does not change.Earlier aging results with higher initial of carbon.The analysis of the aging data for specimens with different carbon content reported abrupt changes in aging kinetics in quenched-in iron alloys which were ascribed to the changes solute carbon contents (Fig.4)is given in Fig.7(a)and (b).The values of n measured from the slopes of these plots are in aging mechanism from nucleation on dislocation only to the nucleation within the matrix and dislocations.[8,9]The given in Table 3.It is interesting to note that the slopes are almost constant for all the specimen groups and are very close nucleation within the matrix is said to be facilitated by the presence of vacancy rings generated due to quenching.[8]Thisto the value of 0.66.This suggests that,even with changingaging process in the ULC and LC steels as a functionof prestrain and temperaturePrestrain,Aging temperature,⌬H,Steel Model%؇C n kcal/molULC Harper21400.6919.61000.70750.76500.7551400.6520.01000.65750.80500.72101400.7320.31000.74750.78500.73Hartley21400.5919.01000.76750.61500.6351400.6519.01000.79750.68500.61101400.8019.71000.71750.64500.66(a)LC Harper2500.5515.75500.6719.21000.80Hartley2500.4616.81000.485500.6421.81000.58was expected in the present case for FC specimens,but thepresent results suggest that,even if such a mechanism ispresent,it does not influence the aging process strongly.Thisis probably due to the fact that even the highest solute carbonin the FC specimen is too low to cause any matrix nucleation.The activation energy for the atmosphere formation processin ULC steel was calculated from Fig.5(a)and(b)using bothHarper and Hartley derivations,and the results are shown inFig.8(a)and(b),respectively,for the two models.The valuesare given in Table2and are in excellent agreement with theactivation energies of18to20.1kcal/mol for diffusion of carbonin bcc iron during strain aging,as published earlier.[6,21]Theactivation energies derived for the LC steel specimens appar-ently show a strong prestrain dependence.At higher prestrain,the activation energy derived(Table2)is close to that fordiffusion of carbon atoms to the dislocations,whereas at lowerprestrain,a much lower activation energy is found.This impliesthat the underlying aging mechanism in the LC steel is not thesame at all prestrains,a fact that was also revealed in the n(b)values(Table2).Fig.6Kinetic analysis of the aging data for prestrained LC specimensusing(a)Harper and(b)Hartley models5.ConclusionsIn the present work,an attempt was made to apply the to the aging results of an ULC and a LC steel,in order tocompare the aging kinetics for these steels and to obtain theavailable analytical models describing kinetics of strain aging(a)(a)(b)Fig.7Kinetic analysis of the aging data of SC,MC,and FC speci-mens using (a )Harper and (b )Hartley modelsTable 3Kinetic parameter n for the strain aging process in the ULC steel as a function of cooling rate(b)Model Prestrain,%Aging temperature,؇C Cooling raten Fig.8Determination of the activation energy of carbon diffusion during strain aging in prestrained ULC specimens using (a )Harper Hartley 55010ЊC иs Ϫ10.63and (b )Hartley model50ЊC иs Ϫ10.60550ЊC иs Ϫ10.63Harper55010ЊC иs Ϫ10.7150ЊC иs Ϫ10.76550ЊC иs Ϫ10.76•The maximum increase in yield stress at atmosphere satura-tion is 30MPa in the ULC steel,and this does not depend on the amount of prestrain or solute content within the range studied.kinetic parameters n and ⌬H .The kinetics were derived through measurement of the increase in yield stress due to aging.The •The activation energy for the atmosphere formation stage following conclusions can be drawn.in the ULC steel has been found to be 19to 20.3kcal/mol,which is in excellent agreement to the activation energy of •The time exponent n evaluated through different kinetic 18to 20.1kcal/mol for diffusion of carbon in bcc iron models suggested that the dislocation pinning by the carbon during strain aging reported previously in the literature.atoms is the dominant mechanism during the strain aging in the ULC BH steel at all prestrain levels and is not affected In the case of the LC BH steel,the dislocation density has by the changes in dislocation structure due to straining.a significant role in determining both the strengthening level after the second stage of aging and the kinetics of the initial •The amount of prestrain up to 10%or the changes in solute carbon content (up to 20wt.ppm)does not influence the aging process.At lower prestrain,the kinetics follows a t 1/2law,and at higher prestrain,the kinetics is governed mainly byaging kinetics in the ULC steel.11.V.T.L.Buono,M.S.Andrade,and B.M.Gonzalez:Metall.Trans.A, the dislocation and carbon atom interaction,which follows a1998,vol.29A,pp.1415-23.t2/3time dependence.12.R.Bullough and R.C.Newman:Proc.R.Soc.,1959,vol.A249,pp.427-40.13.A.K.De,S Vandeputte,and B.C.De Cooman:Scripta Mater.,1999,vol.41,pp.831-37.References14.S.Hartley:Acta Metall.,1966,vol.14,pp.1237-46.15.A.V.Snick,K.Lips,S.Vandeputte,BC De Cooman,and J.Dilewijns: 1.W.C.Leslie:The Physical Metallurgy of Steels,McGraw-Hill,New in Proc.on Processing and Properties,W.Bleck,ed.,Aachen,Ger-York,NY,1982,p.88.many,1998,Modern LC and ULC Sheet Metals for Cold Forming,2.W.Pitsch and K.Lu¨cke:Arch.Eisenhu¨ttenwes.,1956,vol.1,p.45.vol.2,pp.413-24.3.D.V.Wilson and B.Russell:Acta Metall.,1960,vol.8,pp.36-45.16.I.G.Ritchie and Z.Pan:33rd MWSP Conf.Proc.,ISS,Warrendale,4.P.Elsen and H.P.Hougardy:Steel Res.,1993,vol.64,pp.431-36.PA,1992,vol.29,pp.15-25.5.A.H.Cottrell and B.A.Bilby:Proc.Phys.Soc.,1949,vol.A62,pp.17.A.K.De,K.De Blauwe,S.Vandeputte,and B.C.De Cooman:J.49-62.Alloys Compounds,2000,vol.310(1–2),pp.405-10.6.S.Harper:Phys.Rev.,1951,vol.83,pp.709-12.18.J.D.Baird:Iron and Steel,1963,vol.8,pp.400-05.7.J.D.Baird:Iron and Steel,1963,vol.7,pp.368-74.19.E.Kozeschnik and B.Buchmayr:Steel Res.,1997,vol.68(5),pp.8.W.C.Leslie:Acta Metall.,1961,vol.9,pp.1004-22.224-30.9.R.H.Doremus:Trans.AIME,1960,vol.218,pp.596-605.20.T.Obara,K.Sakata,M.Nishida,and T.Irie:Kawasaki Steel TechnicalReport,1985,vol.12,p.25.10.S.I.Neife,E.Pink,and H.P.Stu¨we:Scripta Metall.Mater.,1994,vol.30,pp.361-66.21.C.Wert:Phys.Rev.,1950,vol.79,pp.601-05.。

轮胎压力监测中英文资料外文翻译文献This paper discusses the use of surface acoustic wave (SAW) XXX SAW sensors。

tire pressure XXX during every phase of driving。

The authors present their prototypes for measuring tire pressure。

the SAW sensors used。

and the XXX test rides.ning tires XXX human life while operating a road XXX Nowadays。

car manufacturers try to save space and weight by not including a XXX。

this can lead to XXX and the need for a tire change is still XXX punctures。

it is essential to measure tire XXX 20 grams。

causing high dynamic load。

A few years ago。

XXX。

These devices use an one port SAW delay line XXX devices that do not require a power supply or XXX。

XXX.In this paper。

we discuss XXX discuss the XXX。

XXX driving。

ensuring the safety of drivers and passengers on the road.Passive SAW pressure sensors operate as either a one-port delay line with XXX。

Oxidative stress,AGE,and atherosclerosisE Schleicher1and U Friess11Department for Internal Medicine IV,Clinical Chemistry(Central Laboratory),University of Tuebingen,Tuebingen,GermanyNumerous reports on the molecular mechanism of atherogenesis indicate an increase in oxidative stress, formation of advanced glycoxidation end products(AGEs), chronic inflammation,and activated cellular response particularly in diabetic patients.To elucidate the initiating and early accelerating events this review will focus on the molecular causes of the induction of these stress factors,their interactions,and their contribution to atherogenesis. Metabolic factors such as elevated free fatty acids,high glucose levels or AGEs induce reactive oxygen species(ROS) in vascular cells leading to ongoing AGE formation and to gene induction of proinflammatory cytokines.Vice versa, numerous cytokines found elevated in obesity and diabetes may also induce oxidative stress thus a circulus vitious may be initiated and accelerated.Increased production of ROS, mainly from mitochondria and NAD(P)H oxidase,stimulates signaling cascades including protein kinase C and mitogen-activated protein kinase pathway leading to nuclear translocation of transcription factors such as nuclear factor-j B(NF-j B),activator protein1,and specificity protein1. Subsequently,the expression of numerous genes including cytokines is rapidly induced,which,in turn,may act on vascular cells promoting the deleterious effects.From animal models of accelerated atherosclerosis a causal role ofNAD(P)H oxidase and the AGE/RAGE/NF-j B axis to atherogenesis is suggested.Because all factors involved form a highly interwoven network of interactions,the blockade of ROS or AGE formation at different sites may interrupt the vicious cycle.Promising candidate agents are,currently on trial.Most important to clinical practice,a number of drugs commonly used in the treatment of diabetes,hypertension, or cardiovascular disease,such as angiotensin-converting enzyme inhibitors,AT1receptor blockers,3-hydroxy-3-methyl-glutaryl-CoA reductase inhibitors(statins),and thiazolidindiones have shown promising‘preventive’intracellular antioxidant activity in addition to their primary pharmacological actions.Kidney International(2007)72,S17–S26;doi:10.1038/sj.ki.5002382 KEYWORDS:atherogenesis;cytokine;endothelial dysfunction;NAD(P)H oxidase;reactive oxygen species(ROS);glycation endproducts(AGE)Atherosclerosis and clinical consequences of macrovascular disease,such as myocardial infarction and stroke,are the major cause of death in the Western world and are increasing in developing countries.Numerous complex and multi-factorial risk factors have been documented.These include: genes,hypertension,diet,smoking,physical activity,hyperli-pidemia,insulin resistance,and chronic hyperglycemia. Particularly type II diabetic patients display accelerated and more advanced atherosclerotic lesions and an increased incidence of cardiovascular and cerebrovascular disease.1 The reason for the aggressive vascular disease associated with risk factors such as diabetes is still unclear.There is a large body of literature on the mechanisms of the development of atherosclerosis.Numerous data suggest the possible involvement of reactive oxygen species(ROS) and advanced glycoxidation end products(AGEs)as early initiating events occurring long before morphological changes are obvious.Therefore,as a working hypothesis and as outlined in Figure1,we propose that the vascular wall can readily cope with physiologically occurring stress factors. However,chronic action of stress factors and metabolic conditions,such as hyperglycemia,dyslipidemia or(satu-rated)free fatty acids(FFAs)elevated in diabetes may induce ROS and/or AGEs which via the receptor for AGE(RAGE) form a vicious cycle of increasingly accumulating AGEs, enhanced formation of ROS,and sustained activation of the RAGE axis thus forming a milieu to perpetuate the injurious effects of chronic inflammation within the vascular wall. Therefore,this review will particularly focus on the causes of accelerated atherosclerosis in diabetic patients because understanding the pathogenesis in this disease may help to elucidate the molecular mechanism in general. CONCEPTS OF THE PATHOGENESIS OF ATHEROSCLEROSIS To describe the initial mechanism(s)of the atherosclerotic process,three major concepts have been developed during the last decades.2In the‘Response to injury’hypothesis Ross and Glomset3proposed endothelial denudation as initial step causing increased endothelial adhesiveness for thrombocytes and leukocytes concomitant with a change of the antic-oagulant to a procoagulant milieu.The adherent thrombo-cytes and leukocytes release mitogens and vasoactive factors (cytokines and growth factors),which in turn lead to migration of smooth muscle cell(SMC)into the intima and to recruitment of monocytes/macrophages into the arterial wall.Upon uptake of low-density lipoproteins(LDLs)&2007International Society of NephrologyCorrespondence:Department for Internal Medicine IV,Clinical Chemistry (Central Laboratory),University of Tuebingen,Hoppe-Seyler-Street3,D-72076 Tuebingen,Germany.E-mail:Erwin.Schleicher@med.uni-tuebingen.dethe macrophages form ‘foam cells’,a hallmark of early arterial lesions.However,the observation that an intact endothelium may cover early atherosclerotic lesions has questioned this hypothesis.Besides,LDL entry rate into the arterial wall is dependent on the location of the arteria rather than on the presence of denuded endothelium.Therefore,the ‘Response to retention ’hypothesis has been introduced.This theory is essentially based on observations showing that after entry of LDL into the subendothelial space LDL is tightly linked to components of the extracellular matrix particularly to negatively charged proteoglycans.The LDL particle may then be modified by various lipolytic and proteolytic enzymes rendering these modified LDLs prone to macro-phage and smooth muscle uptake and subsequent formation of foam cells.On the basis of the observation that oxidatively modified but not native LDL lead to cholesterol accumula-tion in macrophages via the scavenger pathway has prompted the ‘Oxidative modification ’hypothesis.In this theory,Steinberg et al .4proposed that oxidative modification of LDL leads to enhanced uptake of modified LDL and subsequent foam cell formation.Furthermore,the modified LDL supports the recruitment of circulating monocytes.Invading monocytes are cytotoxic to endothelial cells thus ‘injuring’the vessel wall.Together,oxidative stress,post-ribosomal modifications and inflammatory reactions are a common feature in the three hypothesis for describing early atherosclerosis.But none of them explains all features of early events of atherosclerosis or the failures of antioxidative treatments,for example vitamin E,in human subjects.2The progression from initial vascular injury towards the fully developed atherosclerotic plaque is better understood:recruited thrombocytes and leukocytes release cytokines,vasoactive peptides,and growth hormones.This promotes the proliferation and migration of SMCs into the intima.The build-up of lipids within the arterial wall is thought to trigger further inflammatory response.Recruited macrophages release ROS and,in turn,take up oxidized lipoproteins.Together,the continuing process of lipid accumulation,foam cell formation,recruitment of inflammatory cells and releaseof ROS,cytokines,growth factors,and proteolytic enzymes perpetuates the inflammatory cascade within the vessel wall and finally sets the stage for cellular necrosis,fibrosis and further progression towards the fibro-fatty plaque,the characteristic lesion of end-stage atherosclerosis.CLINICAL FEATURES OF EARLY STATES OF VASCULAR INJURY:ENDOTHELIAL DYSFUNCTIONThe earliest clinical determinant of arterial changes is endothelial dysfunction as determined by flow-associated vasodilatation 5and endothelial dysfunction emerges as a key component in atherogenesis.Endothelial dysfunction is clearly associated with insulin resistance which occurs particularly in obese individuals and type II diabetes.6Then,what is the link between obesity and atherosclerosis?Subjects with accumulation of visceral adipose tissue show,as well,increased ectopic lipid accumulation in liver and skeletal muscle as determined by 1H-NMR.Particularly,the liver fat contributes to insulin resistance.The ectopic lipid deposition may lead to enhanced production of inflammatory cytokines (e.g.,via enhanced oxidative stress)in the affected organ in parenchymal cells and/or infiltrated macrophages as has been shown to occur in adipose tissue.7The tissue crosstalk occurring via the vascular system may be the link between obesity and macrovascular disease.Is there any relation between endothelial dysfunction and oxidative stress?In vitro and in vivo studies have demonstrated that hyperglycemia-induced endothelial dysfunction and endothelium-depen-dent vasorelaxation is contrasted by antioxidants such as vitamin C.8Furthermore,many of the agents with beneficial effects on endothelial dysfunction including statins,angio-tensin-converting enzyme inhibitors or angiotensin II type 1receptor blockers exert significant intracellular antioxidant action.9Together,accumulating data suggest a link between endothelial dysfunction and oxidative stress supporting a potentially causal role of ROS in the initial events of atherosclerosis.OXIDATIVE STRESS AND ATHEROSCLEROSIS ROSAlthough the term oxidative stress is widely used a generally accepted definition is lacking.While most researchers consider oxidative stress as an increased production of short-lived,highly reactive,oxygen-derived molecule species (see below),Sies 10describes oxidative stress as a ‘disturbance in the pro-oxidant/antioxidant balance in favor to the pro-oxidant state’.Several ROSs have been suggested to play central roles in vascular physiology and pathophysiology,the most important of which are superoxide anion (O 2KÀ),hydrogen peroxide (H 2O 2),nitric oxide (NO),and peroxynitrite (ONOO À).Although O 2KÀper se can affect vascular functions,it is also pivotal in generating other reactive species.When O 2KÀis produced in concert with NO K,they rapidly react to form the highly reactive molecule ONOO KÀ.ONOO KÀis an important mediator of lipid peroxidation and protein modification including oxidationAccelerated atherosclerosisAtherosclerosisRepair/regenerative responseVascular stressVascular stress + tissue injuryVascular stress + tissue injury+ risk factors (diabetes)Figure 1|Working hypothesis for the accelerated atherosclerosis in diabetes.Reactive ROS and AGEs that react via the RAGE are generated if the vascular wall is stressed.Because all three partners are linked a vicious cycle is induced.This cycle is enhanced by further stress factors occurring in diabetes.E Schleicher and U Friess :Oxidative stress,AGE,and atherosclerosisof LDL.In the absence of immediately accessible NO K,O2KÀis rapidly converted into H2O2by superoxide dismutase, which then is converted into H2O by either catalase, peroxiredoxins,or glutathione peroxidase.H2O2can also react with reduced transition metals,for example Fe2þ,to form the highly reactive hydroxyl radical(OH K),or it can be metabolized by myeloperoxidase to form hypochlorous acid (HOCl).In summary,ROS can react rapidly and unspecifically with a wide variety of biomolecules but their lifetime and correspondingly their radius of action is very limited.Apart from the short lifetime of ROS effective antioxidant defense mechanism(s),particularly through the redox enzymes,have been developed to detoxify these agents.Furthermore,recent reports indicate that hemoxygenase1is involved in ROS elimination.In diabetes,however,antioxidative defense is impaired.11Biomarkers of oxidative stress in the atherosclerotic lesion There is definite evidence of oxidative stress in atherogenesis because numerous lipoxidation end products(ALEs)or glycoxidation end products(AGEs)have been found to be increased in atherosclerotic lesions as compared to unaffected tissue.Such biomarkers originate from a wide variety of vascular tissue constituents,such as lipids,fatty acids,sugars, amino aids,proteins,or DNA(reviewed by Stocker and Keaney2).Different types of fatty acid oxidation products have been identified in human atherosclerotic lesion.Of these,hydroxyl products of linoleic acid(C18:2D9,12)or hydroxyoctadecaenoic acids and hydroxyl products of arachidonic acid(C20:4D5,8,11,14)or hydroxyeicosatetrae-noic acids are the most abundant type of oxidized lipids in atherosclerosis,12mostly present as cholesterol ester.Other important lipoxidation products in human atherosclerotic lesions are F2-isoprostanes,which occur at levels20-to40-fold lower than hydroxylated FFAs and represent radical oxidation products of arachidonic acid.12Moreover,several end products of glycoxidation(AGEs)have been found to accumulate in the atherosclerotic lesion,including hydro-imidazolones,pentosidine,and,most abundantly,N e-carboxymethyllysine(CML)as discussed below in detail.The processes of lipoxidation and glycoxidation are closely interlinked by the production of reactive carbonyl inter-mediates:the oxidation of polyunsaturated fatty acids can lead to the formation of glyoxal,methylglyoxal,malondial-dehyde,or4-hydroxynonenal,which are widely used as markers of lipoxidation.Likewise,the oxidative decomposi-tion of reducing sugars,such as glucose,fructose,ribose,and of glycolysis intermediates gives rise to glyoxal,methylglyox-al,and3-deoxyglucosone.13All of the carbonyl intermediates are prone to react with the N e-lysyl residues to modify proteins posttranslationally.Specific antibodies have been raised against the corresponding modifications,for example against4-hydroxynonenal or malondialdehyde adducts of apolipoprotein B100or against the CML modification and used to study the kinetic and distribution of ALE/AGE modifications in atherosclerosis and other processes,includ-ing diabetes,uremia,and aging.14–17Further evidence for oxidative protein modifications in the atherosclerotic lesion comes from amino-acid analysis. Compared with normal arteries,oxidized lipoproteins and atherosclerotic samples contain higher concentration of several oxidative amino-acid products(i.e.methionine sulfoxide,o-tyrosine,m-tyrosine,3-chloro-tyrosine,3-nitro-tyrosine,hydroxyleucine,hydroxyvaline,and N e-CML.2,18–20 To monitor the action of oxidative stress on DNA,markers as 8-oxo-deoxyguanosine,1,N2-etheno-20-deoxyguanosine,and N-carboxyethylguanosine have begun to emerge.21Sources of ROS generationAs indicated in Table1,virtually all types of vascular cells were found to produce O2KÀand H2O2upon specific stimulation or in response to physical or metabolic stress. In addition to mitochondrial sources,O2KÀcan be generated by several enzymes including the vascular isoforms of NAD(P)H-oxidase(vascular NOX isoforms),‘uncoupled’NO synthase(s),cytochrome P450oxidase,xanthine oxidase, and cyclooxygenase.The most important enzymatic source is thought to be membrane-associated NAD(P)H-oxidase,33,34 which generates intracellular O2KÀand H2O2in vascular SMCs,endothelial cells,andfibroblasts.However,Nishikawa et al.24found that mitochondria are a major source of ROS in high glucose-treated endothelial cells.ROS induction by cytokines and growth factors Numerous mediators,cytokines,and growth factors have been shown to induce O2KÀproduction(Table1).These include angiotensin II,tumor necrosis factor-a,interleukin-1, platelet-derived-,epidermal-and transforming growth factor, and macrophage inducing factor1most of which via the activation of NAD(P)H oxidase.Moreover,ligand-stimulated RAGE activation(see below)results in the generation of ROS, at least in part,also via activation of the NAD(P)H oxidase.22 However,the origin of the cytokines is unclear.Different tissues may produce and secrete inflammatory cytokines,for example by adipose tissue(adipokines),by liver(hepato-kines),by muscle(myokines),and other sources.It is estimated that approximately30%of circulating interleukin-6is derived from adipose tissue.ROS induction by FFAsElevated triglycerides and elevated FFAs(and altered FFA pattern)are an earlyfinding in obese individuals with insulin resistance and diabetic patients often correlating with endothelial dysfunction.35In these patients at high risk for developing vascular disease,elevated FFA plasma levels derive from excessive dietary fat intake and/or increased adipose tissue lipolysis.The abnormally high plasma levels of particularly saturated fatty acids,palmitate and stearate, induce ROS in endothelial and aortic SMCs as determined by electron spin resonance30or redox-sensitive dyes.31,29,32 Because ROS formation could be prevented by NAD(P)HE Schleicher and U Friess:Oxidative stress,AGE,and atherosclerosisoxidase inhibition but not by blockade of xanthine oxidase, NO synthase,or mitochondrial electron transport and because ROS formation is also prevented by inhibition of protein kinase C,these results suggest a fatty acid-induced, protein kinase C-mediated NAD(P)H oxidase activation with subsequent ROS generation.Enhanced ROS formation leads to activation of nuclear factor-k B(NF-k B).Moreover,FFA-induced NF-k B activation has been observed to mediate cytokine gene expression(interleukin-6)in endothelial cells and to induce endothelial apoptosis.36,37NF-k B activation induced by saturated fatty acids has also been shown to occur in human myotubes.38Palmitate-induced ROS generation is also occurring in CHO cells.39,40The authors provide evidence that lipotoxicity is prevented if either the metabo-lism of the saturated FFA or the deposition in triglyceride is stimulated.Molecular mechanism of ROS generation by high glucose levels in cultured vascular cellsGlucose is taken up by cells via facilitate glucose transport by plasma membrane integrated glucose transporters(family of GLUTs).The GLUTs differ in their location and particularly in their biochemical characteristics and regulation.While essentially all cells,including vascular cells,express GLUT1 for basal glucose transport,the insulin-sensitive GLUT4is essentially only expressed in muscle and adipose tissue and therefore responsible for the majority of the insulin-stimulated glucose uptake after a meal.Both GLUTs are characterized by a low K m,approximately2–5m M.41Glucose transport rate is at about50–70%of maximal capacity at 5m M,65–82%at10m M,and73–88%at15m M extracellular glucose concentration(Figure2).Thus a doubled or tripled increase in extracellular glucose levels leads to an enhanced glucose uptake rate of approximately22or33%only.In liver glucose is taken up by the low affinity(K m¼22m M),high capacity GLUT2which is essentially dependent on glucose concentration.Although these data indicate that vascular cells are protected from excess glucose uptake when extracellular concentrations are high,several reports show an high glucose-induced increase in endothelial ROS content by electron spin resonance30or an redox-sensitive dye.31To elucidate the molecular mechanism analogs of D-glucose were used.Although elevated L-glucose concentrations wereTable1|Formation of ROS in vascular cells induced by different stimuli and metabolic factors(modified from Stocker and Keaney2and supplemented)Stimulus Cell type Source ROS Effect(s)Reference Ligands of G-protein coupled receptorsAngiotensin II EC,SMC NOX O2KÀ,H2O2NOÁformation,hypertensionCell hypertrophy,p38activation2RAGE/NF-k B activationAGEs,CML-albumin EC NOX,mitochondrial H2O2Proinflammatory,proadhesive andprocoagulant gene expression(e.g.VCAM-1,tissue factor)22,23Growth factorsPDGF SMC,FB NOX O2KÀ,H2O2Growth,induction of MCP-12EGF FB NOX H2O2Unknown2TGF-b1EC,FB NOX H2O2Growth inhibition2 CytokinesTNF-a EC,FB NOX O2KÀ,H2O2Unknown2IL-1b EC,FB NOX O2KÀ,H2O2Unknown2Physical stressShear stress EC eNOS+/Àunknown OONOÀ,O2KÀJNK activation2 Metabolic conditions(high glucose,elevated FFAs)HG EC,SMC,arterial rings NOX,mitochondrial O2KÀ,ROS Activation of PKC,NF-k BUpregulation NOX subunitsAccumulation of AGEs,sorbitol,inhibited bystatins24–28Palmitate EC NOX ROS Upregulation of NOX subunits,inhibited bypitavastatin29PalmitateÀ/+HG,EC,SMC NOX,mitochondrial ROS PKC activationInvolvment of AMP kinase,inhibited bymetformin,30,31FFAsÀ/+HG EC Unknown O2KÀEnhancement of HG induced O2KÀproduction by saturated4unsaturated FFAs32AGE,advanced glycoxidation end product;CML,carboxymethyllysine;EC,endothelial cell;HG,high glucose;FB,fibroblast;FFAs,free fatty acids;NF-k B,nuclear factor-k B; NOS,endothelial nitric oxide synthase;NOX,NAD(P)H oxidase isoforms;PKC,protein kinase C;RAGE,receptor for AGE;ROS,reactive oxygen species;SMC,smooth muscle cell; TNF-b2,tumor necrosis factor-b2.E Schleicher and U Friess:Oxidative stress,AGE,and atherosclerosisineffective,30m M of the non-metabolizable analogs D -2-deoxy-glucose or D -3-O -methyl-glucose,like D -glucose,induced ROS within 15–30min.The ROS induction is mediated by activation of NAD(P)H oxidase because inhibition of this enzyme completely prevented the high glucose-induced endothelial ROS formation (Schleicher and Klopfer,unpublished results).These results indicate that elevated extracellular glucose concentrations need to enter the cell for ROS generation but need not to be metabolized.According results have been reported by Lee et al .42in mesangial cells,however they found,similar to Nishikawa et al .,24that mitochondria are also an important source of ROS under hyperglycemic conditions.Together,the data indicate that hyperglycemia induces ROS in endothelial cells via enhanced uptake of glucose but not through increased metabolism of glucose although the detailed molecular mechanism remains unclear.Biological effects of ROSThe effects of O 2KÀand H 2O 2on vascular function depend critically on the amount produced.Intracellular H 2O 2has been shown to modulate the activity of kinases and phosphatases in several signaling pathways by the oxidation of regulatory cysteine residues and there is emerging evidence that mammalian cells do produce H 2O 2physiologically to regulate cell proliferation,differentiation,and migration.43When formed in subtoxic amounts,H 2O 2can act as a second messenger modulating such responses as growth of vascular SMCs and fibroblasts or the expression of inflammatory mediators and matrix components.34Higher amounts of ROS can cause oxidative modification of lipids,proteins,and DNA resulting in significant toxicity,DNA damage,and even apoptosis.One major consequence of increased oxidative stress wall is the increased formation of ALEs/AGEs (see also below).Several lines of evidence indicate that these products have awide variety of extra-and intracellular effects to pertubate vascular structure and cellular functions as recently reviewed by Goldin et al .44First,AGEs may alter properties of the major matrix proteins collagen,vitronectin,and laminin through AGE–AGE intermolecular crosslinks resulting in increased stiffness of the vasculature.Second,AGE/ALE modification of LDL and apolipoprotein B100leads to altered LDL metabolism.Adducts of lysine residues with malondialdehyde and 4-hydroxynonenal in apolipoprotein B-100alter LDL properties to become a ‘high uptake’form of LDL,which cannot longer interact with the receptor for native LDL and instead is recognized by scavenger receptor on macrophages.2Likewise,it could be demonstrated that CML-modified LDL,formed upon reaction of glyoxal and LDL,is not taken up via the LDL receptor.45Third,AGE formation on the extracellular matrix has been shown to alter not only vascular structure but also cellular function.Glycation of key molecules in the basal membrane such as laminin and type I and type IV collagen weakens adhesion of endothelial cells.AGE modification of matrix glycoproteins leads to disparities in proliferation,growth,and secretory activity of different vascular cell types,46thus resulting in altered liberation of cytokines and growth factors.The other important consequence of increased oxidative stress appears to be the direct modulation of redox-sensitive transcription factors and signaling pathways.Studies have suggested that the affinity of certain transcription factors,particularly NF-k B and activator protein 1(AP-1),for their cognate DNA binding site can also be directly modified by ROS.34,47Intracellular ROS regulate several classes of genes relevant to vascular pathology,including antioxidant en-zymes,vasoactive substances,cytokines,and growth factors.Although the exact mechanisms are not elucidated yet,many of the corresponding promoters contain NF-k B-and/or AP-1-responsive elements.34,47In endothelial cells it has been shown by the use of various antioxidants that cytokine-induced expression of the vascular cell adhesion molecule-1(VCAM-1),intracellular adhesion molecule-1,and the chemotactic factor monocyte chemoattractant protein-1involves NF-k B in a redox-dependent fashion.34,47GLYCATION,GLYCOXIDATION,AND FORMATION OF AGEs Formation of AGEsThe discovery of hemoglobin HbA1c and its structurally identification has shown that glucose reacts non-enzymati-cally with free amino groups of proteins yielding a Schiff base which,after Amadori rearrangement,forms a rather stable 1-amino-fructose moiety.The term glycation has been coined for this post-ribosomal modification of proteins.Further studies showed that this reaction is not restricted to hemoglobin and that most abundant plasma proteins such as albumin,g -globulin,fibrinogen,48high-density lipopro-tein,and LDL 49are glycated according to the average plasma glucose level of the respective patient.Glycation of plasma proteins depends also on their half-lives.50,51Furthermore,extracellular matrix proteins from lung,aorta,nerve,V max20510********0102030405060708090100Extracellular glucose concentration G ex (m M )R e l a t i v e g l u c o s e t r a n s p o r t r a t e V r e l (%)V max K m = 22 m MK m = 5 m M K m= 2 m MGLUT-2GLUT 1/4Figure 2|Vascular cells are protected from excessive glucose uptake.Vascular cells express GLUTs with K m between 2and 5m M (e.g.GLUT 1and GLUT 4),whereas the K m of GLUT 2,which is expressed in liver and the pancreatic b -cell,is 22m M .The graphs show the dependency of the relative glucose transport rate from glucose concentration.Vertical lines indicate glucose concentrations of 4,7.5,and 15m M (80,135,and 270mg/dl),respectively,to indicate the ranges of daily variations of blood glucose levels in nondiabetic and diabetic subjects.E Schleicher and U Friess :Oxidative stress,AGE,and atherosclerosisglomerular basement membrane,52and lens53are glycated to an extent which correlates to previous hyperglycemia. Although various attempts have been made to show an effect of protein glycation on the respective protein functions most of the studies used an extent of glycation which is far beyond that occurring in(patho)physiological conditions.54 Cerami’s group recognized that glycation(fructosamine moiety)is not stable under physiological conditions yielding fluorescent compounds after several weeks.55Because the products were derived from glycated proteins the term advanced glycoxidation end products(AGE product)was coined.Meanwhile,a number of AGE products have been identified(for review see Thornalley56).Because the AGE intermediates glyoxal and methylglyoxal may also be formed from lipoxidation the term ALE,in analogy to AGE,was suggested by Januszewski et al.57In addition methylglyoxal may derive from glycolysis intermediates.58Although a hyperglycemia/diabetes-associated increase in AGE/ALE in various cells/tissues has been shown in vitro and in experimental animals as(reviewed by Thorpe and Baynes59 and Vlassara and Palace60)and particularly in cultured endothelial cells,61,25there are only a few reports on enhanced accumulation of AGEs in the vessel wall.In an immunohis-tochemical study using a polyclonal antiserum specific for the CML modification,a marked increase of CML accumulation was observed in atherosclerotic plaques and in foam cells. Moreover,staining was more pronounced in diabetic patients.16,62CML forms early on in atherogenesis and dense intracellular CML deposits were found in macrophage-derived foam cells in the stages of intimal thickening and fatty streak lesions.15Schalkwijk et al.63studied CML accumulation in diabetic and nondiabetic heart.Staining with a monoclonal CML-antibody was approximately sixfold higher in hearts from diabetic patients as compared to control hearts.CML deposition was localized in endothelial and SMCs of small intramyocardial arteries that showed no morphological abnormalities.In patients with acute myo-cardial infarction,CML depositions were threefold increased compared with controls in the small intramyocardial blood vessels and predominantly colocalized with activated en-dothelium(E-selectin-positive)both in infarction and noninfarction areas.63From their data the authors concluded that CML might reflect an increased risk for acute myocardial infarction rather than being a result.Noteworthy,AGE/ALE modification is not limited to protein DNA may be modified too.Accordingly,N-carboxyethylguanosine staining was increased in aortas of diabetic and uremic patients.21 BIOLOGICAL EFFECTS OF AGE:ACTIVATION OF THE AGE/ RAGE/NF-j B-AXISPrevious results showed that AGE may form crosslinks between proteins leading to alteredflexibility and digestibility of the collagen matrix in vascular wall and in skin.However, with the characterization of the receptor for AGE(RAGE)it became clear that AGE may exert their biological effects via amplification through a receptor-coupled signaling pathway.Although other receptors for AGE have been described,most available data have been collected with RAGE.64,65This receptorfirst described as binding protein for AGE by A.M. Schmidt66is a multiligand member of the immunoglobulin superfamily.AGEs may interact with endothelial RAGE to activate cellular events such as upregulation of the transcrip-tion factor NF-k B,67activation of NAD(P)H-oxidase22and activation of p38MAP kinase and ERK1/2MAP kinase cascades.Recently S100/calgranulins,a family of calcium-binding polypeptides that accumulate at sites of chronic inflammation,and HMGB1(amphoterin)have been identi-fied as important activators of the RAGE/NF-k B axis.NF-k B is translocated to the nucleus where it increases transcription of RAGE itself and a number of genes relevant to atherogenesis including VCAM-1,intracellular adhesion molecule-1,tissue factor,endothelin-1,and likely,proin-flammatory cytokines including interleukin-1b,interleukin-6,and tumor necrosis factor-a(as reviewed by Goldin44). AGE/RAGE-mediated gene expression involves,at least in part,intracellular ROS formation.Basta et al.23reported that exposure to CML-albumin led to an increased surface expression of VCAM-1in endothelial cells and to intracel-lular ROS formation by a RAGE-dependent mechanism.Both effects were inhibited by specific anti-RAGE antibody and by inhibition of the NAD(P)H-oxidase.The stimulation of RAGE gene expression by RAGE-mediated signaling indicates triggering of a vicious cycle.Once increased ROS formation is started it may be perpetuating or even enhanced if the stimuli are chronically present.Taken together,the production of ROS,AGE,cytokines,and other gene products involved in atherogenesis are interlinked and may be as outlined in the scheme shown in Figure3.LESSONS FROM ANIMALSFirst insights on how ROS and AGE/ALE formation might interfere with cellular stress response and inflammation,thus promoting progression of the atherosclerotic lesion,came from experimental animal models.The ApoEÀ/Àmouse model allows to dissect the contribution of hyperglycemia,of ROS generating enzymes(such as NADPH oxidase)or of the AGE/RAGE/NF-k B axis to atherogenesis and was used by several groups.ApoEÀ/Àmice are hyperlipidemic with markedly elevated cholesterol and triglyceride levels and display initial atherosclerotic lesions including fatty streaks and foam cell formation as early as at8weeks of age. Advanced atherosclerosis is evident at15weeks of age. ApoEÀ/Àmice were additionally rendered diabetic,either by streptozotocin,68representing a type I diabetes equivalent,or by breeding into a db/db background,69,70representing a type II diabetes model,to study the influence of hyperglycemia and diabetes as additional risk factors.In both combined hyperlipidemia/diabetes models development of athero-sclerosis was accelerated and more advanced.Increased lesion complexity of the aortic vessel wall was found to be accompanied by enhanced AGE formation,enhanced RAGE expression and increased levels of the proinflammatoryE Schleicher and U Friess:Oxidative stress,AGE,and atherosclerosis。

Aero-thermal performance improvements of unshrouded turbines through management of tip leakage and injection flowsJie Gao *,Qun Zheng,Zhengyi Zhang,Yuting JiangCollege of Power and Energy Engineering,Harbin Engineering University,Harbin 150001,Chinaa r t i c l e i n f oArticle history:Received 25August 2013Received in revised form 12March 2014Accepted 15March 2014Available online 13April 2014Keywords:Cavity tipTip leakage flow Tip injectionTip clearance height Injection mass flow rate Aero-thermal performancea b s t r a c tThe tip leakage flow not only is responsible for a signi ficant amount of aerodynamic losses in a turbine stage,but also leads to high heat-loads on the tip region.The paper presents a numerical investigation of in fluences of tip injection on aero-thermal performance of the tip leakage flow for both flat tip and cavity tip in an unshrouded turbine rotor,in an attempt to improve the turbine blade tip aero-thermal per-formance by the management of tip leakage and injection flows.The effects studied include the sensi-tivities to the geometrical clearance height and the injection mass flow rate.The results show that,at all tip clearances,tip injection has a good effect on the control of the leakage flow,and it signi ficantly re-duces the sensitivities of turbine performances to the effects of the tip clearance height.With tip in-jection,cavity tip geometry does not play a major role in the turbine performance improvement,but it obtains the good film-cooling performance on the blade tip.Tip clearance height,blade tip geometry and injection mass flow rate have related effects on the blade tip aero-thermal performance,and the turbine blade tip obtains the best aero-thermal performance with an optimum injection mass flow rate.Ó2014Elsevier Ltd.All rights reserved.1.IntroductionTurbomachinery has seen widespread use in the industry (Wu et al.[1],Gomes et al.[2]and so on);however,the tip clearances between rotor tips and the casing wall,which are necessary to prevent rubbing,result in an undesirable loss of ef ficiency,driven by the pressure difference between the blade pressure side (PS)and the suction side (SS).And,tip leakage flow is believed to be detri-mental to the performance of turbomachinery (Booth et al.[3]and Mohamed and Shaaban [4]).The work output of the rotor decreases because the leakage flow passes through the tip gap without being properly turned and expanded.Thus,tip leakage flow acts as an obstruction and leads to loss generation inside the rotor passage.Booth et al.stated that a tip gap of about 1%of the rotor span could cause a loss of 1e 3%on the stage ef ficiency.Moreover,another important consequence is the deterioration of the blade tip surface,due to regions of high local heat transfer in the narrow gap caused by flow separation and reattachment.Metzger and Rued [5]per-formed fundamental studies and showed that the generated tip leakage flow increases in the heat transfer of about 200%near thetip gap.Therefore,it is important to take measures to minimize the impact of tip leakage flow.In previous studies,several different methods were investigated to control the effects of the tip leakage flow.These methods are generally divided into passive control (no energy import)and active control (energy import)from the viewpoint of the energy.Most of the passive control methods reported involve modi fication of the tip surface geometry in an effort to decrease the leakage mass flow rate,including winglets (Zhou et al.[6]),squealer tips (Lee and Chae [7]),tip surface optimization (Maesschalck et al.[8]),casing treatments (Gao et al.[9])and so on.Of these methods reviewed only the cavity tip (double squealer tips)finds actual application in service.Double squealer rims offer the advantages of minimizing contact surface area and partly shielding cavity surface from ther-mal degradation.The other methods presented added cooling re-quirements that may not be easily addressed.Generally the active turbine clearance flow control can be ach-ieved by means of mechanical force,plasma actuator and tip in-jection.In order to reduce the blade tip clearance as far as possible,a first generation mechanically actuated active clearance control (ACC)system has been designed and fabricated (DeCastro and Melcher [10]).The system utilizes independent actuators,a segmented shroud structure,and a clearance measurement feed-back to provide fast and precise active clearance control throughout engine operation.In addition,the control of the tip leakage flow*Corresponding author.Tel.:þ8613895710245.E-mail addresses:gaojie_d@ ,gaojie_d@ (J.Gao).Contents lists available at ScienceDirectEnergyjournal ho me page:www.elsevier.co m/locate/energy/10.1016/j.energy.2014.03.0600360-5442/Ó2014Elsevier Ltd.All rights reserved.Energy 69(2014)648e 660using plasma actuators mounted to a turbine blade tip was inves-tigated in a linear cascade facility by Van Ness et al.[11].The active control results indicated that the actuators led to qualitative changes in the structure of downstream wake profiles;the tip leakageflow can lead to a significant increase in emission and specific fuel consumption in the high-pressure turbine.The tip injection has showed potential success for increasing the aerodynamic performance around the blade tip regions.Chen et al.[12]studied the aerodynamics of tip injection using a two-dimensional model.They concluded that when the coolant is ejected normal to the tip,it is able to block the leakageflow.Li et al.[13]carried out a numerical simulation of the effects of tip injec-tion.They suggested that the jet could obstruct the tip leakageflow and weaken the interaction between tip leakageflow and main passageflow,improving the turbine efficiency by0.41%when the clearance gap is small.Rao and Camci[14e16]conducted an experimental study of a turbine tip desensitization method based on tip injection in a large-scale rotating turbine rig,and took the effect of the injection mass flow rate and the injection location into consideration.It was found that tip injection could cause the tip leakage vortex to be reduced, and its associated losses decreased to the level observed for the case with half the tip gap height.And,the tip injection at81%blade axial chord was most successful in reducing the total pressure deficit in the leakage vortex.It was shown that such information would be useful in optimizing tip injection schemes.Niu and Zang [17e19]performed parametric investigations of tip injection in a high-turning axial turbine linear cascade,including injection circumferential angle and injection location in the blade thickness direction.The results showed that the nearer to the PS corner the injection holes are,and the smaller the injection circumferential angle is,the more tip injection can affect the tip leakageflow.Be-sides,injection chordwise location also plays an important role in the redistribution of the secondaryflow within the cascade pas-sage,and the holes located in the aft part of blade can perform much better than those in the front part.Yet,the effect of the relative blade-casing motion on theflowfield and heat transfer was not investigated here.Based on the above investigations,they[20] investigated the influence of tip injection at design and off-design incidences.It was found that even at off-design conditions,tip in-jection can still perform well in reducing the tip leakage massflow and its associated losses.However,the potential thermal benefits were not investigated in their research.The tip coolant injection has also been used to protect the blade tips from the high gas temperatures.Hohlfeld et al.[21]and Couch et al.[22]conducted experimental and computational in-vestigations on blowing from the dirt purge holes of aflat tip in a linear cascade.They stated that theflow ejected from the dirt purge holes is able to block the tip leakageflow.The obstruction effect by tip injection performed quite well for a small tip gap,but it was not effective for a large tip gap.It should be noted that,the main focus of their research was the thermal performance of the tip.Newton et al.[23]provided detailed measurements of heat transfer condition on the tip of a general cooled turbine blade,and reported that the injection at the location of separation bubble could result in a better cooling benefit.Kwak and Han[24]inves-tigated thefilm cooling of gas turbine blade tips in a linear cascade. Blade tip heat transfer coefficients were found to increase with the tip clearance height increasing.Thefilm cooling from radial holes along the camber line was shown to be more effective at larger clearances and at higher blowing ratios;however,the aerodynamic performance of tip cooling is unknown.Also,Acharya et al.[25] computationally simulated thefilm cooling of turbine blade tips. Thefilm cooling at three different tip clearances was simulated.The tip injection was found to alter the leakage vortex and also to decrease the heat transfer coefficient along the coolant trajectory;film-cooling effectiveness was found to increase slightly with the tip clearance size.Nowadays,some modern,unshrouded turbine blades have some sort of squealer tips.Mhetras et al.[26]investigated the ef-fects of tip injection on heat transfer conditions on a squealer tip. They stated that due to the combined effect of squealer tip and tip injection,significant improvements could be obtained on the cavity floor and inner rim walls,and a larger cavity depth or higher blowing ratios could give higher effectiveness on all tip surfaces, whereas the overall mixing of cooling air and hot leakageflow was rarely mentioned.Han et al.[27,28]investigated thefilm cooling of bothflat tip and cavity tip in a linear cascade.They found that the film cooling ejections on both plane tip and squealer tip dramati-cally affect theflow behaviors at the tip region,and the coolant injected from the holes on the blade PS produced an additional reduction in the blade tip heat transfer coefficient.Besides,the cooled cavity tip is superior to the cooledflat tip in terms of the thermal performance;however,this labyrinth andfilm cooling design can lead to a coolant accumulation in the squealer cavity.In order to allow the accumulated coolant in the cavity to escape,and meanwhile to cool the tip TE(the trailing edge),a piece of rim wall of PS squealer near the TE is cut.The literature mentioned above were focused either on the aerodynamic performance of the tip leakageflow with tip injection or on the thermal performance of cooled blade tips.Only few in-vestigations have been focused on the aero-thermal performance of the tip leakageflow with injection as well as the combined effect of cavity tip and tip injection.The objectives of this paper are to systematically investigate the physics and loss mechanisms of the tip leakageflow with tip injection;to assess the combined effect of cavity tip and tip injection;to compare the aero-thermal perfor-mance of the tip leakageflow offlat and cavity tips with injection and to examine the effects of tip clearance height and injection massflow rate.Numerical methods(similar methods have been successfully used for the study on a small scale horizontal axis wind turbine in Ref.[29])have been employed to make a comprehensive understanding of the aero-thermal performance of the tip leakage flow with injection onflat and cavity tips.2.Tip injection configurationsIn this investigation,results are presented for bothflat and cavity tips with and without injection.The design clearance is1.5%Fig.1.Schematic of blade tip geometry and mesh with injection holes.J.Gao et al./Energy69(2014)648e660649of blade span.Fig.1shows the blade tip geometry with injection holes.For the squealer tip,the depth of the cavity is2%of the blade span,and the width of the shoulder is0.77mm,which forms a narrow rim and a wide cavity.For all injection cases,there are four radial injection holes placed on each blade tip.The injection holes are assumed to be1mm in diameter and20mm in length,and are located at10%,30%,50%and70%blade axial chords.The blade geometry chosen for the investigation is taken from GE-E3first stage highly loaded turbine rotor as shown in Fig.2, which represents a modern gas turbine blade geometry.The rotor blades are twisted,and have a constant axial chord length of 28.7mm and an aspect ratio(span to chord)of1.39.The main blade geometry parameters are shown in Table1,and detailed blade geometry and operating conditions can be found in the NASA report(Timko[30]).3.Numerical techniquesThe numerical investigation presented in the paper was per-formed by using ANSYS CFX11.0(AEA[31]),an available general finite volume based Navier e Stokes solver.The solutions are ob-tained by solving the compressible Reynolds-Averaged Navier e Stokes equations using afinite volume method to discretize the equations.The overall accuracy is of the second order.A standard k e u two-equation model developed by Wilcox[32]is used in the investigation for turbulence closure,and no wall functions are used.The commercial structured grid generation package Autogrid5 (preprocessor to NUMECA)was used in the investigation.Fig.2 shows the representative computational grid of the rotor blade used.The channel was discretized into H-type grids,while the re-gions around the blade surface and the tip gap were discretized into O-type grids to ensure high grid quality.Also,details of the cavity tip mesh can be seen more closely as shown in Fig.1.The tip gap and the cavity were modeled with butterfly mesh topology.There are17nodes in the boundary layer to provide grid-independent results,and there are33cell layers distributed from the blade tip to the casing wall with23points across each squealer.Additionally, 27grid nodes per holes diameter had been ensured everywhere inside the injection holes,with33nodes in the hole length direc-tion.There are over100points from the leading edge(LE)toward the trailing edge(TE)along the blade chord.The average of the dimensionless distance,yþ(of the closest grid point from wall)on the wall is less than1,and the maximum yþon the wall is about3. In order to keep the grids for the cases considered in the investi-gation as similar as possible,the same grid topology is adopted for all the cases considered.The computational domain consisted of a single blade channel with periodic boundary conditions imposed along the circumfer-ential direction.At the inlet,which is placed at one axial chord length upstream of the rotor blade,absolute total pressure,absolute total temperature andflow angle are specified along with the tur-bulence intensity.At the exit,which is placed at a distance equal to two axial chord lengths downstream of the rotor blade,static pressure is specified.At the inlet of injection holes,injection mass flow rate and static temperature(350K)are also predefined.The isothermal wall condition(478.1K,a scaled experimental value)is imposed on the blade tip,and the other walls are assumed to be adiabatic.In addition,the no-slip condition is applied to all walls. Detailed boundary conditions are listed in Table1.Convergence criteria for these computations are based on the reduction of RMS residuals below1Â10À5.All results have been checked for convergence by using isentropic efficiency monitoring, and the solution was said to be converged when the isentropic efficiency was not changed more than0.01%during10successive iterations.Typically,500iterations were necessary for convergence.Grid independence is checked by comparing the solutions on a flat tip blade with injection obtained with1.3million nodes and the solutions obtained with1.6million nodes.From the distribution of the heat transfer coefficient on the blade tip,it is found that results from the simulation with 1.6million nodes are in reasonable agreement with that with1.3million nodes.Besides,there are few changes in the tip leakage loss when increasing the grid nodes from 1.3to1.6million.Thus,the total number of grid nodes for theflat tip with injection is1.6million in the investigation.Likewise,the total number of grid points for other cases ranges from1.4million to1.9Fig.2.3D computational grid and details of the LE,TE and tip grids.Table1Blade geometry andflow conditions.Parameters ValuesTip(outer)diameter732mmBlade number76Axial chord28.7mmInlet Mach number(relative)0.37Absolute total pressure ratio 2.25Absolute inlet angle from axial direction(hub)69degAbsolute inlet angle from axial direction(midspan)74.5degAbsolute inlet angle from axial direction(shroud)78.5degRotational speed8450r/minInlet turbulence intensity5%parison of heat transfer coefficient on the blade tip:(a)experimental result,and(b)predicted result.J.Gao et al./Energy69(2014)648e660650million cells,with the more complex tip geometry requiring more cells.In order to validate the ability of the standard k e u turbulence model to predict the blade tip leakage flow and heat transfer,a linear cascade studied by Azad et al.[33]was calculated.The cases is a no-injection flat tip blade with a 1.5%tip clearance and tur-bulence intensity Tu ¼6.1%.Since Azad ’s linear cascade data are often used for validation,only cascade heat transfer dataareFig.4.Streamlines near the blade tip (a)without and (b)with injection at a tip clearance of1.5%.Fig.5.Velocity magnitude contours near the blade tip (a)without and (b)withinjection.Fig.6.Entropy-increase contours near the blade tip (a)without and (b)with injection at 30%,50%,80%and 110%axial chords.J.Gao et al./Energy 69(2014)648e 660651presented here for the sake of brevity.Fig.3compares the measured and predicted heat transfer coefficient distributions on the blade tip without injection,and it should be noted that the uncertainty estimate in the heat transfer coefficient reported by Azad is7.9%.The regions of high heat transfer coefficient are located along the PS where theflow accelerates into the tip-gap region.The location and the size of the predicted“sweet spot”region,as well as the corresponding heat transfer coefficients are in excellent agreement with the measurements.In view of the experimental uncertainties,the measurements and predictions appear to exhibit quite reasonable agreement with each other,with the magnitude of the peak predicted heat transfer coefficient agreeing with the cor-responding measured value to within12%.Although there is no experimental data available to validate the present CFD results for the injection cases,as nowadays accepted in open literature,the use of a proper grid with a reliable turbulence modeling provides a modeling of satisfactory accuracy.4.Results and discussionAs a by-product of the presentflow and heat transfer compu-tations,it is possible to compute the isentropic efficiency and the possible effect of tip coolant injection on turbine efficiency. Despite40years development of gas turbine cooling technology, there is no general agreement on the most appropriate definition of cooled turbine efficiency;the critical issue is the choice of a hypothetical“ideal”process.Based on the review of the definitions in use,Young and Horlock[34]presented new proposals for overcoming the problems to compute the inlet total pressure and inlet total temperature after mixing.Thus,the isentropic efficiency can be calculated in terms of the natural turbine efficiency formula.The mixing process is defined to be adiabatic and the total temperature after mixing T0m can be calculated from the steady-flow energy equation,written for semi-perfect gases as,ð1ÀfÞZ T0mT01g c pgðT0Þd T0þXif iZ T0mT01c ic pcðT0Þd T0¼0(1)The total pressure after mixing p0w is given by,p0w 01g ¼Yip01c i01g!ðfiR C=RÞ(2)where f i¼m c i/(m gþPm c i),f¼P fi and the summation is over then coolant streams.c pg is specific heat related with the total tem-perature T0.R denotes the specific gas constant,and p0is the total pressure.The subscripts c i,g and m denote the i th coolant stream, mainstream gas and mixed conditions.Because the blade tip surface temperature was set at a constant temperature of478.1K,in order to recover adiabatic conditions,the mixed-out total temperature at the exit must be modified to reflect the loss of energy through heat transfer.The change in mass-averaged normalized total temperature at the exit of the blade row can be computed as follows,D T0;ex¼P ZAðm v T=v n j wÞd Am RePr(3)In the above equation m is dimensionless viscosity and n is the normalized distance to the wall;m is the normalized massflow rate through the blade passage and A is the surface area.The integration is performed over all of the heat transfer surfaces.Thus,the mass-averaged value of the exit total temperature is computed as,T0;ex¼T unmodifiedþD T0;ex(4) Also,the effect of non-adiabatic wall conditions must be taken into account in the calculation of entropy-increase D S.4.1.The physics and loss mechanisms of tip leakageflow with tip injectionFlow patterns over theflat tip with and without injection at a tip clearance of1.5%are shown in Fig.4.It shows that the coolant ejected from the injection holes reduces the overall tip leakageflow that enters the tip gap in two ways.Firstly,the coolant is injected normally into the tip gap and has little momentum in the direction of the tip leakageflow.The tip leakageflow that crosses the PS of the tip gap collides with the coolant over the tip gap.As shown in Fig.4(b),several streamlines suddenly change the direction in the tip gap,which is due to the collision with the leakageflow from upstream.This happens in a region where the leakageflow velocity is locally low as shown in Fig.5.Secondly,the coolant mixes with theflow across the tip gap downstream of the injection holes,and it slows the leakageflow.Also,Fig.4shows that several leakage streamlines turn upstream of the injection holes(called the “reverse leakageflow”)with the effect of the coolant.Additionally, it is found that with tip injection,the leakageflow is turned toward the SS edge of the blade,which can also be proven from Fig.5.Fig.5shows velocity magnitude contours on the blade tip sur-face with and without tip injection.In Fig.5(a),theflow accelera-tion in the tip-gap region can clearly be seen(except in the“sweet spot”LE region),and theflow decelerates as the leakageflow rolls up into a vortex along the SS.With tip injection as shown in Fig.5(b),due to the lower injection velocity,the tip-gap region is characterized by the accelerating leakageflow interspersed by lower velocity coolant streaks.The momentum exchange between the coolant and theflow that enters the tip gap is expected to block and deflect the tip leakageflow.Thus,it can be concluded that the coolant jets are turned toward the blade SS and appear to form a film on the tip surface.It should be noted that the injection results in the extra mixing losses,although it blocks the leakageflow.In order to judge the overall effect of tip injection on the tip leakage losses,entropy-increase contours near the blade tip with and without injection at0.20.40.60.81.01.21.4Efficiency-increaseLeakage-decreaseValue(%)Injectionmass flow rateparison of tip leakage-decrease and efficiency-increase between without and with tip injection forflat and cavity tips.J.Gao et al./Energy69(2014)648e660 65230%,50%,80%and 110%axial chords are shown in Fig.6.The injection flow signi ficantly alters the nature of the leakage vortex,and also alters the SS passage flow in the vicinity of the tip.With tip injection,the most noticeable phenomenon is the dramatic reduction of the tip leakage vortex in both its strength and size;the passage vortex is signi ficantly weakened due to the changed endwall flow fields.Overall,the turbine performance has been signi ficantly improved.Denton [35]indicated that,the leakage loss depends both on the proportional leakage flow and on the magnitude of the velocity difference between the leakage flow and the main flow with which it mixes.The coolant is injected normally into the tip gap,and then it mixes with the flow across the tip gap downstream of the in-jection holes.After exiting the tip gap,they mix with the main flow inside the blade passage,forming the tip leakage vortex.It can be concluded that,the tip leakage flow and the coolant have approx-imately the same magnitude of the velocity difference.Therefore,the overall effect of tip injection on the turbine performance de-pends on the relative magnitude of the injection mass flow and the reduction in the tip leakage mass flow.If the tip injection can reduce the flow which enters the tip gap from the PS by more than the injection mass flow rate,the turbine performance would be improved,and vice versa.The above analyses can also be proven from Fig.7.The tip leakage and injection mass flow rates are reported as percentage of total turbine mass flow.The leakage-decrease values were ob-tained by calculating the leakage difference between with and without tip injection;so did the ef ficiency-increase values.It is shown that,with tip injection,the tip leakage flow decreases dramatically,and the reduction in the tip leakage flow is larger than the injection mass flow rate.The results in the reduction in the mass flow that exits out of the tip gap from the SS,which increases the turbine ef ficiency obviously.In addition,the cavity tip with injection performs worse performance in controlling the tip leakage flow than the flat tip with injection.This means that there is a mutual coupling between the tip injection and the blade tip geometry.bined effects of cavity tip and tip injection on aero-thermal performance of tip leakage flowsFig.8shows the comparison of velocity vectors and Mach number distribution on flat and cavity tips with an injection mass flow rate of 0.4%.The cut plane passes through the center of the second hole from the LE,and it is parallel to the direction of thetipparison of velocity vectors and Mach number distribution on flat and cavity tips with injection mass flow rate of 0.4%at tip clearances of 1%and 2%:(a)without injection,1%,(b)with injection,1%,(c)without injection,2%,and (d)with injection,2%.J.Gao et al./Energy 69(2014)648e 660653leakage flow.As shown in Fig.8(a),for a flat tip the flow in the blade tip region is directed primarily from the PS to the SS,with the leakage flow rolling up into a vortex as it encounters the SS passage flow.For a cavity tip,the flow inside the cavity is quite complex andstrongly three-dimensional.The flow that enters the tip gap from the LE is directed toward the PS rim,and bounces off the rim surface toward the SS eventually exiting the gap near or past mid-chord,rolling up into a vortex that createsblockage.parison of heat transfer coef ficient distribution on flat and cavity tips at tip clearances of 1%and 2%:(a)without injection,1%,(b)with injection,1%,(c)without injection,2%,and (d)with injection,2%.x /hEntropy-increase (J/(kg·K))x /hEntropy-increase (J/(kg·K))Fig.10.Mass-averaged entropy-increase distribution along the span at rotor exit for flat and cavity tips:(a)1%and (b)2%tip clearances.J.Gao et al./Energy 69(2014)648e 660654As shown in Fig.8(b),for a flat tip with injection,the coolant is injected normally into the tip gap and counteracts with the tip leakage flow passing the tip region near the PS.This blocks the tip leakage mass flow and leads to low momentum activity between the injection holes and the SS corner.At the same time,coolant jets are turned toward the SS of the tip gap.For a cavity tip with in-jection,the coolant first mixes with the flow in the cavity,then it enters the tip gap,and finally it exits the tip gap along the SS squealer of the cavity tip.Thus,it can be concluded that due to the collision of the coolant and the flow that enters the tip gap from the PS,the momentum exchange is smaller than that in the case of the cooled flat tip.Besides,it can be seen that,over the PS inlet of the tip gap,the colors indicate that when coolant is used,the velocity magnitude is lower than that in the uncooled case.This shows that the leakage flow across the tip PS is partially blocked by the coolant injection.At the tip SS,the velocity magnitude is also low in regions where the coolant exits the tip gap.With the effect of tip clearance heights,it is shown that,for the small tip clearances,the coolant from the injection holes fills the tip gap.For the large tip clearances,however,there is physically enough space for the jets to impinge on the outer shroud and then reattach to the blade tip.This means that,with the same injection mass flow rate,the coolant creates a proportionally larger blockage at the smaller tip clearances.Also,Hohlfeld et al.[21]and Couch et al.[22]found that the blockage effect of the coolant is more evident at the smaller tip clearances.The comparison of the heat transfer coef ficient distribution on flat and cavity tips with an injection mass flow rate of 0.4%is shown in Fig.9.It can be seen that without tip injection,the high heat transfer coef ficient occurs near the LE,where the passage flow is driven into the tip gap by higher pressure near the stagnation point of the blade.For the cavity tip,the separation vortex has the characteristics of separation and reattachment,occupying about half of the cavity region.The point of reattachment corresponds to high heat transfer coef ficients on the cavity floor.When the coolant is injected from the blade tip surface,the heat transfer coef ficient along the coolant trajectory decreases signi ficantly.Particularly for the cavity tip,the heat transfer coef ficient is lower due to coolant accumulation inside the tip cavity.However,as the tip clearance increases,the lateral spreading of the coolant jets becomes small,caused by the increased peak velocities in the leakage flow path.Additionally,for the flat tip,high heat transfer coef ficients still occur near the LE,where no injection holes are located.However,for the cavity tip,the heat transfer coef ficient near the TE of the blade tip is reduced obviously due to coolant accumulation from upstream holes in the tip cavity.This recon firmed that the leakage flow is turned toward the blade SS edge,which reduces the magnitude of the velocity difference between the leakage flow and the main flow,and then the tip leakage losses.It can also be seen that the effect of coolant injection on heat transfer coef ficients is relatively large for the small tip gaps (s /h ¼1%),with a reduced effect on the large tip gaps (s /h ¼2%).This is because,for the small clearance gap case,the amount of leakage flow is small,and the coolant injection has relatively large effect on heat transfer coef ficients.However,for the large clearance gap case,because the amount of leakage flow is large,the effect of coolant injection on heat transfer coef ficients is less important.The dominating secondary flow phenomena in the rotor passage are the tip leakage vortex and the passage vortex,and there ism L (%)τ/h (%)ξ(%)τ/h (%)ξm i x i n g (%)τ/h (%)ηi s (%)τ/h (%)Fig.11.Effects of tip clearance on (a)tip leakage flow rate,(b)losses inside the tip gap,(c)caused by the mixing when the leakage flow leaves the tip gap,and (d)isentropic ef ficiency for flat and cavity tips with and without injection.J.Gao et al./Energy 69(2014)648e 660655。

tpo68三篇托福阅读TOEFL原文译文题目答案译文背景知识阅读-1 (2)原文 (2)译文 (5)题目 (8)答案 (15)背景知识 (17)阅读-2 (20)原文 (20)译文 (24)题目 (27)答案 (35)背景知识 (37)阅读-3 (41)原文 (41)译文 (44)题目 (47)答案 (54)背景知识 (55)阅读-1原文Salt and the Rise of Venice①The city of Venice, on Italy’s coastline, achieved commercial dominance of southern Europe during the Middle Ages largely because of its extensive trade in the valuable commodity of salt. At first, Venice produced its own salt at its Chioggia saltworks. For a time its principal competitor in the region was the town of Cervia, with Venice having the advantage because Chioggia was more productive. But Chioggia produced a fine-grained salt, so when Venetians wanted coarser salt, they had to import it. Then, in the thirteenth century, after a series of floods and storms destroyed about a third of the salt-producing ponds in Chioggia, the Venetians were forced to import even more salt.②That was when the Venetians made an important discovery. More money could be made buying and selling salt than producing it. Beginning in 1281, the government paid merchants a subsidy on salt landed in Venice from other areas. As a result of this assistance, shipping salt to Venice became so profitable that the salt merchants could afford to ship other goods at prices that undersold theircompetitors. Growing fat on the salt subsidy, Venice merchants could afford to send ships to the eastern Mediterranean, where they picked up valuable cargoes of Indian spices and sold them in western Europe at low prices that their non-Venetian competitors could not afford to offer. That meant that Venetians were paying extremely high prices for salt, but they did not mind expensive salt if they could dominate the spice trade and be leaders in the grain trade. When grain harvests failed in Italy, Venice would use its salt income to subsidize grain imports from other parts of the Mediterranean and thereby corner the Italian grain market.③Unlike the Chinese salt monopoly, the Venetian government never owned salt but simply took a profit from regulating its trade. Enriched by its share of sales on high- priced salt, the salt administration could offer loans to finance other trade. Between the fourteenth and sixteenth centuries, a period when Venice was a leading port for grains and spices, 30 to 50 percent of the tonnage of imports to Venice was in salt. All salt had to go through government agencies. The salt administration issued licenses that told merchants not only how much salt they could export but also to where and at what price. The salt administration also maintained Venice’s palatial public buildings andthe complex hydraulic system that prevented the metropolis from washing away. Many of Venice’s grand statues and ornamental buildings were financed by the salt administration.④Venice carefully built its reputation as a reliable supplier, and so contracts with the merchant state were desirable. Venice was able to dictate terms for these contracts. In 1250, when Venice agreed to supply Mantua and Ferrara with salt, the contract stipulated that these cities would not buy salt from anyone else. This became the model for Venetian salt contracts. As Venice became the salt supplier to more and more countries, it needed more and more salt producers from which to buy. Merchants financed by the salt administration went farther into the Mediterranean, buying salt from many distant sources. Wherever they went, they tried to dominate the supply, control the saltworks, and even acquire them if they could.⑤Venice manipulated markets by controlling production. In the late thirteenth century, wishing to raise the world market price, Venice had all saltworks on the Greek island of Crete destroyed, and it banned the local production of salt. The Venetians then brought in all the saltneeded for local consumption, built stores to sell the imported salt, and paid damages to the owners of the saltworks. The policy was designed to control prices and at the same time keep the locals happy. Aiding its ability to ruthlessly manipulate commerce and control territory, Venice maintained the ships of the merchant fleet as a naval reserve and called them into combat when needed. The Venetian fleet patrolled the Adriatic Sea, stopped ships, inspected cargo, and demanded licensing documents to make sure all commercial traffic was conforming with its regulations.译文盐和威尼斯的崛起①位于意大利海岸线上的威尼斯城在中世纪期间在南欧取得了商业主导地位，主要是因为它广泛从事有价值的盐贸易。