This paper presents computational work on the biogas early-phase combustion in

Transcription

1 Full chemcal knetc smulaton of bogas early-phase combuston n SI engnes Abstract Madan Mohan A. 1*, Anand T.N.C 2, and Ravkrshna R. V 1. 1 Combuston & Spray Laboratory, Department of Mechancal Engneerng Indan Insttute of Scence, Bangalore , INDIA. 2 Department of Mechancal Engneerng, IIT Madras, Chenna , INDIA Ths paper presents computatonal work on the bogas early-phase combuston n SI engnes usng detaled chemcal knetcs. Specfcally, the early-phase combuston s studed to assess the effect of varous gnton parameters such as spark plug locaton, spark energy, and number of spark plugs. An ntegrated verson of the KIVA-3V and CHEMKIN codes was developed and used for the smulatons utlzng detaled knetcs nvolvng 325 reactons and 53 speces. The results show that locaton of the spark plug and local flow feld play an mportant role. A central plug confguraton whch s assocated wth hgher local flow veloctes n the vcnty of the spark plug, showed faster ntal combuston. Although a dual plug confguraton shows the hghest rate of fuel consumpton, t s comparable to the rate exhbted by the central plug case. The radcal speces mportant n the ntaton of combuston are dentfed and ther concentratons are montored durng the early phase of combuston. The concentraton of these radcals s also observed to correlate very well wth the above-mentoned trend. Thus, the role of these radcals n promotng faster combuston has been clearly establshed. It s also observed that the mnmum gnton energy requred to ntate a self-sustaned flame depends on the flow feld condton n the vcnty of the spark plug. Increasng the methane content n the bogas has shown mproved combuston. 1

2 *Correspondng author: 1 Introducton Awareness of lmtatons of conventonal fossl fuel reserves has ntensfed the search for renewable fuels for use n nternal combuston engnes. Bogas s one possble alternatve for power generaton due to flexblty n producton from varous feedstocks. Snce bogas can be produced from agrcultural and forest waste, t does not conflct wth food producton nterests 1. The bogas generated from varous feedstocks contans manly methane and carbon doxde. Dependng on the feed stock and the generaton process used, bogas contans % methane, % carbon doxde, and trace amounts of Hydrogen and H2S. The presence of carbon doxde and other dluents such as N2 and H2S wth methane reduces the heatng value of bogas fuel. The presence of carbon doxde also leads to reduced effectve concentratons of the fuel, CH4, and of O2. Carbon doxde also tends to undergo endothermc reactons that ncrease rapdly n ntensty as combuston temperatures are approached. These endothermc reactons are sgnfcant n the early stages of combuston and play a major role on flame propagaton. To determne the optmal gnton strategy for bogas-fuelled engnes, t s requred to understand the n-cylnder aspects of bogas early-phase combuston. Due to the presence of CO2 whch nhbts combuston, gnton delay s expected to be hgher, and the burnng velocty wll be lower. Advancng the gnton tmng wll resolve the problem of longer delay straghtway but may not always be a convenent and vable strategy 2. Addng fuel addtves and ncreasng the gnton source energy are two other possble methods of acceleratng the endothermc phase n the ntal phase of combuston 2. Although, the major part of fuel combuston and heat release take place under hghly turbulent 2

3 condtons, the ntal phase of combuston s very crtcal for ensurng a self sustaned flame and for reducng the tme between the ntaton of spark and onset of robust, self sustaned combuston. Durng ths perod, a pool of radcal and combuston ntermedates must be establshed to feed the man heat release reactons. The progress of the early combuston phase s closely lnked to the detaled chemstry as well as local heat and mass transfer condtons,.e., the local flow feld. Thermodynamc and chemcal propertes of the non-methane component of bogas, whch s prmarly CO2, can also strongly affect the early stages of the combuston due to ts non-combustblty. In lterature, most of the studes on bogas-fuelled engnes are expermental n nature. Karm and Werzba 3 conducted experments usng a methane and carbon doxde mxture as fuel for a sngle-cylnder, spark gnton engne. It was observed that the maxmum power output depends on spark tmng and the concentraton of carbon doxde. Huang and Crookes 4 found that wth bogas fuels contanng a sgnfcant amount of CO2, the engne runs wth close-to-stochometrc mxtures and hgher compresson rato gves good performance. The effect of spark plug locaton and number of plugs n natural gas engnes was dscussed by Mayer et al. 5. It was found that a multple spark plug confguraton s better than a centrally located spark plug for maxmum power, hgher thermal effcency and low HC producton at leaner equvalence ratos 5. Stone et al. 6 found that the presence of carbon doxde n bogas lowers the burnt gas temperature, whch translates nto lower NO emssons. Bertol et al. 7 lnked KIVA- 3V 8 and CHEMKIN II 9 to smulate desel engne combuston wth 57 speces and 290 chemcal reactons. To handle stff dfferental equatons, a specal solver lbrary called VODE was used. Each cell was consdered as a contnuously strred tank (CST) adabatc 3

4 reactor whch s physcally and formally ndependent from the surroundng cells. Smlarly, Yoseff et al. 2 lnked KIVA-3V and CHEMKIN II to smulate a natural gas engne, and found that early stages of combuston are more senstve to the characterstcs of the gnton source than to large changes n fuel composton. Fundamental work on bogas chemcal knetcs was performed n counter flow dffuson flames to study the effect of flame stretch on chemstry 10. Thus, t s observed that most studes on bogasfuelled engnes n lterature have been expermental n nature, wth few computatonal studes. Also, there s a specfc need to study bogas early-phase combuston n engnes snce gnton strateges are strongly affected by ths phase of combuston. Snce bogas early-phase combuston nvolves complex chemcal knetcs, smulatons need to be performed usng full chemstry. Also, t s not suffcent f the chemstry alone s smulated accurately; the flow feld n the engne also needs to be smulated accurately. To the best of our knowledge, the current work s the frst reported study where the full chemcal knetcs has been taken nto account to study the bogas early-phase combuston n SI engnes, and addtonally, realstc n-cylnder flow feld condtons have been utlzed n the smulatons. The man motvaton behnd the present work s to explore gnton strateges for effectve combuston of bogas n SI engne. Current expermental efforts requre a fundamental understandng of gnton and combuston phenomenon of bogas. Ths would help n explorng gnton strateges such as spark tmng, spark energy, spark plug locaton and dual spark plugs to maxmse engne performance. The present study nvolves study of the early-phase of combuston n a sngle cylnder 110-cm 3 four-stroke engne usng bogas as fuel. One of the objectves s to explore change n spark tmng on 4

5 combuston characterstcs and engne performance wth bogas fuel. In order to smulate the ntal phases of engne combuston, KIVA-3V, a specalsed mult-dmensonal engne smulaton program that can smulate turbulence, combuston and movng pston boundary, and CHEMKIN-II, a program for solvng speces concentraton from a multstep chemcal knetc mechansm are used. The next few sectons descrbe the computatonal methodology, and a dscusson of the results followed by a summary of the key fndngs. 2 Computatonal Methodology To smulate detaled chemcal knetcs along wth flud dynamcs, the CHEMKIN II code s ntegrated nto the KIVA-3V code. In ths way, the lnked code offers the well known strengths of KIVA-3V for nternal combuston engne calculatons, whle the treatment of detaled chemstry s addressed by CHEMKIN II. The RNG k-ε model s used to smulate the effect of turbulence on convecton and dffuson transport. The ntegraton s done such a way that KIVA-3V passes thermodynamc data and speces data to CHEMKIN II to obtan a full chemstry soluton. Each grd cell s treated as a constant volume adabatc reactor for the chemstry soluton. Ths s justfed because the cell volume does not change n a computatonal tme step of KIVA-3V, and flud and thermodynamc nteractons are frozen across the boundary durng the chemstry calculaton. Durng the chemstry soluton, the KIVA-3V tme step s taken as the ntegraton tme n the CHEMKIN II solver to get updated mxture condtons. The energy released n each cell due to the chemcal reactons s calculated and added to the specfc nternal energy of that partcular cell. The new gas phase temperature s then updated at the end of tme step n CFD solver whch s consstent wth KIVA-3V 5

6 formulaton. A smlar approach was used by Yossef et al. 3,11,12 to smulate natural gas combuston n SI engnes and by Kong 13 to smulate DME mxture combuston n a HCCI engne. The governng equatons for the constant volume adabatc reactor 14 are shown Appendx I. A flowchart outlnng the computatonal steps, and the KIVA-CHEMKIN couplng, s shown n Fgure 1. Snce the bogas contans methane as the man reactve consttuent, a detaled chemcal knetc mechansm for CH4, GRI , was used to smulate the fuel combuston chemstry. The mechansm conssts of 325 reactons and 53 speces. In a typcal combuston process, ntaton happens when a free radcal attacks the fuel molecule. The concentraton of such free radcals thus plays an mportant role n combuston ntaton and propagaton. From the GRI 3.0 mechansm, the followng reactons are dentfed to be mportant n ths ntaton process: O+CH4<=>OH+CH3 H+CH4<=>CH3+H2 OH+CH4<=>CH3+H2O CH+CH4<=>H+C2H4 CH2+CH4<=>2CH3 HO2+CH4<=> CH3+H2O2 HCO+CH4<=> CH3+CH2O (R1) (R2) (R3) (R4) (R5) (R6) (R7) Thus, the concentratons of radcals HCO, CH, HO2, O, H and CH2 apart from OH whch s an mportant radcal ndcatng the presence of the flame zone, are tracked durng the 6

7 smulatons. The hgh temperature quanttatve reacton path dagram (QRPD) for methane-ar combuston also ndcates that OH, H, and O are mportant radcals that drectly attack the methane molecule 14. The net reacton rates of the other four radcals consdered n the present study are much smaller than those of the above-mentoned radcals. However, they were also ncluded n the dscusson snce the detaled GRI 3.0 mechansm ndcated that these radcals are also nvolved n a drect reacton wth the methane molecule. 3 Results and Dscusson The present study was carred out for a 110 cm 3, sngle-cylnder, SI engne geometry. Fgure 2 shows the geometry of the engne used n the smulatons whch corresponds to the closed-valve part of the engne cycle. The focus of ths work s on the early-phase combuston, and hence, computatons start from the nlet valve closure (IVC). The tme from the ntaton of spark to the tme where 10% of the total fuel burnt s consdered as the ntal phase 16. In ths phase, combuston s predomnantly governed by chemcal knetcs snce the turbulent tme scales are lower when compared to chemcal tme scales. The calculated turbulent tme scales are of the order of 10-4 s, whle the chemcal tme scales are of the order of 10-3 s. The turbulent tme scale s calculated by takng the rato of k/ε whle the chemcal tme scale s the tme for the fuel to reach 1/e of ts ntal value 14,17. Flow at the gnton locaton s crtcal n sustanng the combuston and flame development. The flow feld from a CFD smulaton of the ntake stroke of the engne usng the AVL-FIRE software and the real engne geometry was taken at ntake valve closure and mapped onto the KIVA-3V grd to smulate engne-lke flow condtons. A flow feld mappng algorthm developed by Sharma et.al 18 s used n 7

8 the present study. The mappng algorthm takes the CFD soluton from AVL FIRE and the coordnates of the KIVA-3V mesh as nput. Then, the algorthm determnes the extent of each KIVA cell and dentfes the FIRE mesh nodes lyng wthn each KIVA cell regon. Each of such FIRE nodes represents a FIRE mesh cell. The values for each thermodynamc and flud varable are averaged over the correspondng values at all such FIRE nodes and then assgned to the KIVA node under consderaton. A homogeneous mxture composton was assumed at the ntake valve closure and the ar-fuel rato was calculated usng expermental flow rates of fuel and ar. A constant wall temperature of 450 K was used for cylnder head, wall and pston surface. Specfcatons of ths engne are descrbed n Table 1. All computatons were performed usng a full, 360 o, threedmensonal mesh. As mentoned earler, smulatons were performed wth a detaled knetc mechansm. The mportant assumpton made n the smulatons s that the turbulent mxng tme scales are small compared to the chemcal tme scales, mplyng that the effect of turbulence on the combuston s small. Ths s a reasonable assumpton snce the ntaton of combuston s known to be chemstry-drven 19. Varous strateges such as varaton of spark plug locaton spark tmng, spark energy and dual spark plugs have been studed. Effect of Spark locaton The locaton of spark plug plays an mportant role n the flame propagaton due to the spatal varaton n the local flow feld. The local flow feld affects the mxng of the hot combuston products and the fresh un-burnt mxture. To study the effect of spark locaton, varous strateges ncorporatng sde, central and dual spark plugs have been 8

9 smulated. The spark locatons are shown n Fg. 3 along wth the velocty magntude n that plane. The varatons n the flow feld n a perpendcular plane can also be observed n Fg. 4. The same flow feld s used as the ntal condton for all the cases smulated. However, there s a dfference n the local flow feld n the vcnty of each plug whch s shown quanttatvely n Fg.3. The magntude of the velocty s hgher at the central spark plug locaton compared to sde plug locatons. The asymmetry n the flow feld s observed due to the swrl and tumble moton nsde the cylnder. Computatons are performed for an equvalence rato of 1.1 and spark energy of 150 mj for each plug. For the dual plug case, the total spark energy s 300 mj. The results show a hgher rate of combuston for the central spark plug locaton as compared to the sde spark plug locaton. The dual spark plug confguraton showed a hgher rate of combuston than the other two strateges due to the larger gnton area and energy. The burnt fuel mass fractons for all three cases are shown n Fg. 5. The fracton calculated by takng the rato of the mass of fuel (CH4) burnt from spark gnton to that crank angle, to the ntal mass of the fuel, s referred to as the nstantaneous burnt fracton. The amount of fuel burnt for the dual spark plug case s almost twce as that for the sde spark plug case, but only slghtly hgher compared to that of the central spark plug confguraton. Although spark energy and the area are same for both sde and central plugs, the local flow feld s markedly dfferent wth hgher veloctes n the case of the central plug, leadng to a hgher burnng rate. Ths s a very mportant observaton and underscores the mportance of the couplng between flow and chemstry even durng the ntal reacton phase where chemstry and not turbulence plays a major role. The total concentraton of some of the radcals, specfcally OH, O and H are shown as a functon 9

10 of crank angle n Fg. 6. It s nterestng to note that all the radcals show near-dentcal trends. The other radcals consdered are comparatvely much smaller n concentraton, and hence ther varaton s not shown n ths fgure. Also, t s observed that ntally, for the central plug case, the rate of growth of these radcals s slower compared to that for both the sde and the dual plug cases. However, the radcal concentraton for the central plug becomes faster and hgher as compared to that for the sde plug confguraton as tme progresses. The ntal slow rate for the central plug case s attrbuted to the local flow feld wth hgher veloctes at the spark locaton whch results n a faster dsspaton of the spark energy. Ths makes the gnton progress slower ntally, and once a suffcent spark kernel s formed, ths strong flow feld helps n faster spread of hgh temperature combuston products and consequent faster combuston. To demonstrate ths, the fuel mass fracton contours n the plane of spark are shown n Fg. 7 for varous crank angles from the onset of the spark. The flame s observed to propagate more n one drecton as compared to the other due to flow feld drectonalty. Ths s observed for all the three confguratons. Ths effect can be further seen n Fg. 8 (a) and Fg. 8 (b) whch show fuel mass fracton and temperature contours, respectvely n two perpendcular planes for the central spark plug confguraton. Effect of spark energy The ntal phase of bogas combuston s domnated by endothermc reactons due to the presence of CO2. Ths may cause msfre, f suffcent energy s not suppled durng the gnton process. The effect of gnton energy s studed for sde, central and dual plug confguratons for 30 BTDC spark tmng. For each case, the effect of three 10

11 dfferent gnton energes (100 mj, 150 mj and 200 mj) has been studed. The total spark energy s suppled to the cells, n an area of 1-mm 3 over a sx-degree crank angle perod from the onset of spark. For the sde and dual plug cases, the smulatons show that an ncrease n the gnton energy ncreases the rate of combuston as evdent from the fuel consumpton rate shown n Fg. 9 as well as n-cylnder OH radcal concentratons shown n Fg. 10. However, for the sde plug confguraton, the dfference n the cases wth 150 mj and 200 mj s small. Interestngly, for the central plug case wth 100 mj gnton energy, combuston was not sustaned. As the gnton energy s ncreased to 150 mj, combuston s observed to sustan, and further ncrease n the gnton energy to 200 mj results n a faster burnng rate and s hgher than that for both the sde and dual plug confguratons wth the same gnton energy. Ths s attrbuted to varaton n the local flow feld n the vcnty of the spark. As mentoned earler, the flow velocty s hgher at the central plug locaton compared to the other spark locatons. Ths hgh velocty ads n carryng the heat and hgh temperature products from the spark locaton. For the lower spark energy correspondng to 100 mj, the transport of heat and products away from the spark locaton domnates over local heat release resultng n flame extncton. Effect of equvalence rato In practcal stuatons, the ar-fuel rato of the engne vares wdely dependng on the requrement. The composton of the fuel and ar mxture has a major effect on the combuston velocty and flame development. The varaton n composton along wth the local flow feld alters the ntal flame development stage drastcally. In the present study, the effect of equvalence rato on the ntal stages of combuston at three spark locatons has been studed. Smulatons were carred out for equvalence ratos of 0.85 (fuel-lean), 11

12 1 (stochometrc), 1.05 and 1.1 (fuel-rch). All three confguratons vz. sde, central and dual plugs were consdered for each equvalence rato. In all the cases, the total spark energy suppled was mantaned constant at 150 mj per plug. For the condtons studed, smulaton results show that the stochometrc mxture burns slghtly faster as compared to both lean and rch mxtures. Ths s seen from the amount of fuel mass burnt shown n Fgure 11. Ths trend s n agreement wth the expermental fndngs of Shy et al. 20 for turbulent methane-ar flames. At stochometrc condtons, ntally, the central spark plug showed a smlar rate of fuel consumpton as the dual plug case. However, as tme progresses, the dual spark plug confguraton shows a hgher burnng rate than the central plug confguraton. The reason for ths can be understood by consderng the flow feld at the tme of the spark ntaton shown n Fg. 3. It can be observed that the flow feld drecton at the second spark plug n case of dual spark plug confguraton s towards the cylnder head. Ths makes one of the spark plugs less effectve n propagatng the hot products and radcals outward. At all the remanng equvalence ratos, the dual plug confguraton shows slghtly hgher burnng rates compared to the central plug confguraton. Ths s due to the larger gnton area and presence of two flame fronts. The sum of all the fuel attackng radcal concentratons mentoned earler s shown n Fg. 12. It s nterestng to note that the radcal concentratons agan show the same trend, and are observed to correlate wth the overall fuel consumpton rate. Overall, the central spark plug performs almost as effectvely as the dual spark plug case n consumng fuel n the bogas early-phase combuston across the varous condtons studed n the present nvestgaton. Ths s an mportant and useful fndng especally for small engne 12

13 applcatons where the cylnder head has very lmted space to accommodate an addtonal spark plug. Effect of bogas composton The effect of bogas composton s studed usng varous mxtures of CH4 and CO2. Three dfferent compostons of bogas vz. () 65% CH % CO2, () 75% CH % CO2, and () 85% CH % CO2 are consdered for the current study. Smulatons are performed for the central spark plug case wth 150 mj spark energy. A comparson of burnt fuel fractons for the dfferent fuel compostons at varous equvalence ratos are shown n Fg.13 to Fg.15. Smulatons showed that ncreasng the methane percentage n the bogas mproves the combuston for all the equvalence ratos studed. Increasng the methane content reduces the nert carbon doxde n the fuel and hence mproves the gnton and combuston characterstcs of the fuel. 4 Summary Detaled three-dmensonal smulatons of bogas early-phase combuston wth full chemstry utlzng 325 reactons and 53 speces are reported. Specfcally, the study s conducted for a sngle-cylnder, 110 cm 3, four-stroke engne. Flow feld mappng at the nlet valve closure was performed to smulate realstc ntal condtons nsde the cylnder. The effect of spark plug locaton, spark energy, dual plugs, and equvalence rato on bogas early-phase combuston has been studed. The smulatons show that the flow feld near the spark locaton has a sgnfcant effect on the ntal flame and radcal growth and consequently, the progress of combuston. The central spark plug confguraton showed an mproved ntal flame propagaton compared to the sde spark 13

14 plug case due to hgher flow veloctes n the vcnty of the spark locaton. The dual spark plug confguraton expectedly gave the hghest rate of fuel consumpton, however, the rate exhbted by the central plug s comparable underscorng the mportance of spark locaton. From the detaled mechansm, radcals such as OH, HCO, CH, HO2, O, H and CH2 are montored durng the early-phase combuston. The concentraton of these radcals s also observed to correlate very well wth the above-mentoned trends clearly establshng the role of these radcals n promotng faster combuston. For the central spark plug case, lower gnton energy resulted n flame extncton due to faster transport of combuston products. Thus, t s concluded that the mnmum energy requred to obtan self-sustaned gnton depends strongly on the local flow feld condtons. The results also show that bogas early-phase combuston rates are hgher for stochometrc mxtures as compared to both lean and rch mxtures. Increasng the methane content n the bogas has shown mproved combuston. The results of the study help n understandng the gnton process n bogas-fuelled engnes, and can ad n desgnng engnes wth better performance and lower probablty of msfre. Nomenclature: ATDC: After Top Dead Center BTDC: Before Top Dead Center CFD: Computatonal Flud Dynamcs CST: Contnuously Strred Reactor Cv: Constant volume specfc heat Cv, : Constant volume specfc heat of th speces 14

15 DME: Dmethyl Ether HCCI: Homogeneously Charged Compresson Ignton IVC: Inlet Valve Closure k, : Rate coeffcent of forward reacton f k, : Rate coeffcent of reverse reacton r j m: Mass of the system N: Number of moles N: Number of moles of th Speces n: Stochometrc coeffcent QRPD: Quanttatve Reacton Path Dagram Q : Rate of heat transfer SI : Spark Ignton T: Temperature t: Tme TDC : Top Dead Center u: Specfc nternal energy u : Molar specfc nternal energy u : Molar specfc nternal energy of th speces 15

16 V: Volume of the system W : Rate of work transfer : Net producton rate of the speces X : Speces concentraton References [1] Petrou, E.C.; Papps, C.P. Bofuels: A survey on pros and cons, Energy & Fuels, 2009, 23, [2] Yossef, D.; Maskell, S. J.; Ashcroft, S. J.; Belmont, M. R. A comparson of the relatve effects of fuel composton and the gnton energy on the early stages of combuston n a natural gas spark-gnton engne usng smulaton, Proc. Instn. Mech. Engrs, part D, 2000, 214, [3] Karm, G.A.; Werzba, I. Methane-Carbon Doxde mxtures as a fuel, SAE specal publcatons, 1992, [4] Haung, J.; Crooks R.J; Spark gnton engne performance wth smulated bogas-a comparson wth gasolne and natural gas, Journal of nsttute of energy, 1998, 71, [5] Mayers, D. P.; Kng, S. R.; Mayer, R. C.; Lss, W. E. Effects of spark plug number and locaton n natural gas engnes, ASME Journal of Engneerng for Gas Turbnes and Power, 1992, 114, [6] Stone, G. J.; and Ladommatos N. Analyss of bogas combuston n spark-gnton engnes, by means of expermental data and computer smulaton, Journal of nsttute of energy, 1993, 66, [7] Bertol, C.; DeMarno; P., Belardn, P.; Avolo, G. A New Parallel Numercally Improved KIVA 3V Program for Desel Combuston Computatons. Internatonal mult-dmensonal engne modellng group meetng, March 2003 [8] O Rourke, P. J.; Butler, T. D.; and Amsden, A.A. KIVA 3V-II: A computer program for chemcally reactve flows wth sprays, LA MS; Los Alamos Natonal Laboratory, Los Alamos, NM,

17 [9] Rupley, F. M.; Mller, J. A.; Kee, R. J. Chemkn-II: A FORTRAN chemcal knetcs package for the analyss of gas-phase chemcal knetcs, SAND B; Sanda Natonal Laboratory, Lvermore, CA,1991. [10] Jahangran, S.; Engeda, A.; Wchman, I.S. Thermal and chemcal structure of bogas counter flow dffuson flames, Energy & Fuels, 2009, 23, [11] Yossef, D.; Ashcroft, S. J.; Belmont, M. R.; Abraham M.; Thurley R.W.F.; Maskell, S. J. Early stages of combuston n nternal combuston engnes usng lnked CFD and chemcal knetcs computatons and ts applcaton to natural gas burnng engnes, Combuston Scence and Technology, 1997, 130, [12] Yossef, D.; Maskell, S. J.; Ashcroft, S. J.; Belmont, M. R. Ignton source characterstcs for natural-gas-burnng vehcle engnes, Proc. Instn. Mech. Engrs, part D, 2000, 214, [13] Kong, S.,C. A study of natural gas/dme combuston n HCCI engnes usng CFD wth detaled chemcal knetcs, Fuel, Vol-86, 2007, [14] Turns, S.R., An Introducton to Combuston: Concepts and Applcatons, 2 nd ed., McGraw-Hll: New York, [15] Smth, G.P.; Davd, M.G; Mchael, F.; Ngel, W.M.; Bors, E.; Mkhal, G.; Bowman, C.T.; Hanson, R.K.; Song, S.; Gardner, W.C.Jr.; Lssansk, V.V.; and Zhwe, Q.; GRI-Mech 3.0, Date of access: 08/08/2009. [16] Heywood, J. B., Internal Combuston Engne Fundamentals, 1 st ed., McGraw-Hll: New York, [17] Law, C. K. Combuston Physcs, Cambrdge Unversty Press, [18] Sharma, C.S.; Anand, T.N.C.; Ravkrshna, R.V.; A methodology for analyss of desel engne n-cylnder flow and combuston, Progress n Computatonal Flud Dynamcs, 2010, 10-.3, [19] Ganesan V, Internal combuston Engnes, 3rd ed, McGraw-Hll: New York, [20] Shy, S.S., Ln, W.J.; We, J.C. An expermental correlaton of turbulent burnng veloctes for premxed turbulent methane-ar combuston, Proc. R. Soc. Lond. A, 2000, 456,

18 Appendx I: The dervaton of the governng equatons for the constant volume adabatc reactor s gven n ths secton. Startng wth the rate form of conservaton of energy equaton for a fxed mass system, du Q W m dt (1) where Q s rate of heat transfer to the system, W s the rate of work done, m s the mass of the system and u s the specfc nternal energy. For a constant volume reactor, W 0. The equaton then becomes: du m dt Q (2) The system nternal energy n terms of system chemcal composton can be expressed as 18

19 N U N u u 1 (3) m m where N and u are the number of moles and molar specfc nternal energy of th speces respectvely. Dfferentaton of ths equaton yelds: du 1 dt m dn u dt N du dt (4) Assumng deal gas behavour,.e., T u u, du dt u dt dt cv, (5) T dt dt where, c v, s the molar constant volume specfc heat of speces. The system composton N can be wrtten as: N V X (6) Therefore, dn dt V (7) where s the net producton rate of th speces, calculated from the detaled chemcal mechansm as follows: L q for 1,2,3,... N (8) j1 j j j '' j ' j (9) 19

20 ' '' where L s number of equatons and j and j are stochometrc coeffcents on the reactants and products sde, respectvely, for the th speces and th j reacton. The rate-ofprogress varable, q j, s calculated as: q j k f N 1 j N j X krj X 1 (10) where k f and krj are rate coeffcents of forward and reverse reactons, respectvely. Substtutng Eqns. 5, 6 and 7 nto Eqn. 4 after rearrangement, we get: dt dt Q u V X c v, (11) However, for a constant volume adabatc reactor, Q 0. Therefore, Eqn. 11 becomes: dt dt u X c v, (12) Ths gves the rate of change of temperature n the constant volume adabatc problem. Bore Stroke length Sqush length Swept volume 5.3 cm 5.31 cm 0.08 cm 110 cc Compresson rato

21 Connectng length rod 9.31 cm Engne speed Inlet valve closure Spark Tmng 3175 rpm 25 ABDC 30 ⁰ BTDC Table 1- Engne specfcatons 21

22 Fgure 1: Block dagram of KIVA-3V and CHEMKIN II ntegraton algorthm 22

23 Fgure 2: Computatonal doman correspondng to the n-cylnder geometry at nlet valve closure. 23

24 Fgure 3: Velocty vectors (cm/s) n a plane passng through the spark plugs at the tme of spark ntaton. 24

25 Fgure 4: Velocty magntude (cm/s) n two perpendcular planes at the tme of spark ntaton Fgure 5: Varaton of n-cylnder fuel mass fracton wth crank angle 25

26 Fgure 6: Varaton of n-cylnder radcal concentraton wth crank angle 26

27 Crank angle from the onset of spark Sde Plug Central Plug Dual Plugs

28 Fgure 7: Fuel mass fracton contours n the plane of the spark locaton at varous crank angles from the onset of spark Fuel mass fracton 27 o BTDC 22 o BTDC 17 o BTDC 12 o BTDC 7 o BTDC (a) Temperature n K 27 o BTDC 22 o BTDC 17 o BTDC 28

29 12 o BTDC 7 o BTDC (b) Fgure 8: (a) Fuel mass fracton, (b) Temperature contours n two perpendcular planes at varous crank angles for the central spark plug confguraton. 29

30 Fgure 9: Varaton of n-cylnder burnt fuel mass fracton at dfferent spark energes and spark locaton. 30

31 Fgure 10: Varaton of n-cylnder OH radcal concentraton correspondng to varous levels of spark energy for all three confguratons. 31

32 Fgure 11: Varaton of burnt fuel mass fracton wth crank angle for varous equvalence ratos. 32

33 Fgure 12: In-cylnder fuel attackng radcal concentraton for varous equvalence ratos. 33

34 Fgure 13: Varaton of n-cylnder burnt fuel mass fracton at equvalence rato0.85 Fgure 14: Varaton of n-cylnder burnt fuel mass fracton at equvalence rato

35 Fgure 15: Varaton of n-cylnder burnt fuel mass fracton at equvalence rato