Int. J. Envron. Sc. Technol. (2014) 11:949 958 DOI 10.1007/s13762-013-0273-5 ORIGINAL PAPER Optmzaton of operatonal parameters on performance and emssons of a desel engne usng bodesel K. Svaramakrshnan P. Ravkumar Receved: 3 May 2012 / Revsed: 24 December 2012 / Accepted: 16 March 2013 / Publshed onlne: 18 Aprl 2013 Ó Islamc Azad Unversty (IAU) 2013 Abstract Ths work nvestgates the nfluence of compresson rato on the performance and emssons of a desel engne usng bodesel (10, 20, 30, and 50 %) blendeddesel fuel. Test was carred out usng four dfferent compresson ratos (17.5, 17.7, 17.9 and 18.1). The experments were desgned usng a statstcal tool known as desgn of experments based on response surface methodology. The resultant models of the response surface methodology were helpful to predct the response parameters such as brake specfc fuel consumpton, brake thermal effcency, carbon monoxde, hydrocarbon and ntrogen oxdes. The results showed that best results for brake thermal effcency and brake specfc fuel consumpton were observed at ncreased compresson rato. For all test fuels, an ncrease n compresson rato leads to decrease n the carbon monoxde and hydrocarbon emssons whle ntrogen oxde emssons ncrease. Optmzaton of parameters was performed usng the desrablty approach of the response surface methodology for better performance and lower emsson. A compresson rato 17.9, 10 % of fuel blend and 3.81 kw of power could be consdered as the optmum parameters for the test engne. Keywords Bofuel Compresson rato Energy Karanja Response K. Svaramakrshnan (&) Department of Mechancal Engneerng, Anjala Ammal Mahalngam Engneerng College, Kovlvenn, Inda e-mal: svaporkod2000@yahoo.co.n K. Svaramakrshnan Anna Unversty, Chenna, Inda P. Ravkumar St. Joseph College of Engneerng and Technology, Thanjavur, Tamlnadu, Inda Introducton The combuston of the fossl fuel produced from desel engnes has polluted the envronment through the exhaust emssons of hydrocarbons (HCs), oxdes of ntrogen (NO x ), carbon monoxde (CO), and oxdes of sulfur (SO x ). Moreover, NO x and CO 2 are the green house gases, and SO x causes acd ran. On the other hand, vegetable ols present a very promsng alternatve to desel ol snce they are renewable and have smlar propertes. Many researchers have studed the use of vegetable ols n desel engnes. Vegetable ols offer almost the same power output, wth slghtly lower thermal effcency when used n desel engnes. Reducton of engne emssons s a major research aspect n engne development wth the ncreasng concern on envronmental protecton and the strngent exhaust gas recrculaton. The tremendous growth of vehcular populaton of the world has led to a steep rse n the demand for petroleum products. Bodesel such as Jatropha, Karanja, sunflower, and rapeseed are some of the popular bodesel currently consdered as substtutes for desel. These are clean burnng, renewable, non-toxc, bodegradable, and envronmentally frendly transportaton fuels that can be used n neat form or n blends wth petroleum derved from desel engnes. Methyl and ethyl esters of Karanja ol can be used as fuel n compresson gnton engne. When bodesel s used as a substtute for desel, t s hghly essental to understand the parameters that affect the combuston phenomenon, whch wll n turn have drect mpact on thermal effcency and emsson. In the present energy scenaro, lot of effort s beng focused on mprovng the thermal effcency of IC engnes wth reducton n emssons (Srvastava Prasad 2000; Ramank 2003; Demrbas 2005). The present analyss reveals that bodesel
950 Int. J. Envron. Sc. Technol. (2014) 11:949 958 from unrefned jatropha, Karanja, and polanga seed ol s qute sutable as an alternatve to desel (Sahoo and Das 2009; Agrawal and Agrawal 2007). In developed and developng countres, fossl fuels are used n desel engnes. Desel engnes have a negatve effect on envronment, snce desel fuels nclude hgh amounts of sulfur and aromatcs. CO, SO x,no x, and smoke are produced from fossl fueled desel engne exhaust emssons (Kalam et al. 2003). It has been observed that engne parameters such as njecton tmng, compresson rato (CR) have consderable effects on the performance and emssons of desel engne runnng on bodesel blends. The oxygenated nature of bodesel becomes more advantageous whch tends to result n more complete combustons and reduce the CO emssons (An et al. 2012). Many nnovatve technologes are developed to tackle these problems. Modfcaton s requred n the exstng engne desgns. Some optmzaton approach has to be followed so that the effcency of the engne s not comprsed. As far as the nternal combuston engnes are concerned, the thermal effcency and emsson are the mportant parameters for whch the other desgn and operatng parameters have to be optmzed. The most common optmzaton technques used for engne analyss are response surface method, gray relatonal analyss (Agrawal and Rajamanoharan 2009), non-lnear regresson (Banapurmath et al. 2008), genetc algorthm (Alonso et al. 2007), and Taguch method; Taguch technque has been popular for parameter optmzaton n desgn of experments. Mult objectve optmzaton of parameters usng non-lnear regresson has found optmum value to be 13 % bodesel desel blend wth an njecton tmng of 24 Btdc (Maheswar et al. 2011). Blend of B30 thumba bodesel, a CR of 14, a nozzle openng pressure of 250 bar, and an njecton tmng of 20 produces maxmum multple performance of a desel engne wth mnmum multple emssons from the engne (Karnwal et al. 2011). A thermodynamc model analyss of jatropha bodesel engne n combnaton wth Taguch s optmzaton approach to determne the optmum engne desgn and operatng parameters was found out to maxmze the performance of bodesel engne (Ganapathy et al. 2009). Artfcal neural network (ANN) has been used to predct the performance and exhaust emssons of blended fuels (Xue et al. 2011; Najaf et al. 2009). It was reported that ANN can predct engne emssons and exhaust gas temperature, qute well wth correlaton coeffcents n the range of 0.983 0.996 (Canakc et al. 2006; Sayn et al. 2007; Ganapathy et al. 2009). Many researches about optmzaton and modfcaton on engne, low temperature performances of engne, new nstrumentaton and methodology for measurements should be performed when petroleum desel s substtuted completely by bodesel (Celk and Arcakloglu 2005). From the revew of lterature, t can be seen that whle a lot of work has been carred out to mprove the performance of bodesel fueled compresson gnton, studes on multobjectve optmzaton to determne the most sutable set of operatng varables, wth modern optmzaton technques are not many. Hence, the am of the present research s to set up an expermental study and to study the ndvdual and combned effects of combuston parameters on the performance and emsson characterstcs of the desel engne, employng Karanja bodesel desel blend, usng response surface methodology (RSM)-based expermental desgn, and the other objectve s to determne the optmal values of CR, blend, and power, whch would be resultng n mproved performance wth less emssons usng the desrablty approach. Ths research was done n Research laboratory of IC Engnes at Anjala Ammal Mahalngam Engneerng College Inda from June 2011 to March 2012. Materals and methods Fuel preparaton The vegetable ols were obtaned from commercal sources and used wthout further purfcaton. The samples were converted to methyl esters by alkal catalytc and noncatalytc super crtcal methanol transesterfcaton methods. Transesterfcaton (also called alcoholyss) s the reacton of a fat or ol wth an alcohol to form esters and glycerol (Sngh and Sngh 2010). Therefore, methanol (CH 3 OH) as an alcohol and potassum hydroxde (KOH) as a catalyst were used n the transesterfcaton. Molar rato between alcohol and ol used was 6:1, whereas catalyst amount was 1 % of the ol s weght. The experments were performed n a laboratory scale apparatus. Transesterfcaton was carred out n a 2,000 ml reacton flask, equpped wth reflux condenser, magnetc strrer, and thermometer. The catalyst was dssolved n methanol by strrng n a small flask. About 1,000 g of ol was added to the reacton flask and heated. When the temperature reached 65 C, the alcohol/catalyst mxture was added nto the ol and then the fnal mxture was strred for 3 h. After completon of strrng, the mxture was allowed to settle down for 24 h. After the transesterfcaton, the glycern layer was separated n a separatng funnel. The ester layer was washed wth warm water four tmes. After the fnal washng, the ester was subjected to a heatng at 100 C to remove excess alcohol and water. The fuel blend was prepared just before commencng the experments, to ensure the mxture homogenety. The propertes of the fuel blend and desel have been determned as per the ASTM Standards n an analytcal lab. The fuels propertes were tested usng standard measurng devces shown n Table 1.
Int. J. Envron. Sc. Technol. (2014) 11:949 958 951 Expermental setup The expermental setup conssts of a drect njecton sngle cylnder four stroke cycle desel engne connected to an eddy current type dynamometer for loadng. It s provded wth necessary nstruments for pressure and crank-angle measurements. These sgnals are nterfaced to computer through engne ndcator for P h AND PV dagrams. Provson s also made for nterfacng ar flow, fuel flow, temperatures, and load measurements. Ths setup has stand-alone panel box consstng of ar-flow, fuel measurng unt, transmtters for ar and fuel flow measurements, process ndcator, and engne ndcator. Rotameters are provded for coolng water and calormeter for water flow measurement. Detals of the engne specfcaton are shown n Table 2. The sgnals from the combuston pressure sensor and the crank angle encoder are nterfaced to a computer for data acquston. The control module system was used to control the engne load, montor the engne speed, and measure the fuel consumpton. Wndows based engne performance analyss software package Engne soft was provded for onlne performance evaluaton. HC, CO, CO 2, and K (ar surplus rate) NO x emssons were measured wth an nfra red gas analyzer wth an accuracy shown n Table 4. In every test, volumetrc fuel consumpton and exhaust gas emssons such as CO, HC, and NO x were measured. From the ntal measurement, brake thermal effcency (BTHE), brake specfc fuel consumpton (BSFC), brake power (BP) for dfferent blends and dfferent CR were calculated and recorded. Error analyss Errors and uncertantes n the experments can arse from nstrument selecton, condton, calbraton, envronment, observaton, readng, and test plannng. Errors wll creep nto all experments, regardless of the care whch s exerted. Uncertanty analyss s needed to prove the accuracy of the experments. In any experment, the fnal result s calculated from the prmary measurements. The error n the Table 2 Engne specfcaton Make and model Krloskar model TV 1 Engne type Sngle cylnder four stroke drect njecton Bore 9 stroke 87.5 mm 9 110 mm Maxmum power output 5.2 kw at 1,500 rpm Dsplacement 661 cc CR 17.5 Loadng Eddy current dynamometer, water coolng Fuel njecton 23 btdc Engne speed 1,500 rpm Software used Engne soft Governor type Mechancal centrfugal type Eddy current dynamometer Model AG-10 Type Eddy current Maxmum 7.5 kw at 1,500 3,000 rpm fnal result s equal to the maxmum error n any parameter used to calculate the result. Percentage uncertantes of varous parameters lke total fuel consumpton; BP, BSFC, and BTHE were calculated usng the percentage uncertantes of varous nstruments used n the experment. For the typcal values of errors of varous parameters gven n Table 4, usng the prncple of propagaton of errors, the total percentage uncertanty of an expermental tral can be computed. The total percentage uncertanty = Square root of [(uncertanty of brake power) 2? (uncertanty of SFC) 2? (uncertanty of TFC) 2? (uncertanty of BTHE) 2? (uncertanty of HC) 2? (uncertanty of CO) 2? (uncertanty of NO x ) 2? (uncertanty of pressure pck up) 2 ] =±1.85 %. Response surface methodology Response surface methodology s a collecton of statstcal and mathematcal technques useful for developng, mprovng, and optmzng processes. Table 1 Propertes of bodesel-blends Karanja and desel Fuel blend Knematcs vscosty, m (mm 2 /s) Heatng value, HV (KJ/kg) Flash pont, FP ( C) Densty, q (kg/l) Cetane number Desel 2.71 44,800 55 0.836 51.00 B20 3.04 43,690 96 0.851 51.70 B40 3.51 43,150 99 0.854 52.82 B50 3.62 43,307 106 0.856 53.15 B60 3.81 42,937 0.859 53.86 B100 4.37 42,133 163 0.900 54.53 Measurement and apparatus standard test method Redwood vscometer ASTM D445 Bomb calormeter ASTM D240 Penksy martens ASTM D93 Hydrometer ASTM D941 Ignton qualty tester ASTM D613
952 Int. J. Envron. Sc. Technol. (2014) 11:949 958 Table 3 Expermental desgn matrx Run order Compresson rato Fuel blends Power BTHE BSFC CO HC NO x (%) (kw) (%) (kg/kw h) (%) (ppm) (ppm) 1 17.5 10 3.64 34.84 0.24 0.22 69 1,258 2 17.5 10 4.16 35.42 0.23 0.42 69 1,319 3 17.5 10 4.68 34.94 0.24 1.11 70 1,335 4 17.5 10 5.2 33.41 0.26 2.28 71 1,306 5 17.5 20 3.64 31.31 0.28 0.14 69 1,239 6 17.5 20 4.16 31.63 0.27 0.3 69 1,282 7 17.5 20 4.68 30.89 0.28 0.84 74 1,279 8 17.5 20 5.2 29.10 0.31 1.76 84 1,232 9 17.5 30 3.64 29.45 0.30 0.11 69 1,206 10 17.5 30 4.16 29.50 0.29 0.25 67 1,230 11 17.5 30 4.68 28.50 0.31 0.72 81 1,209 12 17.5 30 5.2 26.45 0.34 1.53 112 1,143 13 17.5 50 3.64 30.68 0.27 0.17 67 1,097 14 17.5 50 4.16 30.22 0.28 0.34 61 1,084 15 17.5 50 4.68 28.69 0.30 0.93 110 1,027 16 17.5 50 5.2 26.12 0.34 1.95 214 924 17 17.7 10 3.64 34.38 0.26 0.12 69 1,061 18 17.7 10 4.16 35.28 0.24 0.26 70 1,173 19 17.7 10 4.68 35.12 0.25 0.74 69 1,241 20 17.7 10 5.2 33.91 0.27 1.58 68 1,264 21 17.7 20 3.64 30.97 0.30 0.06 70 1,075 22 17.7 20 4.16 31.60 0.28 0.17 70 1,169 23 17.7 20 4.68 31.19 0.29 0.55 67 1,218 24 17.7 20 5.2 29.72 0.32 1.20 62 1,223 25 17.7 30 3.64 29.21 0.31 0.04 70 1,074 26 17.7 30 4.16 29.59 0.31 0.13 71 1,150 27 17.7 30 4.68 28.91 0.32 0.47 63 1,181 28 17.7 30 5.2 27.18 0.35 1.04 47 1,167 29 17.7 50 3.64 30.67 0.29 0.08 70 1,031 30 17.7 50 4.16 30.53 0.29 0.20 74 1,070 31 17.7 50 4.68 29.32 0.31 0.62 48 1,064 32 17.7 50 5.2 27.07 0.34 1.34 73 1,013 33 17.9 10 3.64 33.17 0.27 0.10 69 883 34 17.9 10 4.16 34.39 0.26 0.23 69 1,048 35 17.9 10 4.68 34.56 0.25 0.68 72 1,167 36 17.9 10 5.2 33.67 0.27 1.45 78 1,242 37 17.9 20 3.64 29.87 0.31 0.05 69 929 38 17.9 20 4.16 30.83 0.30 0.15 66 1,076 39 17.9 20 4.68 30.74 0.30 0.05 87 1,177 40 17.9 20 5.2 29.59 0.32 1.10 133 1,234 41 17.9 30 3.64 28.23 0.33 0.02 67 962 42 17.9 30 4.16 28.93 0.35 0.11 59 1,090 43 17.9 30 4.68 28.57 0.31 0.42 122 1,173 44 17.9 30 5.2 27.16 0.30 0.95 256 1,211 45 17.9 50 3.64 29.91 0.31 0.06 61 984 46 17.9 50 4.16 30.09 0.30 0.18 33 1,075 47 17.9 50 4.68 29.21 0.32 0.56 248 1,121 48 17.9 50 5.2 27.28 0.35 1.23 70 1,122 49 18.1 10 3.64 31.22 0.29 0.16 69 726 50 18.1 10 4.16 32.77 0.27 0.33 68 942 51 18.1 10 4.68 33.25 0.26 0.91 78 1,114 52 18.1 10 5.2 32.69 0.27 1.90 98 1,240
Int. J. Envron. Sc. Technol. (2014) 11:949 958 953 Table 3 contnued Run order Compresson rato Fuel blends Power BTHE BSFC CO HC NO x (%) (kw) (%) (kg/kw h) (%) (ppm) (ppm) 53 18.1 20 3.64 28.30 0.33 0.10 66 805 54 18.1 20 4.16 29.32 0.31 0.23 56 1,003 55 18.1 20 4.68 29.54 0.31 0.68 133 1,156 56 18.1 20 5.2 28.72 0.32 1.45 298 1,264 57 18.1 30 3.64 26.50 0.35 0.07 60 870 58 18.1 30 4.16 27.52 0.33 0.18 31 1,050 59 18.1 30 4.68 27.49 0.33 0.58 258 1,184 60 18.1 30 5.2 26.40 0.35 1.26 74 1,274 61 18.1 50 3.64 28.41 0.32 0.12 39 957 62 18.1 50 4.16 28.91 0.31 0.27 60 1,100 63 18.1 50 4.68 28.35 0.32 0.76 71 1,198 64 18.1 50 5.2 26.74 0.35 1.61 235 1,251 The most extensve applcatons of RSM are n the partcular stuatons, where several nput varables potentally nfluence some performance measure or qualty characterstc of the process. Thus, performance measure or qualty characterstc s called the response. The nput varables are sometmes called ndependent varables, and they are subject to the control of the scentst or engneer. The feld of RSM conssts of the expermental strategy for explorng the space of the process or ndependent varables, emprcal statstcal modelng to develop an approprate approxmatng relatonshp between the yeld and the process varables, and optmzaton methods for fndng the values of the process varables that produce desrable values of the response. Response surface methodology was employed n the present study for modelng and analyss of response parameters to obtan the characterstcs of the engne. The desgn and analyss of experment nvolved the followng steps: The frst step was the selecton of the parameters that nfluence the performance and emsson characterstcs. In ths study, the CR, fuel blends, and power were consdered as the nput parameters. The CR (denoted by CR ) was vared at four levels n steps of 0.2 from 17.5 to 18.1. The fuel blends (denoted by B ) too was vared from 10 to 50 %. The power (denoted by P ) was vared from 3.64 to 5.2 kw. The advantage of usng desgn of experments s to evaluate the performance of the engne over the entre range of varaton of CR and other parameters wth mnmum number of experments. The desgn matrx was selected based on the 3 level factor desgn of RSM generated from the software Desgn Expert verson 8.0.7.1 of stat ease, US, whch contaned 64 expermental runs as shown n Table 3. As per the run order, the experments were conducted on the engne, and the responses were fed on the responses column. Table 4 The accuraces and uncertanty of the measured and calculated results Measurements Accuracy Percentage uncertanty Engne speed ±1 rpm ±0.2 Temperatures ±1 C ±0.1 Carbon monoxde ±0.02 % ±0.2 Hydrocarbon ±10 ppm ±0.2 Carbon doxde ±0.03 % ±1.0 Ntrogen oxdes ±20 ppm ±0.2 Burette measurement ±2CC ±1.5 Crank angle encoder ±0.5 CA ±0.2 Load ±1 N ±0.2 Calculated results Power ±0.2 Fuel consumpton ±1.5 Brake thermal effcency ±2.58 A multple regresson analyss was carred out to obtan the coeffcents and the equatons can be used to predct the responses. Usng the statstcally sgnfcant model, the correlaton between the process parameters and the several responses were obtaned. Fnally, the optmal values of the CR, fuel blends, and power parameters were obtaned by usng the desrablty approach of the RSM. Desrablty approach The real-lfe problems requre optmzaton wth the multple responses of nterest. Technques lke overlyng the contour plots for each response, constraned optmzaton problems, and desrablty approach are found to have benefts lke smplcty, avalablty n the software, and flexblty n weghtng and gvng mportance for ndvdual response. In the present work, RSM-based,
954 Int. J. Envron. Sc. Technol. (2014) 11:949 958 desrablty approach s used for the optmzaton of nput parameters lke CR, fuel blends, and power for the measured propertes of responses (BTHE, BSFC, CO, HC, and NO x ). The optmzaton analyss s carred out usng Desgn Expert software, where each response s transformed to a dmensonless desrablty value (d) and t ranges between d = 0, whch suggests that the response s completely unacceptable and d = 1, whch suggests that the response s more desrable. The goal of each response can be ether maxmum, mnmum, target, n the range and/or equal to dependng on the nature of the problem. The desrablty of each response can be calculated by the followng equatons wth respect to the goal of each response. For a goal of mnmum, d = 1 when Y B Low ; d = 0 when Y C Hgh and d ¼ Hgh Y wt when Low \ Y \ Hgh Hgh Low For a goal of maxmum, d = 0 when Y B Low ; d = 1 when Y C Hgh and d ¼ Y Low wt when Low \Y \Hgh Hgh Low For goal as target, d = 0, when Y \ Low ; Y [ Hgh. d ¼ Y Low wt1 when Low \ Y \ T T Low d ¼ Y Hgh wt2 when T \ Y \ Hgh T Hgh ; and For the goal wthn the range, d = 1 when low \ Y \ hgh and d = 0. Here ndcates the response, Y the value of response, Low represents the lower lmt of the response, Hgh represents the upper lmt of the response, T means the target value of the response, and wt ndcates the weght of the response. The shape of the desrablty functon can be changed for each response by the weght feld. Weghts are used to gve more emphass to the lower/ upper bounds. Weghts can be ranged from 0.1 to 10; a weght greater than 1 gves more emphass to the goal, weghts less than 1 gve less emphass. When the weght value s equal to one, the desrablty functon vares n a lnear mode. Solvng of multple response optmzatons usng the desrablty approach nvolves a technque of combnng multple responses nto a dmensonless measure of performance called the overall desrablty functon. In the overall desrablty objectve functon (D), each response can be assgned an mportance (r), relatve to the other responses. Importance vares from the least mportant value of 1, ndcated by (?), the most mportant value of 5, ndcated by (?????). A hgh value of D ndcates the more desrable and the best functons of the system, whch s consdered as the optmal soluton. The optmum values of factors are determned from value of ndvdual desred functons (d) that maxmzes D (Pandan et al. 2011). Results and dscusson Analyss of the model The prncpal model analyss was based on the analyss of varance (ANOVA) whch provdes numercal nformaton for the p value. The models found to be sgnfcant as the values of p were less than 0.05. The dfferent models for the responses were developed n terms of actual factors and are gven below as Eqs. (1) (5). BTHE ¼ 2636:22 þ 314:053 A 1:39518 B 37:2864 C þ 0:055847 A B þ 3:09128 A C 0:050291 B C 9:32265 A 2 þ 8:28488 10 4 B 2 1:94840 C 2 ð1þ BSFC ¼ 2:21394 þ 0:084203 A þ 6:26954 10 3 B þ 0:52798 C 1:06040E 004 A B 0:046050 A C þ 7:12892 10 4 B C þ 4:9192 10 3 A 2 1:03086 10 4 B 2 þ 0:032402 C 2 ð2þ Table 5 Response surface model evaluaton Model BTHE BSFC CO HC NO x Mean 30.2839 0.30499 0.630491 152.48 1124.66 SD 2.554 0.032 0.579192 314.8 130.059 R 1.000 1.000 0.9958 0.9392 1.000 Model degree Quadratc Quadratc Modfed Modfed Quadratc Adj. R 2 1.000 1.000 0.9939 0.9307 1.000 Pred. R 2 1.000 1.000 0.9867 0.9147 1.000
Int. J. Envron. Sc. Technol. (2014) 11:949 958 955 CO ¼ 3570:26684 þ 404:90374 A 2:84265 B þ 1012:05362 C þ 0:32413 A B 115:68034 A C 0:042085 B C 11:45397 A 2 þ 1:02510 10 4 B 2 þ 4:23397 C 2 þ 3:29041 A 2 C 0:20116 AC 2 þ 6:44070 10 3 B 2 C ð3þ HC ¼ 81:67 þ 14:19 A þ 1:59 B þ 47:2 C þ 8:84 A B þ 32:45 A C þ 24:05 B C þ 17:60 A 2 15:96 B 2 þ 26:43 C 2 þ 15:02 A B C þ 8:15 A 2 B þ 30:90 A 2 C 10:93 A B 2 þ 17:05 A C 2 29:74 B 2 C þ 12:23 B C 2 þ 1:758E 004 A 3 þ 1:250E 004 B 3 þ 1:758E 004 C 3 : NO x ¼ 1:28147 10 5 11691:4 A 273:185 B 7934:06 C þ 16:3619 A B þ 499:11 A C 3:54044 B C þ 247:846 A 2 0:071363 B 2 83:0965 C 2 : where A CR, B fuel fracton n %, C power n kw Evaluaton of the model ð4þ ð5þ The stablty of the models was valdated usng Analyss of varance (ANOVA). The output showed that the model was sgnfcant wth p values less than 0.0001. The reference lmt for p was chosen as 0.05. The regresson statstcs goodness of ft (R 2 ) and the goodness of predcton (Adjusted R 2 ) are shown n Table 5 for all the responses. The R 2 value ndcates the total varablty of response after consderng the sgnfcant factors. The (adjusted R 2 ) value accounts for the number of predctors n the model. Both the values ndcate that, the model fts the data very well. varaton n CR on the BTHE ndcate that hgher CRs mprove the engne effcency. Ths can be attrbuted to better combuston and hgher lubrcty of bodesel. As seen n Table 3, the ncreased CR ncreased the BTHE by 2 % for B10 compared to the results of orgnal CR. By ncreasng the CR of the engne, the BTHE also gets ncreased for all the fuel types tested. BTHE s drectly proportonate to the CR. Brake specfc fuel consumpton (BSFC) As shown n Table 3, the BSFC generally ncreased wth the ncrease n bodesel percentage n the fuel blend. It can be consdered that the decrease n the lower heatng value of the blends by addng bodesel requres more fuel to be njected nto the cylnder to get the same power output, leadng to the ncrease n the BSFC (Doddayaraganalu et al. 2010). When there s an ncrease n CR,the maxmum cylnder pressure ncreases due to the fuel njected n hotter combuston chamber and ths leads to hgher effectve power. Therefore, fuel consumpton per output wll decrease. As the BSFC s calculated on weght bass, obvously hgher denstes resulted n hgher values for BSFC. As densty of Karanja bodesel was hgher than that of bodesel for the same fuel consumpton, on volume bass, pure bodesel yelds hgher BSFC. The hgher denstes of bodesel blends caused hgher mass njecton, for the same volume, at the same njecton pressure. The calorfc value of the bodesel s less than desel. Due to these reasons, the BSFC for the other blends was hgher than that for desel. Smlar trends of decrease n the BSFC value wth ncreasng load for dfferent bodesel were also reported by other researchers (Baju et al. 2009) whle testng bodesel obtaned from Karanja. Engne emssons Converson of bodesel chemcal energy under hgh pressure and temperature n CI engnes produces emssons Brake thermal effcency (BTHE) Brake thermal effcency evaluates how effcent the engne transforms the chemcal energy of the fuel nto useful work. Ths parameter s determned by dvdng the BP of the engne by the amount of energy nput to the system. The percentage change n the BTHE s shown n Table 3. The BTHE usually ncreases wth the ncrease n bodesel percentage n the fuel blend. Thus, the prmary reason for the decrease n the BTHE of bodesel s the hgher BSFC n spte of lower LHV of bodesels. The maxmum BTHE s 35 % for the CR 17.9 and fuel blend between B10 and B20, whereas low BTHE les n the regon around 17.7 CR and fuel blend between B30 and B40. The effects of the Fg. 1 The HC varatons aganst compresson rato and power
956 Int. J. Envron. Sc. Technol. (2014) 11:949 958 such as CO 2,NO x, PM, CO, HC, and aromatc compounds (Jo-Han et al. 2010). The engne operatng parameters, such as ar fuel equvalence rato, fuel type, combuston chamber desgn, and atomzaton rato affect, wth all emssons emtted by nternal combuston engnes, especally, emssons of CO and unburned HC n the exhaust are very mportant, snce they represent the low chemcal energy that cannot be totally used n the engne. Emssons such as CO 2,NO x emtted by desel engne have mportant effects on ozone layer and human health (Aksoy 2011). The engne emssons wth Karanja bodesel have been evaluated n terms of CO, HC, and NO x at varous CR, at dfferent loadng condtons of the engne. Hydrocarbon (HC) It s seen n Fg. 1 that there s a sgnfcant decrease n the HC emsson level wth Karanja ol as compared to pure desel. The HC emsson s a mnmum of 33 ppm whch occurs at CR (18.1) and blend B30, so the value of HC reduces as CR and fuel blend ncreases. At the hgher CR, UBHC was low. Ths may be due to ncreased temperature and pressure at hgher CR and better combuston can be ensured (Muraldharn and Vasudevan 2011). Hydrocarbon concentraton decreases wth bodesel addton and ths suggests that addng oxygenate fuels can decrease HC from the locally over rch mxture. Furthermore, oxygen enrchment s also favorable to the oxdaton of HC n the expanson and exhaust process (Huang et al. 2005). As confrmed n Fg. 1, ncreased CR reduced the HC emssons by 4 % and reduced CR ncreased the HC emssons. At lower CR, nsuffcent heat of compresson delays gnton, and so HC emssons ncrease (Jndal et al. 2010). These reductons ndcate the more complete combuston of the fuels and, thus, HC level decreases sgnfcantly. The reducton n HC emsson was lnear wth the addton of bodesel for the blends. The maxmum and mnmum UBHC produced s 0.0299 g/kw h and 0.01554 g/kw h whch s less than the EURO-IV norms (0.5 g/kw h). Carbon monoxde (CO) The varaton n CO of the engne s shown n Fg. 2. As vewed n Fg. 2, ncreased CR decreased the CO emssons by 37.09 % and reduced CR ncreased CO emssons by 9.67 % compared to the results of orgnal CR for B 100. At lower CR, nsuffcent heat of compresson delays gnton and so CO emssons ncrease (Sayn et al. 2007). The possble reason for ths trend could be that the ncreased CR actually ncreases the ar temperature nsde Fg. 2 The CO varatons aganst compresson rato and power the cylnder therefore reducng the gnton lag causes better and more complete burnng of the fuel (Raheman and Ghadge 2008), the percentage of CO s less than 0.3 % at CR 17.7, B20 and maxmum percentage of 1.9 % at CR 17.5, B50. The lower CO emssons of bodesel blends may be due to ther more complete oxdaton as compared to desel. Some of the CO produced durng combuston of bodesel mght have been converted nto CO 2 by takng up extra oxygen molecule present n the bodesel chan and thus, reduces CO formaton. The maxmum and mnmum CO produced s 2.886 g/kw h and 0.222 g/kw h, whch s less than the EURO-IV norms (4 g/kw h). Ntrogen oxdes (NO X ) The NO x values for dfferent fuel blends at varous CR are shown n Table 3. The amount of NO x produced for B10 B50 s the range of 720 1,300 ppm as compared to desel whch vares from 300 to 900 ppm. It can be seen that the ncreasng proporton of bodesel n the blends ncreases NO x as compared wth desel. Ths could be attrbuted to the ncreased exhaust temperatures and the fact that bodesel had some oxygen content n t whch facltated NO x formaton. Snce the sze of njected partcles of vegetable ols s bgger than that of desel fuel, combuston effcency and maxmum combuston temperatures wth vegetable ols were lower. Therefore, NO x emssons were lower (Ramadhas et al. 2004). As llustrated n the Fg. 3 ncreased CR ncreased the NO x emssons by 10 % and reduced CR decreased NO x emssons by 12 %, compared to the results of orgnal CR for B50. Reduced CR s to reduce the n-cylnder temperatures and thus flame temperatures durng the combuston to suppress NO x emssons (Raheman and Ghadge 2008). NO x emssons were also hgher at part loads for bodesel. Ths s probably due to hgher bulk modulus of bo-desel, resultng n a dynamc njecton advance, apart from statc njecton advance provded for optmum effcency. Excess oxygen (10 %)
Int. J. Envron. Sc. Technol. (2014) 11:949 958 957 average of expermental values, predcted values and the percentage of error. The valdaton results ndcated that the model developed was qute accurate as the percentage of error n predcton was n a good agreement. Concluson Fg. 3 present n bo-desel would have aggravated the stuaton (Pradeep and Sharma 2007).The maxmum and mnmum NO x produced s 0.3644 and 0.213 g/kw h, whch s less than the EURO-IV norms(3.5 g/kw h). Optmzaton The NO x varatons aganst compresson rato and power The crtera for the optmzaton, such as the goal set for each response for lower and upper lmts used, weght used, and mportance of the factors are presented n Table 6. In desrablty-based approach, dfferent best solutons were obtaned. The soluton wth hgh desrablty was preferred. Maxmum desrablty of 0.978 was obtaned at the followng compresson system parameters lke 17.9 of CR, 10 % of fuel blend, and 3.81 kw of power whch could be consdered as the optmum parameters for the test engne havng 5.2 kw as rated power at 1,500 rpm. Valdaton of optmzed result In order to valdate the optmzed result, the experments were performed thrce at the optmum compresson system parameters. For the actual responses, the average of three measured results was calculated. Table 7, summarzes the Based on the results of ths study, the followng conclusons were drawn n terms of fuel propertes and exhaust emsson characterstcs. Karanja ol methyl ester can be regraded as an alternatve to desel fuel. The desgn of experments was hghly helpful to desgn the experment and the statstcal analyss helped to dentfy the sgnfcant parameters whch are most nfluencng on the performance emsson characterstcs. Ths expermental desgn consderably reduced the tme requred by mnmzng the number of experments to be performed and provded statstcally proven models for all response. It s clear from ths research that CO and HC emssons have been reduced when bodesel s fueled nstead of desel. Advancng the CR from 17.5 to 18.1 helped to decrease the CO and HC emssons. Decreasng the fuel blend ratos contrbuted for better BTHE wth lesser BSFC wth lower CO, HC and NO x values. However, when too low was the blend rato, the results were good. The maxmum BTHE for B10 (35.42 %) was hgher than that of desel at full load. Desrablty approach of the RSM was found to be the smplest and effcent optmzaton technque. A hgh desrablty of 0.97 was obtaned at the optmum engne parameters of CR of 17.9, fuel blend B10, and 3.81 kw power, where the values of the BTHE, BSFC, CO, HC, and NO x were found to be 33.65 %, 0.2718 kg/kw -1 h -1, 0.109 %, 158, and 938 ppm, respectvely. Table 6 Optmzaton crtera and desrablty response Source Lower lmts Upper lmts Weght Importance Goal Desrablty Upper Lower Compresson rato 17.5 18.1 1 1 3 In range 1 Fuel fracton 10 50 1 1 3 In range 1 Power 3.64 5.2 1 1 3 In range 1 BTHE 26.12 35.42 1 0.1 5 Maxmze 0.9963 BSFC 0.234 0.358 0.1 1 5 Mnmze 0.9574 CO 0.028 2.287 0.1 1 5 Mnmze 0.994 HC 31.73 298 0.1 1 5 Mnmze 0.979 NO x 725.6 1334.7 0.1 1 5 Mnmze 0.9648 Combned 0.978
958 Int. J. Envron. Sc. Technol. (2014) 11:949 958 Table 7 Comparson of actual and predcted values S. no. Value Compresson rato Fuel fracton Power (kw) BTHE (%) BSFC (Kg/Kw h) CO (%) HC (ppm) NO x (ppm) 1 Predcted 17.9 10 3.81 33.65 0.2718 0.109 158.03 938.3 2 Actual 17.9 10 3.81 33.24 0.2783 0.127 156.86 940.45 3 Error -0.41 0.0065 0.018-1.17-2.15 Acknowledgments The authors are grateful to the management of Anjala Ammal Mahalngam Engneerng College Kovlvenn, Tamlnadu, for provdng the laboratory facltes to carry out the research. Nomenclature P Power (kw) RSM Response surface methodology B Blend fracton (%) CR Compresson rato BTHE Brake thermal effcency BSFC Brake specfc fuel consumpton (kg kw 1 h 1 ) btdc Before top dead center FF Fuel fracton References Agrawal D, Agrawal AK (2007) Performance and emsson characterstcs of a jatropha ol preheated and blend n a drect njecton compresson gnton engne. Appl Therm Eng 27(13):2314 2323 Agrawal AK, Rajamanoharan K (2009) Expermental nvestgatons of performance and emssons of karanja ol and ts blends n a sngle cylnder agrcultural desel engne. Appl Energy 86(1):106 112 Aksoy F (2011) The effect of opum poppy ol desel fuel mxture on engne performance and emssons. Int J Envron Sc Tech 8(1):57 62 Alonso JM, Alvarruz F, Deantes JM, Hernandez L, Hernandez V, Molto G (2007) Combnng neural networks and genetc algorthms to predct and reduce desel engne emsson. IEEE Trans Evol 11:46 55 An H, Yang WM, Chou SK, Chua KJ (2012) Combuston and emssons characterstcs of desel engne fueled by bodesel at partal load condtons. Appl Energy 99(1):363 371 Baju B, Nak MK, Das LM (2009) A comparatve evaluaton of compresson gnton engne characterstcs usng methyl and ethyl esters of Karanja ol. Renew Energy 34(9):1616 1621 Banapurmath NR, Tewar TG, Hosmath RS (2008) Performance and emsson characterstcs of DI compresson gnton engne operated on honge, jatropha and sesame ol methyl esters. Renew Energy 33(10):1982 1988 Canakc M, Erdl A, Arcaklog E (2006) Performance and exhaust emssons of a bodesel engne. Appl Energy 83(6):594 605 Celk V, Arcakloglu E (2005) Performance maps of a desel engne. Appl Energy 81(3):247 259 Demrbas A (2005) Bodesel producton from vegetable ols va catalytc and non-catalytc supercrtcal methanol trans esterfcaton methods. Int J Energy Combust 31:466 487 Doddayaraganalu C, Vsweswara S, Padmanbha M (2010) Combuston characterstcs of desel engne operatng on jatropha ol methyl ester. Ther Sc 14(4):965 977 Ganapathy TK, Murugesan KRP, Gakkhar RP (2009) Performance optmzaton of jatropha engne model usng Taguch approach. Appl Energy 86(3):2476 2486 Huang Z, Lu H, Jang D, Zeng K, Lu B, Zhang J, Wang X (2005) Performance and emssons of a CI engne. Fueled wth desel/ oxygenate blends for varous fuel delvery advance angles. Energy Fuels 19:403 410 Jndal S, Nandwana BP, Rathore NS, Vashstha (2010) Expermental nvestgaton of the effect of compresson rato and njecton pressure n a DI desel engne runnng on jatropha methyl ester. Appl Ther Eng 30(5):442 448 Jo-Han NG, Hoon Kat NG, Suyn G (2010) Advances n bodesel fuel for applcaton n compresson gnton engnes. Clean Technol Envron Polcy 12(5):459 493 Kalam M, Husnawan A, Masjuk M (2003) Exhaust emsson and combuston evaluaton of coconut ol-powered ndrect njecton desel engne. Renew Energy 16(6):2405 2415 Karnwal M, Hasan N, Kumar A, Sddquee N, Khan ZA (2011) Multresponse optmzaton of desel engne performance parameters usng Thumba bodesel desel blends by applyng the taguch method and grey relatonal analyss. Int J Automot Technol 12(4):599 610 Maheswar N, Balaj C, Ramesh A (2011) A nonlnear regresson based mult-objectve optmzaton of parameters based on expermental data from an IC engne fueled wth bodesel blends. Bomass Boenergy 35(2):2171 2183 Muraldharn K, Vasudevan D (2011) Performance, emsson and combuston characterstcs of a varable compresson rato engne usng esters of waste cookng ol and desel blends. Appl Energy 88(11):3959 3968 Najaf G, Ghobadan B, TavakolButtsworth R, Yusaf TF, Fazollahnaejad M (2009) Performance and exhaust emssons of a gasolne engne wth ethanol blended gasolne fuels usng artfcal neural network. Appl Energy 86(5):630 639 Pandan M, Svaprakasam SP, Udayakumar M (2011) Investgaton on the effect of njecton system parameters on the performance and emsson characterstcs of a twn cylnder compresson gnton drect njecton engne fuelled wth pongama bodesel desel blend usng response surface methodology. Appl Energy 88(8):2663 2676 Pradeep V, Sharma RP (2007) Use of hot EGR for NO x control n a compresson gnton engne fuelled wth bo-desel from jatropha ol. Renew Energy 32(7):1136 1154 Raheman H, Ghadge SV (2008) Performance of desel engne wth bodesel at varyng compresson and gnton tmng. Fuel 87(12):2659 2666 7 pages Ramank PK (2003) Propertes and use of Jatropha curcas ol and desel fuels blends n compresson gnton engne. Int J Renew Energy 28(3):239 248 Ramadhas AS, Jayaraj S, Muraleedharan C (2004) Use of vegetable ols as IC engnes fuels a revew. Renew Energy 29(5):727 742 Sahoo PK, Das LM (2009) Process optmzaton for bodesel producton from Jatropha, Karanja and Polanga ols. Fuel 88(9):1588 1594 Sayn C, Ertunc HM, Hosoz M, Klcaslan I, Canakc M (2007) Performance and exhaust emssons of a gasolne engne usng artfcal neural network. Appl Thermal Eng 27(1):46 54 Sngh SP, Sngh Dpt (2010) Bodesel producton through the use of dfferent sources and characterzaton of ols and ther esters as substtute of desel. Renew Sustan Energy Rev 14(1):200 216 Srvastava Prasad A (2000) Trglycerdes-based desel fuels. Renew sustan Energy Rev 4(1):111 133 Xue J, Tony E, Grf A, Hansen C (2011) Effect of bodesel on engne performances and emssons. Renew Sustan Energy Rev 15(2): 1098 1116