research article quadratic prediction models for the...

Research ArticleQuadratic Prediction Models for the Performance Comparisonof a Marine Engine Fuelled with Biodiesels B5 and B20

Chedthawut Poompipatpong12

1 Automotive Eco-Energy and Industrial Product Standard Research Center King Mongkutrsquos University of Technology North Bangkok1518 Pracharat 1 Wongsawang Bangsue 10800 Bangkok Thailand

2Department of Automotive Engineering Technology King Mongkutrsquos University of Technology North Bangkok1518 Pracharat 1 Wongsawang Bangsue 10800 Bangkok Thailand

Correspondence should be addressed to Chedthawut Poompipatpong chedthawutpkmutnbacth

Received 31 May 2014 Revised 2 September 2014 Accepted 16 September 2014 Published 30 September 2014

Academic Editor Z X Guo

Copyright copy 2014 Chedthawut PoompipatpongThis is an open access article distributed under the Creative CommonsAttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

According toThailandrsquos renewable energy development plan biodiesel is one of the interesting alternative energies In this researchbiodiesels B5 and B20 are tested in amarine engineThe experimental results are then compared by using three different techniquesincluding (1) the conventional technique (2) average of the point-to-point comparisons and (3) a comparison by using quadraticpredictionmodelsThis research aims to present the procedures of these techniques in-depthThe results show that the comparisonby using quadratic prediction models can accurately predict ample amounts of results and make the comparison more logicalThe results are compatible with those of the conventional technique while the average of the point-to-point comparisons showsdiverse results These results are also explained on the fuel property basis confirming that the quadratic prediction model and theconventional technique are practical but the average of the point-to-point comparison technique presents an inaccurate resultThebenefit of this research shows that the quadratic prediction model is more flexible for future science and engineering experimentaldesign thus reducing cost and time usage The details of the calculation results and discussion are presented in the paper

1 Introduction

Alternative energy is one of the most interesting issues in thecurrent situation Biodiesel diesohol and pyrolysis oil havebeen improved and are replacing the usage of conventionalautomotive diesel fuel because some of them show ecologicalbenefits and some economic benefits An important topicof the investigation of alternative fuels is the differences ofengine performance and emission concentrations [1 2] Theexperimental results of conventional fuel are the referencevalues Then the results of alternative fuels are compared tothe reference values The literature generally reports thesedifferences in units of percentage for ease of understanding[3 4]

For example in order to compare engine performancewhen operated with two different fuels the output perfor-mances over thewhole test condition of each fuel are averagedand then these values are compared This process is doneby research [5 6] and is called the ldquoconventional techniquerdquo

in this research This conventional comparison techniqueis straightforward and acceptable However the weak pointis that the number of experimentations is limited in manyscientific and engineering researches due to the cost andtime consumption Hence this technique works only onexperimental results and it cannot be assured that the numberof experimentations will be sufficient to be the representativeof the entire engine performance and emission

This research experimentationwas done on a six-cylindermarine diesel engine fuelled with biodiesels B5 and B20since biodiesel B5 is currently used in Thailand and theofficial government plan is approaching biodiesel B20 Sincethe objective of this research was to present the details ofthree different comparison techniques engine torque wasselected as the example throughout the paper because it isone of the most important diesel engine parameters [7] Themost outstanding benefits of this research are that (1) thedifference of using biodiesels B5 and B20 is depicted in termsof engine performance and (2) the details of three comparison

Hindawi Publishing CorporationInternational Journal of Engineering MathematicsVolume 2014 Article ID 104989 6 pageshttpdxdoiorg1011552014104989

2 International Journal of Engineering Mathematics

(1) Diesel engine (8) Hot water reservoir(2) Intake manifold (9) Cooling tower(3) Exhaust manifold (10) Airflow meter(4) Engine dynamometer (11) Exhaust gas temperature sensor(5) Dynamometer controller (12) Engine speed sensor(6) Water supply tank (13) Inlet coolant temperature(7) Engine coolant supply tank (14) Outlet coolant temperature

14

7

89

6

5

12

10

11

13

14

Figure 1 Experimental setup

techniques are presented This will assist researchers inselecting the most appropriate comparison technique forreliable and accurate results in the future

2 Experimental Setup and Engine Testing

In order to accomplish the main objective of this researcha GARDNER 6-cylinder 10450 cc marine diesel engine wasinstalled on a Clayton water brake dynamometer which sup-plied the engine load The fuel consumption was measuredon a geometric basis using a digital scale and a stop watchA KIMOCP200 and a Debimo airflowmeasuring blade wereattached at the intake manifold for the airflow measurementThermocouples were connected at several positions in orderto collect the raw data A closed-loop water cooling systemwith a cooling towerwas utilizedThe engine temperaturewascontrolled by a coolant temperature controller The diagramof the experimental setup is shown in Figure 1

The experimentationwas conducted at full load conditionover the engine speed range between 1000 and 1400 rpmwith an interval of 100 rpm Thirty raw data were recordedfor every test condition and the average values were pre-sented The engine was first operated with biodiesel B5 forexperimentation After the test the engine was switched tobiodiesel B20 and run for a while in order to flush out thebiodiesel B5 from the fuel line and injection system Finallybiodiesel B20 was tested at exactly the same condition Thebiodiesels B5 and B20 in this research were supplied by thePetroleum Authority of Thailand Public Company Limited

Table 1 Fuel properties

Properties Regulation forbiodiesel B5

Biodiesel B20 fromPTT

Cetane number ge50 5734Specific gravity From 081 to 087 08408Viscosity at 40∘C From 18 to 41 3774Flash point (∘C) ge52 790Pour point (∘C) le10 9Water and sediment( vol) le005 lt0005

Heating value (Jg) 4123855 4523333

(PTT) The fuel properties conformed to the regulations ofthe Department of Energy Business Thailand Ministry ofEnergy The details of the fuel specifications are shown inTable 1

3 Comparison Techniques and the Results

Raw data were collected from the engine experiment andwere then analyzed and compared by using three differentcomparison techniques The results of every technique werereported in percentage units () because this method ismore reasonable and comparable than when the informationis presented in raw data Nevertheless there are severalmethods that are used to calculate percentages and whichprovide dissimilar results This paper presents the details ofthree methods including one conventional method and twonew comparison methods which are (1) ldquothe comparison ofthe average values (conventional technique)rdquo (2) ldquothe averageof the point-to-point comparisons (the new technique 1)rdquo and(3) ldquothe comparison by applying quadratic prediction model(the new technique 2)rdquo It is also important to note that theoutput torque was selected as the representative of the engineperformance throughout the paper

31 Methodology of Conventional Technique A Comparison ofthe Average Values This technique is the simplest and mostcommonly used comparison technique It consists of onlytwo main steps which are the averaging of the output dataand then the average data of biodiesel B20 is compared tothat of biodiesel B5 (reference value) by using the percentagedeviation method (PD) This method is called ldquoaveragingbefore comparingrdquo or ldquothe comparison of the average valuesrdquoThe detailed methodology is shown in Figure 2

Figure 3 shows that biodiesel B5 produces slightly higheroutput torque than that of biodiesel B20 All of the experi-mental data in Figure 5 are also summarized in Table 2 Theaverage output torques over the speed conditions of 1000to 1400 rpm were 54950Nm and 50806Nm for biodieselsB5 and B20 respectively The difference was basically deter-mined using percentage deviation method The comparisonresults using the conventional technique show that the enginefuelled with biodiesel B20 generates 754 lower torque thanthat of the engine fuelled with biodiesel B5

International Journal of Engineering Mathematics 3

Start

End

Comparison by percentage deviation method

Average torque from B5 Average torque from B20

Figure 2 Methodology of the conventional technique

7851073528

64012

41510

17189

7391168880

61073

35438

14730100

200

300

400

500

600

700

800

900

1000

950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Torq

ue (N

m)

Engine speed (rpm)

Torque-biodiesel B5Torque-biodiesel B20

Figure 3 Experimental results

32 Methodology of the New Technique 1 The Average of thePoint-to-Point Comparisons The purpose of presenting thistechnique in this paper is to depict its advantage Point-to-point comparison reports a set of detailed results forresearchersThis technique is very similar to the conventionaltechnique The only difference is in the sequence betweenaveraging and comparing The two important steps of thistechnique are first the point-to-point comparison betweenbiodiesel B5 and biodiesel B20 Then the results of thepoint-to-point comparisons are averaged This means thatthis technique employs the ldquocomparing before averagingrdquomethod This method is more complicated especially whenthere are many experimental data The methodology of thistechnique is presented in Figure 4

Table 3 shows that the output torques of biodiesels B5and B20 were compared in every tested conditionThereforethe process included five pair-wise comparisons Finally theresults were averaged It can be concluded concerning thenew technique 1 that the engine fuelled with biodiesel B20produces 914 lower torque than that of the engine fuelledwith biodiesel B5

33 Methodology of the New Technique 2 The Compari-son by Using Quadratic Prediction Model According to the

Start

End

Point-to-point comparison by using percentage deviation method

Calculation for the average or mean value

Figure 4 Methodology of the new technique 1

Table 2 The comparison result using the conventional technique

Speed (rpm) Output torque (Nm)Biodiesel B5(reference) Biodiesel B20

1000 78510 739111100 73528 688801200 64012 610731300 41510 354381400 17189 14730Average 54950 50806

PD = (50806 minus 54950)54950

times 100 = minus754

experimentation all of the calculations using conventionaltechniques and the new technique 1 were based on onlyfive levels of speed 1000 1100 1200 1300 and 1400 rpmIt is understandable that only five levels of speed cannotbe representative of all of the values between 1000 and1400 rpm and this limitation puts the comparison resultsin doubt This is the reason why the new technique 2 isrelatively different from the previous techniquesmdashit is morecomplex but also more flexible and reliable The procedureis shown in Figure 5 This technique will help researchers inovercoming this limitation by systematically integrating fourbasic mathematical steps as follows

331 Mathematical Modelling and Quadratic PredictionModel Plot Engine torque ismodelled as a function of enginespeed (119909) The second-order quadratic model is normallysuggested since it is sufficient to fit the data in most cases[8ndash10] without overfitting [11] The quadratic model containsa squared term (1199092) a linear term (119909) and a constant term(119888) which have coefficients of 119886 and 119887 as shown in (1) Atthe end of this step two quadratic prediction models weregenerated One was the representative of biodiesel B5 andthe other was of biodiesel B20 They are shown in (2) and(3) respectively The prediction curves are also presented inFigure 6 Consider

Torque = 1198861199092 + 119887119909 + 119888 (1)


Table 3 The comparison result by using the new technique 1

Speed (rpm) Output torque (Nm) Point-to-point percentage deviationBiodiesel B5 (reference) Biodiesel B20

1000 78510 73911 minus5861100 73528 68880 minus6321200 64012 61073 minus4591300 41510 35438 minus14631400 17189 14730 minus1431

Mean percentage deviation (MPD) minus914

Start

End

Mathematical modeling andquadratic prediction model plot

Accuracy check

Determination of the areacovered by the equation of B5

Comparison by percentage deviation

No

Yes



0

100

200

300

400

500

600

700

800

900

Torq

ue (N

m)


950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Engine speed (rpm)

y = minus000360157x2+ 709718665x minus 2708842

y = minus000360157x2+ 71257666x minus 27845680

Figure 6 The plots of the quadratic prediction models

Torque1198615

= minus0003601571199092

+ 709718665119909 + 2708842

(2)

Torque11986120

= minus0003601571199092

+ 71267666119909 + 2784568 (3)

332 Accuracy Check Each level of speed was placed intothe equations and the predicted values were calculated Thedifferences between the measured and predicted values arethe errors which can identify the accuracies of the generatedquadratic models Literature [12 13] suggests that meanabsolute percentage error (MAPE) is generally used becauseit is a unit-free and is an easily understandable method TheMAPE is defined in (4) One has

MAPE = 1119899

119899

sum119894=1

1003816100381610038161003816measured value minus predicted value1003816100381610038161003816measured value

times 100

(4)

Each level of engine speed was placed into (2) and thepredicted torque of biodiesel B5 was calculated The resultsare shown in Table 4 The same process was also carried outwith (3) and the predicted torque of biodiesel B20 is shown inTable 5 Tables 4 and 5 also present the existing errors It wasfound that the mean absolute percentage errors were 195and 538 for (2) and (3) respectively An MAPE of lowerthan 10 confirmed that both quadratic models were veryaccurate (excellent prediction) [13ndash15]

Moreover Tables 4 and 5 show themean percentage error(MPE) of minus0083 and minus0664 respectively These valuesindicate that the predicted values were on average 0083and 0664 lower than the measured values As a resultthe holistic accuracies of the generated quadratic predictionmodels were 99917 and 99336 respectively

333 Determination of Area Covered by the Quadratic ModelAfter the quadratic models were validated they were used asrepresentatives of the engine torque for biodiesels B5 and B20operations These models can accurately predict numerousresults within the testing range Then the areas covered byboth quadratic prediction curves were determined using thedefinite integration method between 1000 and 1400 rpmTheresults were determined as follows

Area1198615= int1400

1000

(minus0003601571199092

+ 709718665119909

minus2708842) 119889119909

Area1198615= 22939986


Table 4 Accuracy check of the quadratic prediction model for biodiesel B5

Speed (rpm) Torque-B5 (Nm) ErrorMeasured value Predicted value (Nm) Percentage Absolute percentage

1000 78510 78677 167 021 0211100 73528 74016 488 066 0661200 64012 62152 minus1860 minus291 2911300 41510 43085 1575 379 3791400 17189 16814 minus375 minus218 218

Average or mean 0008 minus0083 195Interpretation 99917 accuracy Excellent prediction




Mean 0004 minus0664 5381Interpretation 99336 accuracy Excellent prediction

Area11986120= int1400

1000

(minus0003601571199092

+ 71267666119909

minus2784568) 119889119909

Area11986120= 21282783

(5)

334 The Comparison of Engine Torque The differencebetween biodiesel B5 and B20 was then determined bythe percentage deviation method This is shown in (6)The comparison results show that the engine fuelled withbiodiesel B20 produces 722 lower torque than that of theengine fuelled with biodiesel B5 Consider

PD = (21282783 minus 22939986)22939986

times 100

PD = minus722(6)

4 Discussions of the Results

The results from the three comparison techniques revealedthat biodiesel B20 produced lower torque than that ofbiodiesel B5 at approximately 754 914 and 722 asconcluded in Table 6 The results show very comparableresults between the conventional technique and the newtechnique 2 while the new technique 1 presents an obviouslydifferent result

In order to discuss which techniques can present thereasonable and reliable results two important technicalexplanations are addressed as follows Firstly biodiesel B5 hasan 883 higher heating value than that of biodiesel B20 [16]

Table 6 The results from the three comparison techniques

Technique Result Interpretations(1) Conventional technique minus754 754 lower than B5(2) New technique 1 minus914 914 lower than B5(3) New technique 2 minus722 722 lower than B5

Hence the output torque of biodiesel B5 should be higherthan that of biodiesel B20 All the comparison techniquespresent the differences in the range between 722 and 914which are relatively logical

Secondly literatures have reported that biodiesel B20 hasbetter lubricating ability and a higher cetane number whichleads tomore efficient combustion [17ndash19]These benefits canslightly compensate for the poor heating value of biodieselB20 Consequently the different output torques betweenbiodiesels B5 and B20 of less than 883 were reliable andexpected

In accordance with the explanation the new comparisontechnique 1 revealed a difference of 914 greater than 883and so it can be reported that the new comparison technique1 is not an appropriate comparison technique On the otherhand the conventional technique and the new technique 2reported the values of 754 and 722 respectively Thesetwo techniques report reasonable and logical results

The new technique 2 is based on statistical and math-ematical procedures The method can accurately predict asufficient number of experimental outputs which will lead tomore comprehensive flexible and accurate results But theweak point of the new comparison technique 2 regards thecomplication of its methodology


5 Conclusions and Recommendations

This research aimed to present the detail of three engineperformance comparison techniques which are (1) ldquothecomparison of the average values (conventional technique)rdquo(2) ldquothe average of the point-to-point comparisons (the newtechnique 1)rdquo and (3) ldquothe comparison by applying quadraticprediction model (the new technique 2)rdquo The experimentwas conducted on a six-cylinder marine engine fuelledwith biodiesels B5 and B20 All the comparison techniquesreported that biodiesel B20 produced lower torque than thatof biodiesel B5 However further analysis showed that thenew technique 1 was an inappropriate comparison techniquewhile the results from the conventional technique and thenew technique 2 were very similar and acceptable

The contribution of this research is that it provides a newtechnique for the comparison of engine performance whichcan be applied in various fieldsThemost outstanding benefitof the new technique 2 takes place when it is integrated intothe experimental design process Since the technique canaccurately predict the output results in-depth the number ofexperiments can be reduced In other words this decreasesthe time and cost usage

Further investigations on the accuracy of the new tech-nique 2 are also very crucial The relation of the MPE MPEand comparison result should be determined Furthermorethe maximum values of the MPE and MAPE those stillprovide an accurate and acceptable comparison result wouldbe specified The investigation of a number of studies invarious fields and statistical analysis is also needed

Conflict of Interests

The author declares that there is no conflict of interestsregarding the publication of this paper

References

[1] A M Ashraful H H Masjuki M A Kalam et al ldquoProductionand comparison of fuel properties engine performance andemission characteristics of biodiesel from various non-ediblevegetable oils a reviewrdquo Energy Conversion and Managementvol 80 pp 202ndash228 2014

[2] A ImranMVarmanHHMasjuki andMAKalam ldquoReviewon alcohol fumigation on diesel engine a viable alternativedual fuel technology for satisfactory engine performance andreduction of environment concerning emissionrdquoRenewable andSustainable Energy Reviews vol 26 pp 739ndash751 2013

[3] M El-Kassaby and M A Nemit-Allah ldquoStudying the effect ofcompression ratio on an engine fueled with waste oil producedbiodieseldiesel fuelrdquo Alexandria Engineering Journal vol 52no 1 pp 1ndash11 2013

[4] P McCarthy M G Rasul and S Moazzem ldquoAnalysis andcomparison of performance and emissions of an internalcombustion engine fuelled with petroleum diesel and differentbio-dieselsrdquo Fuel vol 90 no 6 pp 2147ndash2157 2011

[5] K Cheenkachorn C Poompipatpong and C G Ho ldquoPerfor-mance and emissions of a heavy-duty diesel engine fuelled withdiesel and LNG (liquid natural gas)rdquo Energy vol 53 pp 52ndash572013

[6] S B Han Y H Chang G H Choi Y J Chung C Poompipat-pong and S Koetniyom ldquoEffect of the intake valve timing andthe injection timing for a miller cycle enginerdquo Journal of EnergyEngineering vol 19 pp 32ndash38 2010

[7] C Poompipatpong and A Kengpol ldquoA group decision supportmethodology to weight diesel enginersquos operating parameters byusing analytical hierarchy process and Delphirdquo InternationalJournal of Industrial Engineering and Technology vol 5 pp 47ndash60 2013

[8] N Maheshwari C Balaji and A Ramesh ldquoA nonlinear regres-sion based multi-objective optimization of parameters basedon experimental data from an IC engine fueled with biodieselblendsrdquo Biomass amp Bioenergy vol 35 no 5 pp 2171ndash2183 2011

[9] J B Hirkude and A S Padalkar ldquoPerformance optimization ofCI engine fuelled with waste fried oil methyl ester-diesel blendusing response surface methodologyrdquo Fuel vol 119 pp 266ndash273 2014

[10] H T M Tran B Cheirsilp K Umsakul and T BourtoomldquoResponse surface optimisation for acetone-butanol-ethanolproduction from cassava starch by co-culture of Clostridiumbutylicum and Bacillus subtilisrdquo Maejo International Journal ofScience and Technology vol 5 no 3 pp 374ndash389 2011

[11] I Pardoe Applied Regression Modeling A Business ApproachJohn Wiley amp Sons Hoboken NJ USA 2006

[12] P Yenradee A Pinnoi and A Charoenthavornying ldquoDemandforecasting and production planning for highly seasonaldemand situations case study of a pressure container factoryrdquoScienceAsia vol 27 pp 271ndash278 2001

[13] A Poompipatpong A Kengpol and T Uthistham ldquoThe effectsof diesel-waste plastic oil blends on engine performance char-acteristicsrdquo KMUTNB International Journal of Applied Scienceand Technology vol 7 pp 37ndash45 2014

[14] N Udomsri A Kengpol K Ishii and Y Shimada ldquoThe designof a forecasting support models on demand of durian for exportmarkets by time series and ANNsrdquo Asian International Journalof Science and Technology in Production and ManufacturingEngineering vol 4 pp 49ndash65 2011

[15] Y F Chang C J Lin J M Chyan I M Chen and J EChang ldquoMultiple regression models for the lower heating valueof municipal solid waste in Taiwanrdquo Journal of EnvironmentalManagement vol 85 no 4 pp 891ndash899 2007

[16] C Poompipatpong N Krasaelom P Triwong W Puttaviteeand P Wongtharua ldquoAn experimental study of performancesand emissions in a small fishery boatrsquos engines fuelled withbiodiesel B 5 and B 20rdquo The Journal of Industrial Technologyvol 10 2014

[17] D Altiparmak A Keskin A Koca and M Guru ldquoAlternativefuel properties of tall oil fatty acid methyl ester-diesel fuelblendsrdquo Bioresource Technology vol 98 no 2 pp 241ndash246 2007

[18] M Lapuerta O Armas and J Rodrıguez-Fernandez ldquoEffect ofbiodiesel fuels on diesel engine emissionsrdquo Progress in Energyand Combustion Science vol 34 no 2 pp 198ndash223 2008

[19] C Carraretto A Macor A Mirandola A Stoppato and STonon ldquoBiodiesel as alternative fuel experimental analysis andenergetic evaluationsrdquo Energy vol 29 no 12ndash15 pp 2195ndash22112004

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttpwwwhindawicom

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of


Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Stochastic AnalysisInternational Journal of


(1) Diesel engine (8) Hot water reservoir(2) Intake manifold (9) Cooling tower(3) Exhaust manifold (10) Airflow meter(4) Engine dynamometer (11) Exhaust gas temperature sensor(5) Dynamometer controller (12) Engine speed sensor(6) Water supply tank (13) Inlet coolant temperature(7) Engine coolant supply tank (14) Outlet coolant temperature

14

7

89

6

5

12

10

11

13

14

Figure 1 Experimental setup

techniques are presented This will assist researchers inselecting the most appropriate comparison technique forreliable and accurate results in the future

2 Experimental Setup and Engine Testing

In order to accomplish the main objective of this researcha GARDNER 6-cylinder 10450 cc marine diesel engine wasinstalled on a Clayton water brake dynamometer which sup-plied the engine load The fuel consumption was measuredon a geometric basis using a digital scale and a stop watchA KIMOCP200 and a Debimo airflowmeasuring blade wereattached at the intake manifold for the airflow measurementThermocouples were connected at several positions in orderto collect the raw data A closed-loop water cooling systemwith a cooling towerwas utilizedThe engine temperaturewascontrolled by a coolant temperature controller The diagramof the experimental setup is shown in Figure 1

The experimentationwas conducted at full load conditionover the engine speed range between 1000 and 1400 rpmwith an interval of 100 rpm Thirty raw data were recordedfor every test condition and the average values were pre-sented The engine was first operated with biodiesel B5 forexperimentation After the test the engine was switched tobiodiesel B20 and run for a while in order to flush out thebiodiesel B5 from the fuel line and injection system Finallybiodiesel B20 was tested at exactly the same condition Thebiodiesels B5 and B20 in this research were supplied by thePetroleum Authority of Thailand Public Company Limited

Table 1 Fuel properties

Properties Regulation forbiodiesel B5

Biodiesel B20 fromPTT

Cetane number ge50 5734Specific gravity From 081 to 087 08408Viscosity at 40∘C From 18 to 41 3774Flash point (∘C) ge52 790Pour point (∘C) le10 9Water and sediment( vol) le005 lt0005

Heating value (Jg) 4123855 4523333

(PTT) The fuel properties conformed to the regulations ofthe Department of Energy Business Thailand Ministry ofEnergy The details of the fuel specifications are shown inTable 1

3 Comparison Techniques and the Results

Raw data were collected from the engine experiment andwere then analyzed and compared by using three differentcomparison techniques The results of every technique werereported in percentage units () because this method ismore reasonable and comparable than when the informationis presented in raw data Nevertheless there are severalmethods that are used to calculate percentages and whichprovide dissimilar results This paper presents the details ofthree methods including one conventional method and twonew comparison methods which are (1) ldquothe comparison ofthe average values (conventional technique)rdquo (2) ldquothe averageof the point-to-point comparisons (the new technique 1)rdquo and(3) ldquothe comparison by applying quadratic prediction model(the new technique 2)rdquo It is also important to note that theoutput torque was selected as the representative of the engineperformance throughout the paper

31 Methodology of Conventional Technique A Comparison ofthe Average Values This technique is the simplest and mostcommonly used comparison technique It consists of onlytwo main steps which are the averaging of the output dataand then the average data of biodiesel B20 is compared tothat of biodiesel B5 (reference value) by using the percentagedeviation method (PD) This method is called ldquoaveragingbefore comparingrdquo or ldquothe comparison of the average valuesrdquoThe detailed methodology is shown in Figure 2

Figure 3 shows that biodiesel B5 produces slightly higheroutput torque than that of biodiesel B20 All of the experi-mental data in Figure 5 are also summarized in Table 2 Theaverage output torques over the speed conditions of 1000to 1400 rpm were 54950Nm and 50806Nm for biodieselsB5 and B20 respectively The difference was basically deter-mined using percentage deviation method The comparisonresults using the conventional technique show that the enginefuelled with biodiesel B20 generates 754 lower torque thanthat of the engine fuelled with biodiesel B5


Start

End




7851073528

64012

41510

17189

7391168880

61073

35438

14730100

200

300

400

500

600

700

800

900

1000

950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Torq

ue (N

m)

Engine speed (rpm)






Start

End






1000 78510 739111100 73528 688801200 64012 610731300 41510 354381400 17189 14730Average 54950 50806

PD = (50806 minus 54950)54950




Torque = 1198861199092 + 119887119909 + 119888 (1)






Start

End


Accuracy check



No

Yes



0

100

200

300

400

500

600

700

800

900

Torq

ue (N

m)


950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Engine speed (rpm)

y = minus000360157x2+ 709718665x minus 2708842

y = minus000360157x2+ 71257666x minus 27845680


Torque1198615

= minus0003601571199092

+ 709718665119909 + 2708842

(2)

Torque11986120

= minus0003601571199092

+ 71267666119909 + 2784568 (3)


MAPE = 1119899

119899

sum119894=1


times 100

(4)




Area1198615= int1400

1000

(minus0003601571199092

+ 709718665119909

minus2708842) 119889119909

Area1198615= 22939986










Area11986120= int1400

1000

(minus0003601571199092

+ 71267666119909

minus2784568) 119889119909

Area11986120= 21282783

(5)


PD = (21282783 minus 22939986)22939986

times 100

PD = minus722(6)

















References



























Volume 2014




Journal of











Journal of


Function Spaces






Algebra










Start

End




7851073528

64012

41510

17189

7391168880

61073

35438

14730100

200

300

400

500

600

700

800

900

1000

950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Torq

ue (N

m)

Engine speed (rpm)






Start

End






1000 78510 739111100 73528 688801200 64012 610731300 41510 354381400 17189 14730Average 54950 50806

PD = (50806 minus 54950)54950




Torque = 1198861199092 + 119887119909 + 119888 (1)






Start

End


Accuracy check



No

Yes



0

100

200

300

400

500

600

700

800

900

Torq

ue (N

m)


950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Engine speed (rpm)

y = minus000360157x2+ 709718665x minus 2708842

y = minus000360157x2+ 71257666x minus 27845680


Torque1198615

= minus0003601571199092

+ 709718665119909 + 2708842

(2)

Torque11986120

= minus0003601571199092

+ 71267666119909 + 2784568 (3)


MAPE = 1119899

119899

sum119894=1


times 100

(4)




Area1198615= int1400

1000

(minus0003601571199092

+ 709718665119909

minus2708842) 119889119909

Area1198615= 22939986










Area11986120= int1400

1000

(minus0003601571199092

+ 71267666119909

minus2784568) 119889119909

Area11986120= 21282783

(5)


PD = (21282783 minus 22939986)22939986

times 100

PD = minus722(6)

















References



























Volume 2014




Journal of











Journal of


Function Spaces






Algebra














Start

End


Accuracy check



No

Yes



0

100

200

300

400

500

600

700

800

900

Torq

ue (N

m)


950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450

Engine speed (rpm)

y = minus000360157x2+ 709718665x minus 2708842

y = minus000360157x2+ 71257666x minus 27845680


Torque1198615

= minus0003601571199092

+ 709718665119909 + 2708842

(2)

Torque11986120

= minus0003601571199092

+ 71267666119909 + 2784568 (3)


MAPE = 1119899

119899

sum119894=1


times 100

(4)




Area1198615= int1400

1000

(minus0003601571199092

+ 709718665119909

minus2708842) 119889119909

Area1198615= 22939986










Area11986120= int1400

1000

(minus0003601571199092

+ 71267666119909

minus2784568) 119889119909

Area11986120= 21282783

(5)


PD = (21282783 minus 22939986)22939986

times 100

PD = minus722(6)

















References



























Volume 2014




Journal of











Journal of


Function Spaces






Algebra


















Area11986120= int1400

1000

(minus0003601571199092

+ 71267666119909

minus2784568) 119889119909

Area11986120= 21282783

(5)


PD = (21282783 minus 22939986)22939986

times 100

PD = minus722(6)

















References



























Volume 2014




Journal of











Journal of


Function Spaces






Algebra
















References



























Volume 2014




Journal of











Journal of


Function Spaces






Algebra
















Volume 2014




Journal of











Journal of


Function Spaces






Algebra









research article quadratic prediction models for the...

Documents