Biodiesel Blends by

Subscriber access provided by BOSTON UNIV

Article

Fast determination of 2-ethylhexyl nitrate diesel/ biodiesel blends by distillation curves and chemometrics Matías Insausti, and Beatriz S Fernández Band Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.5b02995 • Publication Date (Web): 28 Jun 2016 Downloaded from http://pubs.acs.org on June 29, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

Fast determination of 2-ethylhexyl nitrate diesel/biodiesel blends by

2

distillation curves and chemometrics

3

Matías Insausti*, Beatriz Susana Fernández Band

4

Lab FIA, INQUISUR -CONICET, Departament of Chemistry, Universidad Nacional del Sur,

5

Bahía Blanca, Buenos Aires, Argentina (Av. Alem 1253, B8000CPB).

6

* Corresponding author. Tel.: +54 291 4595100; fax: +54 291 4595160.

7

E-mail addresses: [email protected].

8

9 10

Abstract A new method for the quantification of 2-EHN (2-ethylhexyl nitrate) was developed and validated.

11

In order to speed up and simplify the chemical analysis of diesel samples

12

throughout inspection procedures, we propose a single analytical method to

13

determine 2-ethylhexyl nitrate by using distillation curves routine assay used to

14

evaluate the quality of diesel (ASTM D86). This test was allied with multivariate

15

calibration based on PLS (partial least-squares) regression.

16

The results were comparable with reference methodology. By using

17

distillation curves of commercial diesel samples results into a prediction of the

18

cetane improver content with Relative Standard Deviations lower than 12% for

19

all fuel samples.

20

Since this correlation was established using commercial samples, the new

21

approach is immediately applicable in the petrochemical industry which needs

22

an adaptation to biodiesel/diesel blends.

23

1 ACS Paragon Plus Environment

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Keywords: Diesel/Biodiesel Blends; Distillation curves; Chemometrics; Cetane

2

improver; 2-ethylhexyl nitrate.

3

1. Introduction

Page 2 of 19

4

In order to improve the fuel properties of diesel, several kinds of chemicals

5

such as nitrates, ether nitrates or nitroso compounds are added. It has been

6

checked that these chemical additives produce an increasing effect on the

7

cetane number, which is associated to burning of fuel in the engine. Commonly,

8

the additive most used is the 2-ethylhexyl nitrate (EHN), due to the

9

improvement on the combustion characteristics, shortening ignition delay and

10

the start of combustion. The ignition delay period is counted from the injection

11

start to the sharp rise of in-cylinder pressure. Ignition delay period of diesel

12

engine is mainly influenced by physical–chemical characteristics of the fuel. A

13

fuel with a high cetane number has a short ignition delay and starts to burn

14

soon after it is injected in an engine [1]. The time to vaporize the fuel and mix

15

with the air content in cylinder, and the time to react through free radical

16

processes determine the ignition delay. Under normal conditions, if the ignition

17

delay is excessively lengthy in the diesel engine, outcome unburned fuel, low

18

power, and formation of particulates that increase engine noise and wear [2].

19

Reported mechanisms include the decomposition of cetane improvers into

20

free radicals and gas-phase catalysts such as NO2. This generated species are

21

involved in the fuel-air reaction. There are also reactions which inhibit free-

22

radical scavengers found in the fuel [3, 4].

23

The fuel from petroleum distillation is obtained by mixing several fractions

24

from the processing stages of crude oil. The ratio of these components in diesel

25

is made so as to frame the normative specifications which enable good


Page 3 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

performance of the product, control of toxic exhaust emissions, and minimize

2

wear of engines and components [5]. Distillation is a physicochemical assay

3

used to measure the volatility of the sample components of a complex liquid

4

mixture. The boiling range gives information on the composition, the properties,

5

and the behavior of the fuel during storage and use [6]. This assay is used to

6

verify the right proportions of the light and heavy fractions of fuel in order to

7

attain good performance. The assay results in a matrix with samples in rows

8

and the temperature reached by recovering 10, 20, 30, 40, 50, 60, 70 ,80, 85,

9

90, 95% of initial fuel volume in columns, allowing the use of multivariate

10

models.

11

The multivariate models associated to analytical techniques are an

12

advantageous alternative to predict physicochemical parameters, because they

13

are easy to apply, fast, low-cost and useful for online determinations. Recent

14

studies have shown the great potential of distillation curves, or a few specific

15

points, for the analysis of different parameters of petroleum products [7-12],

16

specific gravity [13], kinematic viscosity [14], octane numbers [15], cetane index

17

[16], flash point, ethanol and biodiesel content [17]. Bruno and coworkers

18

demonstrated that different additive concentrations in fuel produce modifications

19

in distillation curves [18].

20

The worldwide output of this nitrated additive is approximated to be about

21

100,000 tons per year, is a large-scale commodity. For a long time has been

22

regarded as not involving particular risks to human health. In spite of this, the

23

substance shows no evidence of biodegradability in water [19], is completely

24

miscible in fat and has potential for bioaccumulation and may form a film on

25

water affecting the oxygen transfer.


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 19

1

Decomposition products of the EHN form nitric oxides (NO and NO2), giving

2

as a result an additional source of NOx. Approximately the third part of the

3

nitrogen in the cetane improver goes to the exhaust as NOx in pure diesel

4

combustion [20]. In diesel/biodiesel blends, owing to the lower cetane number

5

(CN) of pure biodiesel (B100) [21], the addition of EHN is necessary to increase

6

the CN. By considering that the percentage of biodiesel in the blends was

7

increasing in recent years, the concentrations of EHN so did. Therefore, the

8

determination of EHN in blends is very important in order to control the

9

presence of this contaminant, which causes pollution of air and soils, dangerous

10

to life [22].

11

The American Standard Test Method D4046 indicates the methodology to

12

establish the alkyl nitrate quantity in fuel samples. The norm includes a

13

wearisome liquid-liquid extraction by using organic solvents, a derivatization

14

and the spectrophotometric measure. The procedure starts with a hydrolysis in

15

nitric acid solution, then the reaction with 2,4-dimethylphenol. The reaction

16

product is extracted by 2,2,4-trimethylpentane followed by addition of sodium

17

hydroxide. The measure is done in a spectrophotometer at 452nm. All these

18

steps results in several methodology disadvantages, such as organic solvent

19

consumption, large time expenditure and relatively high values of relative

20

standard deviations for repeatability and reproducibility. In the literature, there

21

are some alternatives by using headspace gas chromatographic/mass

22

spectroscopy assay [23], a chemiluminiscence detection of a derived product

23

[24] or using infrared spectrometry [25]. The developed GC/ MS and infrared

24

spectrometry methods, although it avoids the problems of the standardized


Page 5 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

method, requires costly instruments not available in every laboratory and

2

qualified operators.

3

In the daily work of quality control in the automotive and petrochemistry

4

industries request the development of new low cost and rapid methods for

5

determining the cetane improver. Thus, the aim of this work is to propose a

6

simple and rapid analytical method to predict the EHN concentration in blends

7

through a chemometric model and the data obtained from the routine distillation

8

curves

9 10

2. Experimental

11

Commercial diesel samples were collected within the year 2010 and 2015,

12

when the Argentinean fuel policy was changing, so the biodiesel content of the

13

120 diesel samples vary between 0 and 7% (v/v). In this country the biodiesel

14

added to blend the petroleum diesel fuel is produced from soybean oil.

15

The samples were distilled with automatic ISL AD 86 5G according to ASTM-

16

D86. Distillation curves (distillation temperature vs. the recovered volume) were

17

recorded as a daily routine in PETROBRAS laboratory (Bahía Blanca). This test

18

is performed in all fuel laboratories, providing very important information to

19

establish not only the quality of fuel but its price in the market. The distillation

20

curves data are also useful for determining cetane index (ASTM D4737).

21

In fuel laboratories, the EHN determination is not usually done as routine

22

analysis because the ASTM Standard Method has a lot of step works which

23

makes complicated, as said above.

24

Therefore, the EHN content in the 120 fuel samples were measured with the

25

Eraspec Diesel Fuel Analysis from Eralytics Company of PETROBRAS (Bahía


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 19

1

Blanca) [26]. This instrument is a NIR/MID-FTIR interferometer which lets the

2

determination of different parameters showing repeatability and reproducibility

3

values comparable with ASTM Method.

4

All distillation curves and EHN values were acquired in the same day that the

5

samples leave the refinery to go to gas stations. This information was registered

6

in the daily fuel quality reports.

7 8

2.1. Construction of chemometric model

9

The multivariate analysis was carried out using Kennard-Stone algorithm

10

(KS) [27] for sample selection and Partial Least Squares (PLS) [28] for

11

calibration modeling.

12

From data of distillation curves and EHN contents obtained for the 120

13

samples, were separated in 3 sets 60 for calibration, 30 for validation and 30 for

14

prediction.

15

The calibration set could be selected randomly from the whole set of 120

16

samples. Nevertheless, this procedure not warrants the fairly election of the set

17

over the entire sampling range. To solve this problem, the KS algorithm ensure

18

the representative choice of the prediction set and guarantee the predictive

19

ability of the calculated model along the whole calibration range. The PLS

20

model proposes data decomposition in score and loadings (latent variables)

21

matrices, and model relationships between sets of observed data. In order to

22

choose the correct number of latent variables to model the distillation curve

23

data, so-called validation process, were used two strategies “Test set” and

24

“Cross-validation”. In the “Test set” process the calculation of the latent

25

variables and scores were done using the calibration set of samples, and the


Page 7 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

validation was done predicting the validation set of samples. In a “Cross-

2

validation” process the calibration and validation set were the same group of

3

samples. In this work was used a “Full Cross-Validation Leave-One-Out”, all

4

calibration-validation samples were predicted by a model calculated with the

5

remainder samples of the group.

6

Then, the 30 data of distillation curves selected as a prediction set were

7

introduced into the calculated PLS models. Thus, the EHN concentrations can

8

be obtained through the models. It worth nothing that the 30 samples of the

9

prediction set were not use in any step of the process to calculate the

10 11

chemometric models. The predicted values of EHN concentration were compared with those

12

obtained experimentally with the ERASPEC instrument.

13

3. Results and discussion

14

Figure 1 A shows the distillation curves for different EHN concentration

15

range. Three curves (an average curve of each range) were plotted in order to

16

show the significant displacement of the curves, due to the presence of EHN.

17

This effect can be best seen in Figure 1 B, which were plotted the mean

18

centered (subtracting the mean temperature value of each recovery point).

19

INSERT FIGURE 1

20

The fuel additive presence retards the vaporization of lighter compounds, in

21

the blend diesel fuel samples this components would be normally registered

22

early in the distillation curve. An example of this effect is concentration found of

23

diethyl carbonate at 10, 20 and 30% (v/v) in the work of Bruno et. al. [18].

24

Another example of measure an additive through a distillation curve comes from


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 19

1

the measurement of molar concentration of Tetraethyl Lead in commercial

2

aviation gasoline approaching the trace concentration level [29].

3

Figure 2 presents a full-cross-validation (leave-one-out) of the whole set of

4

120 samples, differentiating groups of samples used in calibration, validation

5

and prediction sets. Each sample was predicted using 4 latent variable of a

6

model calculated with the 119 remaining samples. In this plot of predicted vs.

7

reference values, is outlined as the KS algorithm separated the samples, and

8

the linear behavior of the 120 samples.

9

INSERT FIGURE 2

10

The PLS modeling correlates the distillation curve data with the EHN content,

11

giving the ability to predict additive concentration in unknown fuel samples

12

assaying only with the data obtained from distillation curve test.

13

Table 1 presents the analytical figures of merit of the two strategies to

14

validate the chemometric models, using a full cross-validation leave-one-out

15

(CV) of 90 samples (calibration and validation sets founded with KS algorithm)

16

and a test set validation (using calibration set of 60 samples and 30 samples of

17

the validation set chosen by KS too). Both ways of validation had comparable

18

results, it can be shown with the Relative Error of Prediction (%), for CV=

19

13.18% and for test set= 14.81%. However as the number of samples used in

20

calibration step was higher for CV the result were little better. This result may

21

indicate that larger set of samples for calibration process could improve even

22

better the prediction results. For all the 3 calculated models, four latent

23

variables were chosen. Also, Table 1 presents the prediction results for the

24

external set of 30 samples which had not been used in the mathematical

25

calculation of the chemometric models. This means that once the statistical


Page 9 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

relationship between distillation curve data and EHN content was obtained, the

2

chemometric technique can be used from the results of the distillation curves, in

3

order to calculate the concentration of the analyte.

4

INSERT TABLE 1

5

The limit of quantification (LOQ) was defined as the lowest amount of an

6

analyte in a sample that can be determined quantitatively with convenient

7

precision and accuracy. It was calculated as LOQ = 10 x S, S is the standard

8

deviation of twenty prediction values for samples with 0% (v/v). The LOQ

9

founded for our methods is 0.0298% (v/v). The detection limit, lower limit of

10

detection (LOD), is the lowest quantity of a substance that can be distinguished

11

from the absence of that substance. It was calculated as LOD = 3.28 x S. The

12

LOD founded for our methods is 0.0101% (v/v). The resultant LOD and LOQ

13

are comparable with those founded in the bibliography [23- 25].

14

The EJCR of the regression [30] of predicted versus nominal concentrations

15

in the prediction set was studied for the two calibration models. The

16

corresponding plots are shown in Figure 3. All confidence regions contain the

17

ideal point of unit slope and zero intercept (indicating accuracy) at 95% of

18

confidence level, and the elliptic sizes obtained were comparable, suggesting

19

that both chemometric methodologies shown similar predictive ability. INSERT FIGURE 3

20 21 22

4. Conclusions

23

By using the distillation curve data and multivariate calibration was

24

predicted efficiently the 2-ethylhexyl nitrate content, which is an important


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 19

1

additive to improve cetane number. The distillation curves were obtained

2

following the ASTM D86 specifications.

3

The proposed method reduces the time and costs of analysis since

4

distillation assays are within the scope of laboratory analysis. The advantage of

5

this quantify methodology is the use of data that is obligatorily acquire in daily

6

routine analysis in every refinery. This new method avoids the need to perform

7

another experimental analysis.

8

The ability to relate the changing composition with distillation curves is critical

9

since many quality parameters could be estimated exploiting the data recorded

10

in the ASTM D86 assay.

11

As the developed method is cheap and do not use any organic compound

12

can be proposed like an alternative to the established standard method of EHN

13

determination.

14

The refineries and petrochemical industries (where EHN has to be

15

determined) could benefit of taking advantage of the present method.

16

Commonly, adding biodiesel to diesel produces a decrease of the cetane

17

number, thus, the improver is added. The new determining method could be

18

helpful in adapting out of date laboratories to test biodiesel/diesel blends.

19

20

Acknowledgements

21

The authors acknowledge the support of CONICET (research funds).

22 23

References


Page 11 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

[1] Ruina L., Zhong E., Peiyong N., Yang Z., Mingdi L., Lilin L. Effects of cetane

2

number improvers on the performance of diesel engine fuelled with

3

methanol/biodiesel blend. Fuel 2014; 128:180-187.

4

[2] Burger J.L., Lovestead T.M., Gough R.V., Bruno T.J. Characterization of the

5

Effects of Cetane Number Improvers on Diesel Fuel Volatility by Use of the

6

Advanced Distillation Curve Method. Energy Fuels 2014; 28:2437–2445.

7

[3] Suppes G.J., Rui Y., Rome C;. Chen Z. Cetane-Improver analysis and

8

impact of activation Energy on the Relative Performance of 2-Ethylhexyl Nitrate

9

and Tetraethylene Glycol Dinitrate. Ind. Eng. Chem. Res. 1997;36:4397-4404.

10

[4] Clothier P., Aguda B., Moise A., Pritchard H. How do dieselfuel improvers

11

work?. Chem. Soc. Rev. 1993;22:101-108.

12

[5] Albahri T.A. Developing correlations for the properties of petroleum fuels and

13

their fractions. Fluid Phase Equilibr 2012; 315:113–25.

14

[6] ASTM D86:2010 Standard Test for Distillation of Petroleum Products at

15

Atmospheric Pressure. Washington, DC: American Society for Testing and

16

Materials.

17

[7] Oliveira F.S., Teixeira L.S.G., Araújo M.C.U., Korn M. Screening analysis to

18

detect adulterations in Brazilian gasoline samples using distillation curves. Fuel

19

2004;83:917–23.

20

[8] Bruno T.J. Improvements in the measurement of distillation curves. Part 1: A

21

composition-explicit approach. Ind Eng Chem Res 2006;45:4371–80.

22

[9] Bruno T.J., Smith B.L. Advanced distillation curve measurement with a

23

model predictive temperature controller. Int J Thermophys 2006;27:1419–34.

24

[10] Barbeira P.J.S., Aleme H.G., Costa L.M. Determination of gasoline origin

25

by distillation curves and multivariate analysis. Fuel 2008;87:3664–8.


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 19

1

[11] Burger J., Lovestead T.M., Windom B.C., Bruno T.J. Characterization of

2

renewable fuels and additives with the advanced distillation curve method. Abstr

3

Pap Am Chem Soc 2011;242:145.

4

[12] Maheshwari A.S., Chellani J.G. Correlations for pour point and cloud point

5

of middle and heavy distillates using density and distillation temperatures. Fuel

6

2012; 98:55–60.

7

[13] Barbeira P.J.S., Aleme H.G., Costa L.M. Determination of ethanol and

8

specific gravity in gasoline by distillation curves and multivariate analysis.

9

Talanta 2009;78:1422–8.

10

[14]

Barbeira P.J.S., Aleme H.G., Assunção R.A.,

11

Determination of specific gravity and kinematic viscosity of diesel using

12

distillation

13

2012;102:90–5.

14

[15] Barbeira P.J.S., Aleme H.G., Mendes G. Determination of octane numbers

15

in gasoline by distillation curves and partial least squares regression. Fuel

16

2012;97:131–6.

17

[16] Barbeira P.J.S., Aleme H.G. Determination of flash point and cetane index

18

in

19

2012;102:129–34.

20

[17] Barbeira P.J.S., Aleme H.G. Determination of biodiesel content in diesel

21

using distillation curves and multivariate calibration. Energy Fuel 2012;26:5769–

22

74.

23

[18] Bruno T.J., Ott L.S., Lovestead T.M., Huber M.L. The composition-explicit

24

distillation curve technique: Relating chemical analysis and physical properties

25

of complex fluids. J. Chromat. A 2010;1217:2703-2715.

diesel

curves

using

and

multivariate

distillation

curves

calibration.

and

Fuel

multivariate

Carvalho

Process

M.M.O.

Technol

calibration.

Fuel


Page 13 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1

[19] American Chemistry Council Petroleum Additives Panel (2006) High

2

production volume: challenge program for nitric acid, 2-ethylhexylester.

3

[20] Ickes A.M., Bohac S.V., Assanis D.N. Effect of 2-Ethylhexyl Nitrate Cetane

4

Improver on NOx

5

Combustion. Energy Fuels 2009;23:4943–4948.

6

[21] Nadai D., Simões J.B., Gatts C.E.N., Miranda P.C.M.L. Inference of the

7

biodiesel cetane number by multivariate techniques. Fuel 2013;105:325-330.

8

[22] Day D.A., Liu S, Russel L.M., Ziemann P.J. Organonitrate group

9

concentrations in submicron particles with high nitrate and organic fractions in

Emissions from Premixed Low-Temperature Diesel

10

coastal southern California. Atmospheric Environment 2010;44:1970-1979.

11

[23] Dvořák B., Bajerová P., Eisner A., Nykodýmová O., Ventura K.

12

Determination of 2-ethylhexyl nitrate in diesel fuel. J Sep Sci 2011;34:1664–

13

1668.

14

[24] Wang C.X., Firor R.. (2010) Analysis of Trace 2-Ethylhexyl Nitrate in Diesel

15

Using Chemiluminescence Detector. Agilent Technologies, Application Brief.

16

[25] Bajerová P., Bajer T., Adam M., Eisner A., Ventura K. Fast determination of

17

2-ethylhexyl nitrate in diesel oils by infrared spectrometry. Fuel 2014;117:911-

18

916.

19

[26] PAC [Internet], Advance Analytical Instrumentation for Lab and Process

20

Application, http://www.paclp.com/.

21

[27] Soares S.F.C., Gomes A.A., Galvão Filho A.R., Araujo M.C.U., Galvão

22

R.K.H. The successive projections Algorithm. TrAC 2013;42:84-98.

23

[28] Wold S., Sjöström M., Eriksson L. PLS-regression: a basic tool of

24

chemometrics. Chemometrics and Intell Lab Sys 2001;58:109–130.


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 19

1

[29] Lovestead T.M., Bruno T.J. Application of the Advanced Distillation Curve

2

Method to the Aviation Fuel Avgas 100LL. Energy Fuels 2009;23:2176–2183.

3

[30] Riu J., Rius F.X. Method comparison using regression with uncertainties in

4

both axes. TrAC 1997;16:211-216.

5


Page 15 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

1 2

Table 1. Analytical performance of PLS models to predict the different set of samples. All samples

Validation set

Prediction set

CV

CV

Test set

CV

Test set

120

90

30

30

30

Slope

0.9478

0.9489

0.8207

0.9654

0.9045

Offset

0.0090

0.0089

0.0256

0.0059

0.0212

Correlation

0.9730

0.9736

0.9720

0.9701

0.9717

R-square

0.9476

0.9490

0.9216

0.9404

0.9374

RMSE*

0.0186

0.0189

0.0200

0.0178

0.0200

SE**

0.0187

0.0190

0.0193

0.0183

0.0195

Bias

0.00046

Samples

-0.00005 0.00599 0.00026 0.00572

3

*RMSE: Root Mean Squared Error, [% EHN (v/v)].

4

**SE: Standard Error, [% EHN (v/v)].


Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Page 16 of 19

Figure captions

2

Figure 1. (●) 0.03-0.12% EHN (v/v), (▲)0.12-0.2% EHN (v/v), (■)0.21-0.3%

3

EHN (v/v). A) Distillation curves data. B) Behavior of distillation curves mean

4

centered.

5

Figure 2. Cross validation prediction versus reference value.

6

Figure 3. Elliptical joint regions (at 95% confidence level) for the slope and

7

intercept of the regression of Test set model (solid line) and CV model (dashed

8

line) results. Black cross marks the theoretical (intercept = 0, slope = 1) point.

9


Page 17 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 1. (■) 0.3-0.21% EHN (v/v), (▲) 0.2 -0.12% EHN (v/v), (●) 0.12-0.03 % EHN (v/v). A) Distillation curves data. B) Behavior of distillation curves mean centered.

ACS Paragon Plus Environment

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Cross validation prediction versus reference value. 259x168mm (96 x 96 DPI)


Page 18 of 19

Page 19 of 19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 3. Elliptical joint regions (at 95% confidence level) for the slope and intercept of the regression of Test set model (solid line) and CV model (dashed line) results. Black cross marks the theoretical (intercept = 0, slope = 1) point. 179x129mm (96 x 96 DPI)


Biodiesel Blends by

Recommend Documents