- Email: [email protected]
Evaluation of sound quality in a tractor driver cabin based on the effect of biodiesel fatty acid composition
Fuel 118 (2014) 194–201
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Evaluation of sound quality in a tractor driver cabin based on the effect of biodiesel fatty acid composition M.D. Redel-Macías a, R.D. Rodríguez-Cantalejo b, S. Pinzi c, A.J. Cubero-Atienza a, M.P. Dorado c,⇑ a
Dep. Rural Engineering, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain Dep. Informatics and Numerical Analysis, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain c Dep. Physical Chemistry and Applied Thermodynamics, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain b
h i g h l i g h t s Biodiesel chemical composition affects sound quality and engine comfort. Characterization of sound quality in a driver cabin of a tractor has been performed using a scale model. A strong correlation between engine power, loudness, cetane number and unsaturation degree has been found. Considering cetane number and viscosity, loudness shows the opposite effect to roughness. A good sound quality requires a compromise between biodiesel viscosity and cetane number.
a r t i c l e
i n f o
Article history: Received 25 August 2013 Received in revised form 29 September 2013 Accepted 29 October 2013 Available online 8 November 2013 Keywords: Acoustic pollution Roughness Cetane number Viscosity Loudness
a b s t r a c t Nowadays, one of the main concerns for vehicle customers is comfort. This fact, together with the constraints imposed by legal regulations on noise emissions and human exposure to noise, have led to consider noise and vibration among the most important design criteria for agricultural machinery cabins. In this context, experimental analysis procedures for rapid prototyping and prediction models for early design assessment are crucial. This work presents a study of sound quality inside a tractor cabin mock up when the engine is fueled with different biodiesel/diesel fuel blends. In this study, the influence of the chemical properties of biodiesel has been correlated to sound quality metrics, i.e. loudness and roughness, and thus, their influence on the comfort of the tractor driver cabin. It has been found that cetane number and unsaturation degree of biodiesel are strongly related to loudness, whereas viscosity and unsaturation degree are strong indicators of roughness. Finally, to predict loudness and roughness based on biodiesel chemical properties, surface response models have been developed. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Biodiesel has the opportunity to contribute to the gradual substitution of fossil fuels in order to achieve the 10% renewable energy target imposed by the EU [1]. An additional advantage of biodiesel utilization is its potential domestic application by farmers to operate their agricultural machinery. One problem associated to drivers of agricultural machinery is the high noise and vibration levels to which they are often exposed. Subsequently, drivers are suffering from fatigue, hearing damage and other health problems [2]. In this sense, the design of the tractor cabin, following noise and vibration comfort criteria, may be an important factor to take into account by manufacturers. ⇑ Corresponding author. Tel.: +34 957218332; fax: +34 957218417. E-mail address: [email protected] (M.P. Dorado). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.10.067
For these reasons, to provide the optimal cabin design, the assessment of conceptual design alternatives in an early design phase is required. In fact, in recent years, this practice has become extensively used, particularly in the automotive industry [3]. Some authors have found meaningful differences from the noise radiated by an engine depending on the fuel used. It seems to be due to the combustion process, that it is strongly dependent on the biodiesel chemical properties, which affect the fuel-burning rate and injection timing advance, among other parameters [4– 7]. When combustion takes place, pressure and mechanical forces act over the engine block, producing the vibration of the block wall. In addition, the pressure gradients produced during combustion also cause the resonant oscillation of the gas in the combustion chamber [8,9]. It depends mainly on the bowl geometry and the gas temperature, being the last one dependent on the chemical properties of the fuel. The effect of engine vibration causes the
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
cabin indoor noise by means of airborne excitation. Airborne excitation consists of sound that impinges on the exterior of the cabin and introduces cabin vibrations, which transmit sound to the interior. Therefore, it can be inferred that if fuel chemical properties, i.e. cetane number, cinematic viscosity, bulk modulus, iodine value or oxygen content among others, have influence on the noise radiated by the engine, thus they also may affect the noise perceived by the driver. Although few decades ago research was focused on automotive noise reduction using active noise control techniques, nowadays efforts are mainly made to improve the sound quality inside the vehicle. Regarding this subject, research is mainly concentrated on the study of the radiated engine noise [10–12]. Since auditory impressions are subjective, different metrics are used to describe them [13–15]. In this sense, Sato and Miura [16] found out that comfortability in trucks running on diesel fuel is strongly related to Zwicker loudness. Moreover, Lee et al. [17] proposed to improve the combustion noise of an engine fueled with diesel fuel, considering its direct relationship with emissions. Although, these procedures are frequently used to study sound quality in an engine when it is fueled with diesel fuel, research concerning the influence of the chemical properties of biodiesel on sound quality in a tractor driver cabin is insufficient. Moreover, due to the EU legislation related to the increasing use of renewable energy, more research is needed. For this reason, an experimental analysis for the evaluation of the airborne excitation in a tractor cabin model (based on sound quality metrics) for different biodiesel/diesel fuel blends at several engine operating condition has been developed. Moreover, the chemical properties of biodiesel blends have been correlated with the comfort inside the cabin using sound quality metrics. Finally, to find out the relationship between biodiesel chemical properties and the perception of noise in a driver cabin based on sound quality parameters, several simple response surface models have been proposed. 2. Experimental layout In a previous phase, the noise radiated by the engine fueled with biodiesel blends and diesel fuel at several engine operating conditions was recorded. Next, recordings were used to both simulate the noise radiated by the engine and evaluate the effect of the use of biodiesel blends compared to diesel fuel on the perceived sound quality inside the tractor cabin. For this reason, a tractor cabin scale model was built.
195
A Soundbook instrumentation series recorded the noise radiated by the engine, by means of a portable and multi-channel meter. The measuring device was a GRAS prepolarized free-field half-inch microphone. This system was equipped with the software Samurai v1.7 from SINUS Messtechnik GmbH (Germany). Recorded measurements were processed in Matlab R2008a, from MathWorks Inc (USA). The microphone was calibrated before each test using a B&K calibrator model 4321. Recordings were carried out following ISO 362-1:2007 and used to simulate the noise emitted by the scale model engine.
2.2. Description of biodiesel properties In a previous work [6], the influence of the degree of unsaturation over emissions has been tested and analyzed. In this work, only the most popular biodiesel blends, according to the international market, have been selected. In this sense, 20% and 50% vo./ vol. biodiesel/diesel fuel blends were produced and analyzed as commented in a previous work [6]. Some of the most important fuel properties are depicted in Table 1.
2.3. Tractor driver cabin scale model The geometry of the tractor driver cabin scale model has been kept as simple as possible and following similar works [2,20]. The cabin scale model consists of two parts, the driver cabin (DC) and the engine compartment (EC), which are connected through a flexible joint allowing noise generated in the EC to be transmitted to the DC. A sound source placed in the EC is used to simulate a primary excitation disturbance source. This source is connected to a computer where the measurement of the noise radiated by the engine is stored. DC comprises five flat, rectangle shaped panels, stuck together as shown in Fig. 1. Panels present different thicknesses and are made of different types of plexiglass material to simulate the diverse structural parts of a cabin (roof, floor, doors, windshield, etc.), each one showing different dynamic properties. Material properties are summarized in Table 2. EC consists of four wood panels, as described in Table 2. The cabin mock up is placed in a wrinkled rubber mat, placed inside a semianechoic test room, as shown in Fig. 2. One microphone is placed inside the cabin at 130 cm height, 71 cm distant to the front part of the cabin and 61 cm distant to the left side. The characteristics of the microphone, calibrator and the measuring equipment are detailed in Section 2.1.
2.1. Description of the engine set up 2.4. Noise characterization A four-stroke three-cylinder direct injection diesel engine from an agricultural tractor has been tested. The maximum torque provided by the engine was 162.8 Nm, achieved at 1300 rpm, while the maximum engine power was 44 kW at 2132 rpm (DIN 6270-A). The engine was coupled to a Froment XT200 electric dynamometer, being the maximum engine power 136 ± 1.44 kW of accuracy at 100% engine speed (provided by the National Institute of Agricultural Engineering, UK) [18]. For each fuel noise testing, different engine operating conditions following the 8-mode cycle, according to ISO 1878-4, were considered [18]. At the beginning, the engine was operated with straight diesel fuel, followed by the blends at 20% and 50% of methyl esters with diesel fuel. To check the influence of the fatty acid composition, biodiesel was made of olive pomace oil (OPME), palm oil (PME), sunflower oil (SME), coconut oil (CME) and linseed oil (LME). After each test, the engine fueling system was cleaned and the engine was stabilized to the new condition [19].
As previously mentioned, sound quality is related to the human appreciation of a certain audio stimulus. Psychoacoustics go beyond the mathematical interpretation of pressure signals, correlating acoustic stimuli with hearing sensation [20]. As a first step, the most appropriate set of metrics needs to be defined. In the present work, the procedure described in ISO 532B was followed [21]. According to this procedure, the specific loudness is presented in Sone/Bark units. It is important to notice that loudness and roughness are the most common metrics to describe engine noise [22,23], being specific and Zwicker loudness used to evaluate subjective perception of noise by the human ear. Zwicker loudness is defined as the integral of the specific loudness over Bark and is expressed in Sone. Roughness is related to high frequency modulation of the sound and its unit is the Asper. In this work, roughness has been calculated according to the model proposed by Aures [24].
196
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Table 1 Chemical and physical properties of the tested fuels and comparison with diesel fuel in terms of total loudness and roughness (%). Fuel
Higher calorific value, HCV (J/g)
C (%)
H (%)
O (%)
Cetane number, CN
Kinematic viscosity, l at 40 °C (mm2/s)
Change in mean total loudness compared to diesel fuel (%)
Change in mean total roughness compared to diesel fuel (%)
Low-sulfur diesel fuel OPME20 OPME50 SME20 SME50 LME20 LME50 CME20 CME50 PME20 PME50
42,700 42,204 41,460 42,147 41,317 42,078 41,144 41,804 40,459 42,154 41,336
86.50 84.59 81.72 84.65 81.87 84.72 82.05 83.97 80.18 84.48 81.45
13.50 13.23 12.83 13.17 12.67 13.09 12.47 13.24 12.85 13.26 12.90
0.00 2.18 5.45 2.18 5.45 2.19 5.48 2.79 6.97 2.26 5.64
53.9 55.5 57.8 53.6 53.2 51.8 48.8 55.8 58.6 57.5 63.0
2.467 3.014 3.834 2.913 3.582 2.788 3.270 2.704 3.059 3.102 4.054
– 1.494 1.818 0.061 1.207 1.003 0.091 1.677 1.895 1.786 1.982
– 2.890 8.825 3.5173 6.353 7.516 9.563 3.162 6.872 5.001 7.463
OPME20: 20% olive pomace oil methyl ester/80% diesel fuel; OPME50: 50% olive pomace oil methyl ester/50% diesel fuel; SME20: 20% sunflower oil methyl ester/80% diesel fuel; SME50: 50% sunflower oil methyl ester/50% diesel fuel; LME20: 20% linseed oil methyl ester/80% diesel fuel; LME50: 50% linseed oil methyl ester/50% diesel fuel; CME20: 20% coconut oil methyl ester/80% diesel fuel; CME50: 50% coconut oil methyl ester/50% diesel fuel; PME20: 20% palm oil methyl ester/80% diesel fuel; PME50: 50% palm oil methyl ester/50% diesel fuel.
Fig. 1. Description of the cabin mock up.
2.5. Multiple response surface models to simulate the relation between physical properties and sound quality metrics
Zwicker loudness ðSoneÞ ¼ 43:20861 þ 0:02504CN
To describe the relationship between the most significant fuel physical properties, Zwicker loudness and roughness measured inside the tractor cabin mock up, some multiple linear surface models have been developed. For each analyzed sound quality metric, two models have been proposed: one based on CN and other on viscosity, as both are among the main physical properties affecting biodiesel behavior. In the experimental design, a holdout crossvalidation procedure using different functions from Matlab R2008a was carried out. The whole dataset n was divided into 3n/4 for training and n/4 for generalization purposes. Each model was performed using 88 points measured under the engine 8-mode cycle for 10 biodiesel blends and diesel fuel. Moreover, 30 holdout processes with different random splits of the data (by means of iterative loop) provided 30 models.
Zwicker loudness ðSoneÞ ¼ 43:95457 þ 0:20398l
2.5.1. Prediction of Zwicker loudness and roughness In order to predict Zwicker loudness and roughness, four important properties have been considered, namely CN, l, engine power (N) and engine speed (n), the last two ones due to their influence when different engine operating conditions are considered. The response surface models are the following:
Results on specific loudness under high engine power values are depicted in Fig. 3(a–d), while those for medium and low engine power are shown in Fig. 3(e–h). In agreement with previous results, the most critical bands are located below 5 Bark or 570 Hz for high power values, extending to 7 Bark (840 Hz) for medium power values [4,5]. As a general trend and considering high engine
þ 0:00108nðrpmÞ 0:03034NðkWÞ
þ 0:00108nðrpmÞ 0:03023NðkWÞ
ð1Þ
ð2Þ
RoughnessðAsperÞ ¼ 0:35086 0:00122CN þ 0:00049nðrpmÞ 0:00187NðkWÞ
ð3Þ
Roughness ðAsperÞ ¼ 0:43248 0:00508l þ 0:00049nðrpmÞ 0:00189NðkWÞ
ð4Þ
where n, engine speed (rpm), N, engine power (kW) and CN, cetane number. Both Zwicker loudness models show a good correlation, with a value of R2 of 99.26% for CN and 99.27% for l. 3. Results and discussion
197
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201 Table 2 Material dimensions and properties. Component of the tractor DC-windshield DC-backside DC-left/right side DC-roof Firewall EC-front/back side EC-left/right side EC-roof
Surface dimensions (mm2) 1440 1500 1440 1500 1320 1500 1320 1440 598 1040 598 1040 1950 1040 598 1950
Young’s modulus, E (N/m2) 9
2.3 10 4.6 109 1.8 109 2.9 109 187.5 109 3.0 109 3.0 109 3.0 109
Fig. 2. Vibro-acoustic cabin to test the model.
power values, the higher the biodiesel cetane number, CN the higher the specific loudness; for medium power values, the opposite effect is observed. It seems mainly due to the reduction of the maximum engine pressure, resulting from the improved CN and the reduced ignition delay period of the biodiesel/diesel fuel blends. However, there are some exceptions to this rule at specific engine operating conditions, i.e. mode # 6 from the 8-modes cycle, where OPME, CME and PME biodiesel blends show a higher increment of specific loudness. In this case, it may be attributed to their high oxygen content that provides a more complete combustion. This causes a higher in-cylinder pressure due to better fuel–air mixing, at medium engine power values. At low power values, blends depict a similar behavior with regards to specific loudness. Similar results on noise radiation are in agreement with previous results [25]. Fig. 4(a–c) illustrates the trend of Zwicker loudness considering high, medium and low engine power values. In general terms, at high engine power, Zwicker loudness increases with the increase of CN, excepting for LME50 (CN of 48.8), which may be due to its high unsaturation degree. Moreover, the increase of Zwicker loudness indicates the influence of the higher
Density, q (kg/m3)
Thickness, e (m)
1138 1208 1217 1309 8000 755 755 755
3.0 103 2.3 103 6.0 103 2.3 103 15 103 20 103 20 103 20 103
CN of the blends on the advance of the fuel injection timing, thus increasing the maximum pressure rise rate, which is expected to occur at high engine power. In general terms, at medium and low engine power, Zwicker loudness decreases with the increase of CN. In the same way, as the increase of CN is responsible of the decreased ignition delay period and subsequently, the maximum pressure rise rate, it is also responsible of an increase of the Zwicker loudness. At medium engine power values, there are some exceptions corresponding to blends of coconut oil with diesel fuel, CME20 (CN of 55.8) and CME50 (CN of 58.6), which show the lowest unsaturation degree. Fig. 5(a–c) shows the evolution of the roughness against kinematic viscosity, l. This figure shows that roughness increases slightly with l at high and medium engine power. High viscosity leads to a poor atomization and mixing, the fuel air mixture burning slowly and releasing less heat [8]. The opposite tendency is noticed for low power values, where biodiesel blends with higher viscosity show the biggest decrease of roughness. In this sense, PME20 (l of 3.102 mm2/s) and CME20 (l of 2.704 mm2/s) show the lowest roughness at high and medium engine power. However, PME50 (l of 4.054 mm2/s) decreases roughness at low engine power, thus reducing vibrations and noise. Additional results of specific roughness for the tested biodiesel blends, sorted by their unsaturation degree at different engine operating conditions, are presented in Fig. 6(a–h). At high engine power, the frequencies causing higher vehicle inner noise are in the range from 15 to 17 critical band rates (Bark), that is 2700– 4000 Hz. It may be seen that higher unsaturation degree increases specific roughness. The higher the unsaturation degree of biodiesel the higher the ignition delay and the advance of the injection, consequently increasing specific roughness. At medium and low engine power, the most critical band rates (Bark) are the 5th (510 Hz) and the 7th (700 Hz), showing the same trend as the unsaturation degree. It should be stressed that although roughness increases with the increase of the unsaturation degree, for some engine operating conditions the roughness value is very low (below 0.10 Asper). Therefore, the real sound quality equivalence should be verified with a jury test, which will be performed in future works. Table 1 depicts the comparison of the tested fuels with diesel fuel in terms of total loudness and roughness (%). It may be inferred that, in general terms, the higher the CN the higher the reduction of mean total loudness compared to diesel fuel. The most significant reduction (1.982%) was achieved with PME50. On the other hand, the higher the viscosity the higher the increase of total roughness. In sum, a reduction of mean total loudness and roughness was achieved when biodiesel blends were used instead of diesel fuel.
3.1. Multiple response surface models to simulate the relation between physical properties and sound quality metrics For each model, the mean square error (MSE) and the standard deviation (SD) were recorded. The best models were selected based on the value of these parameters in the generalization set, provided
198
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Fig. 3. Spectrogram of specific loudness (Sone/Bark) for biodiesel blends at several engine operating conditions (a–h).
Fig. 4. Zwicker loudness level (Sone) for biodiesel blends with different CN, (a) at high engine power, (b) at medium engine power and (c) at low engine power.
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Fig. 5. Roughness level (Asper) for biodiesel blends with different l, (a) at high engine power, (b) at medium engine power and (c) at low engine power.
Fig. 6. Spectrogram of specific roughness level (Asper/Bark) for different biodiesel blends at several engine operating conditions.
199
200
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Table 3 Comparative performance of the models based on the value of MSE of the generalization set. Method
MSEG (mean ± SD)
Zwicker loudness based on CN Zwicker loudness based on l Roughness based on CN Roughness based on l
2.591 ± 0.0594 2.590 ± 0.0618 0.008 ± 0.0002 0.007 ± 0.0001
Where G is the generalization set.
Table 4 Summary of the modeling results. MSE
R2 (%)
Fig. 8. Box plot for roughness models.
Multiple response surface models
99.26 floudness(x) = 43.20861 + 0.02504 CN + 0.00108 n (rpm) 0.03034 N (kW) 2.48 99.27 floudness(x) = 43.95457 + 0.20398 l + 0.00108 n (rpm) 0.03023 N (kW) 0.0076 91.02 froughness(x) = 0.35086 0.00122 CN + 0.00049 n (rpm) 0.00187 N (kW) 0.0076 91.00 froughness(x) = 0.43248 0.00508 l + 0.00049 n (rpm) 0.00189 N (kW)
2.44
by Table 3. Training results have not been considered since they are upward-biased. Table 4 depicts the results of the multiple response surface models applied to Zwicker loudness and roughness. The evaluation of these models was based on the MSE, coefficient of determination (R2) and P-value in the generalization set. The P-value was lower than 0.0001 for all proposed models, meaning that there are statistically significant relationships between the variables, with a confidence level above 99.9%. Results of the different models against MSE are plotted in Figs. 7 and 8. Box plots depict algorithm results according to the smallest observation, lower quartile, median, upper quartile and largest observation. 3.1.1. Prediction of Zwicker loudness As can be observed, both Zwicker loudness models are analogous and depict similar coefficients for engine speed and power. This indicates a clear effect of the engine speed and power on the Zwicker loudness. In fact, at high speed, the Zwicker loudness increases, while at high power it decreases proportionally to CN and l. Although, the higher the CN the higher the Zwicker loudness value, the value of the coefficient of l indicates that it is more significant to loudness than CN. 3.1.2. Prediction of roughness Table 4 shows a value of R2 of 91.02% and 91.00% for response surface models of roughness. Considering the Zwicker loudness
models, it can be seem that the speed and power present similar coefficient values in the models. However, the coefficients of CN and l of the roughness models are lower than those depicted by the loudness models. This means that the effect of these properties is stronger considering loudness. The negative sign of CN and l coefficients in the models means that roughness decreases as these properties increase. This statement is in agreement with the above explanation: CN is a relative measure of the delay between injection and auto-ignition of the fuel, while l affects the atomization quality. High cetane number guarantees good start behavior and a smooth run of the engine. In contrast, fuels with low cetane number tend to increase the in-cylinder pressure due to incomplete combustion. On the other hand, fuels with high l tend to form larger droplets during injection, which can cause poor fuel atomization of the spray, increase the engine deposits, need more energy to pump the fuel, wear fuel pump elements and injectors and consequently lead to poor combustion [26,27]. The main cause of roughness is due to the medium vibration originated by the moving parts of the engine, i.e. crankshaft and camshaft. This vibration is transmitted via the structure-borne to the tractor cabin and it penalizes the sound quality and comfort inside the driver cabin. Therefore, it may be inferred that higher viscosity penalizes the comfort by means of roughness increase. 4. Conclusions This work describes an experimental procedure used in a scale model, which can be used for the characterization of the sound quality in the driver cabin of a tractor. In general terms, at high engine power, the higher the cetane number the higher the loudness, excepting for biodiesel/diesel fuel blends that show a high unsaturation degree. At medium and low engine power values, the increase of cetane number decreases loudness, with the exception of blends with the lowest unsaturation degree (coconut oil biodiesel). For roughness, a strong correlation with the unsaturation degree and viscosity has been found. Results provided by multiple response surface models show that the higher the cetane number and viscosity the higher the loudness and the lower the roughness. In sum, the use of biodiesel blends is recommended to improve sound quality inside the tractor cabin, making the interior more comfortable for the driver. To achieve a good sound quality, a compromise between viscosity and cetane number of biodiesel blends is recommended. Acknowledgments
Fig. 7. Box plot for Zwicker loudness models.
Authors are grateful to the financial support by the Spanish Ministry of Education and Science (ENE2010-15159) and the Andalusian Economy, Innovation and Science Council, Spain (TEP-4994).
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
References [1] Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the promotion of the use of energy from renewable sources and amending and subsequently repealing Directives 2001/77/EC and 2003/30/EC; 2009. [2] Desmet W, Pluymers B, Sas P. Vibro-acoustic analysis procedures for the evaluation of the sound insulation characteristics of agricultural machinery cabins. J Sound Vib 2003;266:407–41. [3] de Oliveira LPR, Varoto PS, Sas P, Desmet W. A state-space modeling approach for active structural acoustic control. Shock Vib 2009;16:607–21. [4] Redel-Macías MD, Pinzi S, Ruz MF, Cubero-Atienza AJ, Dorado MP. Biodiesel from saturated and monounsaturated fatty acid methyl esters and their influence over noise and air pollution. Fuel 2012;97:751–6. [5] Redel-Macias MD, Pinzi S, Leiva D, Cubero-Atienza AJ, Dorado MP. Air and noise pollution of a diesel engine fueled with olive pomace oil methyl ester and petrodiesel blends. Fuel 2012;95:615–21. [6] Redel-Macías MD, Pinzi S, Leiva-Candia DE, Cubero-Atienza AJ, Dorado MP. Influence of fatty acid unsaturation degree over exhaust and noise emissions through biodiesel combustion. Fuel 2013;109:248–55. [7] Ramos MJ, Fernandez CM, Casas A, Rodriguez L, Perez A. Influence of fatty acid composition of raw materials on biodiesel properties. Bioresour Technol 2009;100:261–8. [8] Caresana F. Impact of biodiesel bulk modulus on injection pressure and injection timing. The effect of residual pressure. Fuel 2011;90:477–85. [9] Gumus M, Sayin C, Canakci M. The impact of fuel injection pressure on the exhaust emissions of a direct injection diesel engine fueled with biodieseldiesel fuel blends. Fuel 2012;95:486–94. [10] Alt NW, Nehl J, Heuer S, Schlitzer MW. Prediction of combustion process induced vehicles interior noise. SAE paper 2003-01-1435; 2003. [11] Genuit K. The sound quality of vehicle interior noise: a challenge for the NVHengineers. Int J Veh Noise Vib 2004;1:11. [12] Nor MJM, Fouladi MH, Nahvi H, Ariffin AK. Index for vehicle acoustical comfort inside a passenger car. Appl Acoust 2008;69:343–53. [13] Papenfus T, Genuit K, Blaschke P, Kang KT, Jung I, Jin J. Method of NVH quality rating of diesel combustion noise using typical driving modes. SAE paper 2009-01-2078; 2009.
201
[14] Payri F, Broatch A, Margot X, Monelletta L. Sound quality assessment of diesel combustion noise using in-cylinder pressure components. Meas Sci Technol 2009:20. [15] Payri F, Torregrosa AJ, Broatch A, Monelletta L. Assessment of diesel combustion noise overall level in transient operation. Int J Automot Technol 2009;10:761–9. [16] Sato N, Miura Y. Study on exterior idling sound quality evaluation method for diesel engine trucks. SAE Technical Papers; 1999. [17] Lee HS, Jung IS, Yoo DK, Kang KT. The combustion noise optimization and improvement for a 3.0L V6 diesel vehicle 2006; 284–92. [18] Dorado MP, Ballesteros E, Arnal JM, López-Giménez FJ. Exhaust emissions from a diesel engine fueled with transesterified waste olive oil. Fuel 2003;82:1311–5. [19] Dorado MP, Ballesteros E, Arnal JM, Gomez J, López-Giménez FJ. Testing waste olive oil methyl ester as a fuel in a diesel engine. Energy Fuels 2003;17:1560–5. [20] de Oliveira LPR, Janssens K, Gajdatsy P, Van der Auweraer H, Varoto PS, Sas P, et al. Active sound quality control of engine induced cavity noise. Mech Syst Signal Process 2009;23:476–88. [21] Zwicker E, Fastl H, Widmann U, Kurakata K, Kuwano S, Namba S. Program for calculating loudness according to DIN 45631 (ISO 532B). J Acoust Soc Jpn 1991;12:3. [22] Brassow B, Clapper M. Powertrain sound quality development of the Ford GT. SAE paper 2005-01-2480; 2005. [23] de Oliveira LPR, da Silva MM, Sas P, Van Brussel H, Desmet W. Concurrent mechatronic design approach for active control of cavity noise. J Sound Vib 2008;314:507–25. [24] Aures W. A procedure for calculating auditory roughness. Acustica 1985;58:268–81. [25] Haik YHY, Selim MYE, Abdulrehman T. Combustion of algae oil methyl ester in an indirect injection diesel engine. Energy 2011;36:1827–35. [26] Ramírez-Verduzco LF, García-Flores BE, Rodríguez-Rodríguez JE, Del Rayo Jaramillo-Jacob A. Prediction of the density and viscosity in biodiesel blends at various temperatures. Fuel 2011;90:1751–61. [27] Ramírez-Verduzco LF, Rodríguez-Rodríguez JE, Jaramillo-Jacob ADR. Predicting cetane number, kinematic viscosity, density and higher heating value of biodiesel from its fatty acid methyl ester composition. Fuel 2012;91:102–11.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Evaluation of sound quality in a tractor driver cabin based on the effect of biodiesel fatty acid composition M.D. Redel-Macías a, R.D. Rodríguez-Cantalejo b, S. Pinzi c, A.J. Cubero-Atienza a, M.P. Dorado c,⇑ a
Dep. Rural Engineering, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain Dep. Informatics and Numerical Analysis, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain c Dep. Physical Chemistry and Applied Thermodynamics, Ed Leonardo da Vinci, Campus de Rabanales, Universidad de Córdoba, Campus de Excelencia Agroalimentario Internacional ceiA3, 14071 Cordoba, Spain b
h i g h l i g h t s Biodiesel chemical composition affects sound quality and engine comfort. Characterization of sound quality in a driver cabin of a tractor has been performed using a scale model. A strong correlation between engine power, loudness, cetane number and unsaturation degree has been found. Considering cetane number and viscosity, loudness shows the opposite effect to roughness. A good sound quality requires a compromise between biodiesel viscosity and cetane number.
a r t i c l e
i n f o
Article history: Received 25 August 2013 Received in revised form 29 September 2013 Accepted 29 October 2013 Available online 8 November 2013 Keywords: Acoustic pollution Roughness Cetane number Viscosity Loudness
a b s t r a c t Nowadays, one of the main concerns for vehicle customers is comfort. This fact, together with the constraints imposed by legal regulations on noise emissions and human exposure to noise, have led to consider noise and vibration among the most important design criteria for agricultural machinery cabins. In this context, experimental analysis procedures for rapid prototyping and prediction models for early design assessment are crucial. This work presents a study of sound quality inside a tractor cabin mock up when the engine is fueled with different biodiesel/diesel fuel blends. In this study, the influence of the chemical properties of biodiesel has been correlated to sound quality metrics, i.e. loudness and roughness, and thus, their influence on the comfort of the tractor driver cabin. It has been found that cetane number and unsaturation degree of biodiesel are strongly related to loudness, whereas viscosity and unsaturation degree are strong indicators of roughness. Finally, to predict loudness and roughness based on biodiesel chemical properties, surface response models have been developed. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Biodiesel has the opportunity to contribute to the gradual substitution of fossil fuels in order to achieve the 10% renewable energy target imposed by the EU [1]. An additional advantage of biodiesel utilization is its potential domestic application by farmers to operate their agricultural machinery. One problem associated to drivers of agricultural machinery is the high noise and vibration levels to which they are often exposed. Subsequently, drivers are suffering from fatigue, hearing damage and other health problems [2]. In this sense, the design of the tractor cabin, following noise and vibration comfort criteria, may be an important factor to take into account by manufacturers. ⇑ Corresponding author. Tel.: +34 957218332; fax: +34 957218417. E-mail address: [email protected] (M.P. Dorado). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.10.067
For these reasons, to provide the optimal cabin design, the assessment of conceptual design alternatives in an early design phase is required. In fact, in recent years, this practice has become extensively used, particularly in the automotive industry [3]. Some authors have found meaningful differences from the noise radiated by an engine depending on the fuel used. It seems to be due to the combustion process, that it is strongly dependent on the biodiesel chemical properties, which affect the fuel-burning rate and injection timing advance, among other parameters [4– 7]. When combustion takes place, pressure and mechanical forces act over the engine block, producing the vibration of the block wall. In addition, the pressure gradients produced during combustion also cause the resonant oscillation of the gas in the combustion chamber [8,9]. It depends mainly on the bowl geometry and the gas temperature, being the last one dependent on the chemical properties of the fuel. The effect of engine vibration causes the
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
cabin indoor noise by means of airborne excitation. Airborne excitation consists of sound that impinges on the exterior of the cabin and introduces cabin vibrations, which transmit sound to the interior. Therefore, it can be inferred that if fuel chemical properties, i.e. cetane number, cinematic viscosity, bulk modulus, iodine value or oxygen content among others, have influence on the noise radiated by the engine, thus they also may affect the noise perceived by the driver. Although few decades ago research was focused on automotive noise reduction using active noise control techniques, nowadays efforts are mainly made to improve the sound quality inside the vehicle. Regarding this subject, research is mainly concentrated on the study of the radiated engine noise [10–12]. Since auditory impressions are subjective, different metrics are used to describe them [13–15]. In this sense, Sato and Miura [16] found out that comfortability in trucks running on diesel fuel is strongly related to Zwicker loudness. Moreover, Lee et al. [17] proposed to improve the combustion noise of an engine fueled with diesel fuel, considering its direct relationship with emissions. Although, these procedures are frequently used to study sound quality in an engine when it is fueled with diesel fuel, research concerning the influence of the chemical properties of biodiesel on sound quality in a tractor driver cabin is insufficient. Moreover, due to the EU legislation related to the increasing use of renewable energy, more research is needed. For this reason, an experimental analysis for the evaluation of the airborne excitation in a tractor cabin model (based on sound quality metrics) for different biodiesel/diesel fuel blends at several engine operating condition has been developed. Moreover, the chemical properties of biodiesel blends have been correlated with the comfort inside the cabin using sound quality metrics. Finally, to find out the relationship between biodiesel chemical properties and the perception of noise in a driver cabin based on sound quality parameters, several simple response surface models have been proposed. 2. Experimental layout In a previous phase, the noise radiated by the engine fueled with biodiesel blends and diesel fuel at several engine operating conditions was recorded. Next, recordings were used to both simulate the noise radiated by the engine and evaluate the effect of the use of biodiesel blends compared to diesel fuel on the perceived sound quality inside the tractor cabin. For this reason, a tractor cabin scale model was built.
195
A Soundbook instrumentation series recorded the noise radiated by the engine, by means of a portable and multi-channel meter. The measuring device was a GRAS prepolarized free-field half-inch microphone. This system was equipped with the software Samurai v1.7 from SINUS Messtechnik GmbH (Germany). Recorded measurements were processed in Matlab R2008a, from MathWorks Inc (USA). The microphone was calibrated before each test using a B&K calibrator model 4321. Recordings were carried out following ISO 362-1:2007 and used to simulate the noise emitted by the scale model engine.
2.2. Description of biodiesel properties In a previous work [6], the influence of the degree of unsaturation over emissions has been tested and analyzed. In this work, only the most popular biodiesel blends, according to the international market, have been selected. In this sense, 20% and 50% vo./ vol. biodiesel/diesel fuel blends were produced and analyzed as commented in a previous work [6]. Some of the most important fuel properties are depicted in Table 1.
2.3. Tractor driver cabin scale model The geometry of the tractor driver cabin scale model has been kept as simple as possible and following similar works [2,20]. The cabin scale model consists of two parts, the driver cabin (DC) and the engine compartment (EC), which are connected through a flexible joint allowing noise generated in the EC to be transmitted to the DC. A sound source placed in the EC is used to simulate a primary excitation disturbance source. This source is connected to a computer where the measurement of the noise radiated by the engine is stored. DC comprises five flat, rectangle shaped panels, stuck together as shown in Fig. 1. Panels present different thicknesses and are made of different types of plexiglass material to simulate the diverse structural parts of a cabin (roof, floor, doors, windshield, etc.), each one showing different dynamic properties. Material properties are summarized in Table 2. EC consists of four wood panels, as described in Table 2. The cabin mock up is placed in a wrinkled rubber mat, placed inside a semianechoic test room, as shown in Fig. 2. One microphone is placed inside the cabin at 130 cm height, 71 cm distant to the front part of the cabin and 61 cm distant to the left side. The characteristics of the microphone, calibrator and the measuring equipment are detailed in Section 2.1.
2.1. Description of the engine set up 2.4. Noise characterization A four-stroke three-cylinder direct injection diesel engine from an agricultural tractor has been tested. The maximum torque provided by the engine was 162.8 Nm, achieved at 1300 rpm, while the maximum engine power was 44 kW at 2132 rpm (DIN 6270-A). The engine was coupled to a Froment XT200 electric dynamometer, being the maximum engine power 136 ± 1.44 kW of accuracy at 100% engine speed (provided by the National Institute of Agricultural Engineering, UK) [18]. For each fuel noise testing, different engine operating conditions following the 8-mode cycle, according to ISO 1878-4, were considered [18]. At the beginning, the engine was operated with straight diesel fuel, followed by the blends at 20% and 50% of methyl esters with diesel fuel. To check the influence of the fatty acid composition, biodiesel was made of olive pomace oil (OPME), palm oil (PME), sunflower oil (SME), coconut oil (CME) and linseed oil (LME). After each test, the engine fueling system was cleaned and the engine was stabilized to the new condition [19].
As previously mentioned, sound quality is related to the human appreciation of a certain audio stimulus. Psychoacoustics go beyond the mathematical interpretation of pressure signals, correlating acoustic stimuli with hearing sensation [20]. As a first step, the most appropriate set of metrics needs to be defined. In the present work, the procedure described in ISO 532B was followed [21]. According to this procedure, the specific loudness is presented in Sone/Bark units. It is important to notice that loudness and roughness are the most common metrics to describe engine noise [22,23], being specific and Zwicker loudness used to evaluate subjective perception of noise by the human ear. Zwicker loudness is defined as the integral of the specific loudness over Bark and is expressed in Sone. Roughness is related to high frequency modulation of the sound and its unit is the Asper. In this work, roughness has been calculated according to the model proposed by Aures [24].
196
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Table 1 Chemical and physical properties of the tested fuels and comparison with diesel fuel in terms of total loudness and roughness (%). Fuel
Higher calorific value, HCV (J/g)
C (%)
H (%)
O (%)
Cetane number, CN
Kinematic viscosity, l at 40 °C (mm2/s)
Change in mean total loudness compared to diesel fuel (%)
Change in mean total roughness compared to diesel fuel (%)
Low-sulfur diesel fuel OPME20 OPME50 SME20 SME50 LME20 LME50 CME20 CME50 PME20 PME50
42,700 42,204 41,460 42,147 41,317 42,078 41,144 41,804 40,459 42,154 41,336
86.50 84.59 81.72 84.65 81.87 84.72 82.05 83.97 80.18 84.48 81.45
13.50 13.23 12.83 13.17 12.67 13.09 12.47 13.24 12.85 13.26 12.90
0.00 2.18 5.45 2.18 5.45 2.19 5.48 2.79 6.97 2.26 5.64
53.9 55.5 57.8 53.6 53.2 51.8 48.8 55.8 58.6 57.5 63.0
2.467 3.014 3.834 2.913 3.582 2.788 3.270 2.704 3.059 3.102 4.054
– 1.494 1.818 0.061 1.207 1.003 0.091 1.677 1.895 1.786 1.982
– 2.890 8.825 3.5173 6.353 7.516 9.563 3.162 6.872 5.001 7.463
OPME20: 20% olive pomace oil methyl ester/80% diesel fuel; OPME50: 50% olive pomace oil methyl ester/50% diesel fuel; SME20: 20% sunflower oil methyl ester/80% diesel fuel; SME50: 50% sunflower oil methyl ester/50% diesel fuel; LME20: 20% linseed oil methyl ester/80% diesel fuel; LME50: 50% linseed oil methyl ester/50% diesel fuel; CME20: 20% coconut oil methyl ester/80% diesel fuel; CME50: 50% coconut oil methyl ester/50% diesel fuel; PME20: 20% palm oil methyl ester/80% diesel fuel; PME50: 50% palm oil methyl ester/50% diesel fuel.
Fig. 1. Description of the cabin mock up.
2.5. Multiple response surface models to simulate the relation between physical properties and sound quality metrics
Zwicker loudness ðSoneÞ ¼ 43:20861 þ 0:02504CN
To describe the relationship between the most significant fuel physical properties, Zwicker loudness and roughness measured inside the tractor cabin mock up, some multiple linear surface models have been developed. For each analyzed sound quality metric, two models have been proposed: one based on CN and other on viscosity, as both are among the main physical properties affecting biodiesel behavior. In the experimental design, a holdout crossvalidation procedure using different functions from Matlab R2008a was carried out. The whole dataset n was divided into 3n/4 for training and n/4 for generalization purposes. Each model was performed using 88 points measured under the engine 8-mode cycle for 10 biodiesel blends and diesel fuel. Moreover, 30 holdout processes with different random splits of the data (by means of iterative loop) provided 30 models.
Zwicker loudness ðSoneÞ ¼ 43:95457 þ 0:20398l
2.5.1. Prediction of Zwicker loudness and roughness In order to predict Zwicker loudness and roughness, four important properties have been considered, namely CN, l, engine power (N) and engine speed (n), the last two ones due to their influence when different engine operating conditions are considered. The response surface models are the following:
Results on specific loudness under high engine power values are depicted in Fig. 3(a–d), while those for medium and low engine power are shown in Fig. 3(e–h). In agreement with previous results, the most critical bands are located below 5 Bark or 570 Hz for high power values, extending to 7 Bark (840 Hz) for medium power values [4,5]. As a general trend and considering high engine
þ 0:00108nðrpmÞ 0:03034NðkWÞ
þ 0:00108nðrpmÞ 0:03023NðkWÞ
ð1Þ
ð2Þ
RoughnessðAsperÞ ¼ 0:35086 0:00122CN þ 0:00049nðrpmÞ 0:00187NðkWÞ
ð3Þ
Roughness ðAsperÞ ¼ 0:43248 0:00508l þ 0:00049nðrpmÞ 0:00189NðkWÞ
ð4Þ
where n, engine speed (rpm), N, engine power (kW) and CN, cetane number. Both Zwicker loudness models show a good correlation, with a value of R2 of 99.26% for CN and 99.27% for l. 3. Results and discussion
197
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201 Table 2 Material dimensions and properties. Component of the tractor DC-windshield DC-backside DC-left/right side DC-roof Firewall EC-front/back side EC-left/right side EC-roof
Surface dimensions (mm2) 1440 1500 1440 1500 1320 1500 1320 1440 598 1040 598 1040 1950 1040 598 1950
Young’s modulus, E (N/m2) 9
2.3 10 4.6 109 1.8 109 2.9 109 187.5 109 3.0 109 3.0 109 3.0 109
Fig. 2. Vibro-acoustic cabin to test the model.
power values, the higher the biodiesel cetane number, CN the higher the specific loudness; for medium power values, the opposite effect is observed. It seems mainly due to the reduction of the maximum engine pressure, resulting from the improved CN and the reduced ignition delay period of the biodiesel/diesel fuel blends. However, there are some exceptions to this rule at specific engine operating conditions, i.e. mode # 6 from the 8-modes cycle, where OPME, CME and PME biodiesel blends show a higher increment of specific loudness. In this case, it may be attributed to their high oxygen content that provides a more complete combustion. This causes a higher in-cylinder pressure due to better fuel–air mixing, at medium engine power values. At low power values, blends depict a similar behavior with regards to specific loudness. Similar results on noise radiation are in agreement with previous results [25]. Fig. 4(a–c) illustrates the trend of Zwicker loudness considering high, medium and low engine power values. In general terms, at high engine power, Zwicker loudness increases with the increase of CN, excepting for LME50 (CN of 48.8), which may be due to its high unsaturation degree. Moreover, the increase of Zwicker loudness indicates the influence of the higher
Density, q (kg/m3)
Thickness, e (m)
1138 1208 1217 1309 8000 755 755 755
3.0 103 2.3 103 6.0 103 2.3 103 15 103 20 103 20 103 20 103
CN of the blends on the advance of the fuel injection timing, thus increasing the maximum pressure rise rate, which is expected to occur at high engine power. In general terms, at medium and low engine power, Zwicker loudness decreases with the increase of CN. In the same way, as the increase of CN is responsible of the decreased ignition delay period and subsequently, the maximum pressure rise rate, it is also responsible of an increase of the Zwicker loudness. At medium engine power values, there are some exceptions corresponding to blends of coconut oil with diesel fuel, CME20 (CN of 55.8) and CME50 (CN of 58.6), which show the lowest unsaturation degree. Fig. 5(a–c) shows the evolution of the roughness against kinematic viscosity, l. This figure shows that roughness increases slightly with l at high and medium engine power. High viscosity leads to a poor atomization and mixing, the fuel air mixture burning slowly and releasing less heat [8]. The opposite tendency is noticed for low power values, where biodiesel blends with higher viscosity show the biggest decrease of roughness. In this sense, PME20 (l of 3.102 mm2/s) and CME20 (l of 2.704 mm2/s) show the lowest roughness at high and medium engine power. However, PME50 (l of 4.054 mm2/s) decreases roughness at low engine power, thus reducing vibrations and noise. Additional results of specific roughness for the tested biodiesel blends, sorted by their unsaturation degree at different engine operating conditions, are presented in Fig. 6(a–h). At high engine power, the frequencies causing higher vehicle inner noise are in the range from 15 to 17 critical band rates (Bark), that is 2700– 4000 Hz. It may be seen that higher unsaturation degree increases specific roughness. The higher the unsaturation degree of biodiesel the higher the ignition delay and the advance of the injection, consequently increasing specific roughness. At medium and low engine power, the most critical band rates (Bark) are the 5th (510 Hz) and the 7th (700 Hz), showing the same trend as the unsaturation degree. It should be stressed that although roughness increases with the increase of the unsaturation degree, for some engine operating conditions the roughness value is very low (below 0.10 Asper). Therefore, the real sound quality equivalence should be verified with a jury test, which will be performed in future works. Table 1 depicts the comparison of the tested fuels with diesel fuel in terms of total loudness and roughness (%). It may be inferred that, in general terms, the higher the CN the higher the reduction of mean total loudness compared to diesel fuel. The most significant reduction (1.982%) was achieved with PME50. On the other hand, the higher the viscosity the higher the increase of total roughness. In sum, a reduction of mean total loudness and roughness was achieved when biodiesel blends were used instead of diesel fuel.
3.1. Multiple response surface models to simulate the relation between physical properties and sound quality metrics For each model, the mean square error (MSE) and the standard deviation (SD) were recorded. The best models were selected based on the value of these parameters in the generalization set, provided
198
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Fig. 3. Spectrogram of specific loudness (Sone/Bark) for biodiesel blends at several engine operating conditions (a–h).
Fig. 4. Zwicker loudness level (Sone) for biodiesel blends with different CN, (a) at high engine power, (b) at medium engine power and (c) at low engine power.
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Fig. 5. Roughness level (Asper) for biodiesel blends with different l, (a) at high engine power, (b) at medium engine power and (c) at low engine power.
Fig. 6. Spectrogram of specific roughness level (Asper/Bark) for different biodiesel blends at several engine operating conditions.
199
200
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
Table 3 Comparative performance of the models based on the value of MSE of the generalization set. Method
MSEG (mean ± SD)
Zwicker loudness based on CN Zwicker loudness based on l Roughness based on CN Roughness based on l
2.591 ± 0.0594 2.590 ± 0.0618 0.008 ± 0.0002 0.007 ± 0.0001
Where G is the generalization set.
Table 4 Summary of the modeling results. MSE
R2 (%)
Fig. 8. Box plot for roughness models.
Multiple response surface models
99.26 floudness(x) = 43.20861 + 0.02504 CN + 0.00108 n (rpm) 0.03034 N (kW) 2.48 99.27 floudness(x) = 43.95457 + 0.20398 l + 0.00108 n (rpm) 0.03023 N (kW) 0.0076 91.02 froughness(x) = 0.35086 0.00122 CN + 0.00049 n (rpm) 0.00187 N (kW) 0.0076 91.00 froughness(x) = 0.43248 0.00508 l + 0.00049 n (rpm) 0.00189 N (kW)
2.44
by Table 3. Training results have not been considered since they are upward-biased. Table 4 depicts the results of the multiple response surface models applied to Zwicker loudness and roughness. The evaluation of these models was based on the MSE, coefficient of determination (R2) and P-value in the generalization set. The P-value was lower than 0.0001 for all proposed models, meaning that there are statistically significant relationships between the variables, with a confidence level above 99.9%. Results of the different models against MSE are plotted in Figs. 7 and 8. Box plots depict algorithm results according to the smallest observation, lower quartile, median, upper quartile and largest observation. 3.1.1. Prediction of Zwicker loudness As can be observed, both Zwicker loudness models are analogous and depict similar coefficients for engine speed and power. This indicates a clear effect of the engine speed and power on the Zwicker loudness. In fact, at high speed, the Zwicker loudness increases, while at high power it decreases proportionally to CN and l. Although, the higher the CN the higher the Zwicker loudness value, the value of the coefficient of l indicates that it is more significant to loudness than CN. 3.1.2. Prediction of roughness Table 4 shows a value of R2 of 91.02% and 91.00% for response surface models of roughness. Considering the Zwicker loudness
models, it can be seem that the speed and power present similar coefficient values in the models. However, the coefficients of CN and l of the roughness models are lower than those depicted by the loudness models. This means that the effect of these properties is stronger considering loudness. The negative sign of CN and l coefficients in the models means that roughness decreases as these properties increase. This statement is in agreement with the above explanation: CN is a relative measure of the delay between injection and auto-ignition of the fuel, while l affects the atomization quality. High cetane number guarantees good start behavior and a smooth run of the engine. In contrast, fuels with low cetane number tend to increase the in-cylinder pressure due to incomplete combustion. On the other hand, fuels with high l tend to form larger droplets during injection, which can cause poor fuel atomization of the spray, increase the engine deposits, need more energy to pump the fuel, wear fuel pump elements and injectors and consequently lead to poor combustion [26,27]. The main cause of roughness is due to the medium vibration originated by the moving parts of the engine, i.e. crankshaft and camshaft. This vibration is transmitted via the structure-borne to the tractor cabin and it penalizes the sound quality and comfort inside the driver cabin. Therefore, it may be inferred that higher viscosity penalizes the comfort by means of roughness increase. 4. Conclusions This work describes an experimental procedure used in a scale model, which can be used for the characterization of the sound quality in the driver cabin of a tractor. In general terms, at high engine power, the higher the cetane number the higher the loudness, excepting for biodiesel/diesel fuel blends that show a high unsaturation degree. At medium and low engine power values, the increase of cetane number decreases loudness, with the exception of blends with the lowest unsaturation degree (coconut oil biodiesel). For roughness, a strong correlation with the unsaturation degree and viscosity has been found. Results provided by multiple response surface models show that the higher the cetane number and viscosity the higher the loudness and the lower the roughness. In sum, the use of biodiesel blends is recommended to improve sound quality inside the tractor cabin, making the interior more comfortable for the driver. To achieve a good sound quality, a compromise between viscosity and cetane number of biodiesel blends is recommended. Acknowledgments
Fig. 7. Box plot for Zwicker loudness models.
Authors are grateful to the financial support by the Spanish Ministry of Education and Science (ENE2010-15159) and the Andalusian Economy, Innovation and Science Council, Spain (TEP-4994).
M.D. Redel-Macías et al. / Fuel 118 (2014) 194–201
References [1] Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the promotion of the use of energy from renewable sources and amending and subsequently repealing Directives 2001/77/EC and 2003/30/EC; 2009. [2] Desmet W, Pluymers B, Sas P. Vibro-acoustic analysis procedures for the evaluation of the sound insulation characteristics of agricultural machinery cabins. J Sound Vib 2003;266:407–41. [3] de Oliveira LPR, Varoto PS, Sas P, Desmet W. A state-space modeling approach for active structural acoustic control. Shock Vib 2009;16:607–21. [4] Redel-Macías MD, Pinzi S, Ruz MF, Cubero-Atienza AJ, Dorado MP. Biodiesel from saturated and monounsaturated fatty acid methyl esters and their influence over noise and air pollution. Fuel 2012;97:751–6. [5] Redel-Macias MD, Pinzi S, Leiva D, Cubero-Atienza AJ, Dorado MP. Air and noise pollution of a diesel engine fueled with olive pomace oil methyl ester and petrodiesel blends. Fuel 2012;95:615–21. [6] Redel-Macías MD, Pinzi S, Leiva-Candia DE, Cubero-Atienza AJ, Dorado MP. Influence of fatty acid unsaturation degree over exhaust and noise emissions through biodiesel combustion. Fuel 2013;109:248–55. [7] Ramos MJ, Fernandez CM, Casas A, Rodriguez L, Perez A. Influence of fatty acid composition of raw materials on biodiesel properties. Bioresour Technol 2009;100:261–8. [8] Caresana F. Impact of biodiesel bulk modulus on injection pressure and injection timing. The effect of residual pressure. Fuel 2011;90:477–85. [9] Gumus M, Sayin C, Canakci M. The impact of fuel injection pressure on the exhaust emissions of a direct injection diesel engine fueled with biodieseldiesel fuel blends. Fuel 2012;95:486–94. [10] Alt NW, Nehl J, Heuer S, Schlitzer MW. Prediction of combustion process induced vehicles interior noise. SAE paper 2003-01-1435; 2003. [11] Genuit K. The sound quality of vehicle interior noise: a challenge for the NVHengineers. Int J Veh Noise Vib 2004;1:11. [12] Nor MJM, Fouladi MH, Nahvi H, Ariffin AK. Index for vehicle acoustical comfort inside a passenger car. Appl Acoust 2008;69:343–53. [13] Papenfus T, Genuit K, Blaschke P, Kang KT, Jung I, Jin J. Method of NVH quality rating of diesel combustion noise using typical driving modes. SAE paper 2009-01-2078; 2009.
201
[14] Payri F, Broatch A, Margot X, Monelletta L. Sound quality assessment of diesel combustion noise using in-cylinder pressure components. Meas Sci Technol 2009:20. [15] Payri F, Torregrosa AJ, Broatch A, Monelletta L. Assessment of diesel combustion noise overall level in transient operation. Int J Automot Technol 2009;10:761–9. [16] Sato N, Miura Y. Study on exterior idling sound quality evaluation method for diesel engine trucks. SAE Technical Papers; 1999. [17] Lee HS, Jung IS, Yoo DK, Kang KT. The combustion noise optimization and improvement for a 3.0L V6 diesel vehicle 2006; 284–92. [18] Dorado MP, Ballesteros E, Arnal JM, López-Giménez FJ. Exhaust emissions from a diesel engine fueled with transesterified waste olive oil. Fuel 2003;82:1311–5. [19] Dorado MP, Ballesteros E, Arnal JM, Gomez J, López-Giménez FJ. Testing waste olive oil methyl ester as a fuel in a diesel engine. Energy Fuels 2003;17:1560–5. [20] de Oliveira LPR, Janssens K, Gajdatsy P, Van der Auweraer H, Varoto PS, Sas P, et al. Active sound quality control of engine induced cavity noise. Mech Syst Signal Process 2009;23:476–88. [21] Zwicker E, Fastl H, Widmann U, Kurakata K, Kuwano S, Namba S. Program for calculating loudness according to DIN 45631 (ISO 532B). J Acoust Soc Jpn 1991;12:3. [22] Brassow B, Clapper M. Powertrain sound quality development of the Ford GT. SAE paper 2005-01-2480; 2005. [23] de Oliveira LPR, da Silva MM, Sas P, Van Brussel H, Desmet W. Concurrent mechatronic design approach for active control of cavity noise. J Sound Vib 2008;314:507–25. [24] Aures W. A procedure for calculating auditory roughness. Acustica 1985;58:268–81. [25] Haik YHY, Selim MYE, Abdulrehman T. Combustion of algae oil methyl ester in an indirect injection diesel engine. Energy 2011;36:1827–35. [26] Ramírez-Verduzco LF, García-Flores BE, Rodríguez-Rodríguez JE, Del Rayo Jaramillo-Jacob A. Prediction of the density and viscosity in biodiesel blends at various temperatures. Fuel 2011;90:1751–61. [27] Ramírez-Verduzco LF, Rodríguez-Rodríguez JE, Jaramillo-Jacob ADR. Predicting cetane number, kinematic viscosity, density and higher heating value of biodiesel from its fatty acid methyl ester composition. Fuel 2012;91:102–11.