Ignitibility and Combustibility of Bunker Fuel Oil -An Experimental Evaluation Using an Optical Combustion Analyzer

Transcription

1 P3: Ignitibility and ombustibility of unker Fuel Oil Ignitibility and ombustibility of unker Fuel Oil y iji Tomita *1, Yusuke Maeda *1, Takeshi Imahashi *1, Hiroshi Morinaka * and Yoshio Yamamoto * *1: Okayama University, TSUSHIM-NK, 3-1-1, KIT-KU, OKYM, 7-53 JPN -mail tomita@mech.okayama-u.ac.jp, Phone *: iwa-giken, o. Ltd., KNMOTO 77-, NK-KU,OKYM, 73-7 JPN -mail yamamoto@eiwagiken.com Phone Nowadays, low-speed two-stroke cycle diesel engines are used for marine transportation because of their high thermal efficiency, thus utilizing cheap fuel, of course good from the economical viewpoint. However, some marine diesel engines have been damaged when low grade bunker oil is used. Here, ignitability and combustibility of 17 samples of bunker fuel oil was evaluated. Visualization via a high-speed color video camera was conducted with signals of photo-sensors, rate of heat release and analysis of fuel properties. It was found that ignition delay and after-burning do differ, fuel to fuel. y changing the gas temperature at injection time, the degree of ignitability for the sample fuel could be clearly judged. Short or long after-burning was shown in spite of the same ignition delay. In some fuel oils, the combustion was promoted after ignition, even if the ignition delay was long. Some fuel oils, those with low sulfur content of less than 1.%, and a higher nitrogen content with a I value between and, presented both worse ignitability and worse combustibility. However, on the other hand, some fuel oils showed good combustibility, even in these properties. Therefore, this equipment will be a strong tool to judge the fuel quality of combustion before bunkering fuel oil. 1. Introduction Low-speed two-stroke cycle diesel engines are currently used for marine transportation because of their high thermal efficiency, their use of inexpensive fuel, and their economical operation. However, some marine diesel engines have been damaged by low-grade bunker fuel oil. Scuffing of the sliding parts of the piston rings and cylinder liners is a particular problem; however, the engine must continue to operate until the ship enters the port, and any schedule delays cause economic harm. Finding a solution to this problem is very important, but the cause of the scuffing is not clear. The design and manufacture of the engine, heat resistance and deterioration of the lubricating oil, operation and maintenance of the engine, and fuel ignitibility and combustibility are all interrelated in a complex manner. Of these factors, the ignitibility and combustibility characteristics of the fuel oil that causes damage are unclear and unpredictable. Fluid catalytic crackers that extract gasoline and other petrochemical products from heavy fuel oil are often used and make the quality of the residual oil worse. unker fuel oil is made of this residual oil combined with low-viscosity oil with a higher aromatic component. unker fuel oils have a wide range of combustion properties. t present, the easiest way to judge the quality of fuel is to investigate the actual combustion of the fuel in a vessel. Some research on the combustion of bunker fuel oil has taken place because the prediction of good or bad fuel based on general properties is difficult [1]. fuel ignition analyzer (FI) determines ignition delay and combustion characteristics from pressure history in a constant-volume vessel. When the ignition delay is long, a correlation is found between the ignition delay and engine damage. However, the accuracy of ignition delay and after-burning measurements is not high because of the instability of the injection system. Therefore, other systems are used to visualize the flame with optical analysis [ 5] and to analyze the rate of heat release with an excellent injection system []. In this study, the characteristics of ignition and combustion were examined using an optical combustion analyzer (O), which analyzes light emission from the flame and the rate of heat release through flame visualization. Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

2 P3: Ignitibility and ombustibility of unker Fuel Oil Number of sample Number of sample Sulfur, % I ensity@15deg., kg/m 3 Viscosity@5deg., mm /s Fig.1 haracteristics of bunker fuel oil tested in this study Table 1 xperimental conditions onstant-volume vessel φ1 [mm] 33 [mm] mbient temperature at injection [K] mbient pressure [MPa] Oxygen concentration [%] Injection pressure [MPa] Injection period [ms] Injection nozzle diameter [mm] Kinematic viscosity [mm /s] Injection system lectronic controlled system Fuel tank lectronically controlled spool valve ccumulator P, MPa Fuel pump Oil pump 5 3 ompressor Photo sensor T P xhaust Fuel gas P T Needle lift sensor Injector Mass flow controller O&N 容, 動 Spark ignition High-speed video camera Fig. Schematic diagram of experimental apparatus Fuel injection Spray combustion nalysis system. xperiment with Optical ombustion haracteristics.1 Fuel tested in this study In this study, 17 samples of bunker fuel oil used in marine engines were analyzed. In particular, many low sulfur fuels were chosen because they have the reputation of being bad fuel oils. Figure 1 shows the sulfur content, calculated carbon aromaticity index (I) [7], density, and kinematic viscosity for the samples t, s Fig.3 xample of vessel pressure history Start of injection nd of injection onstant pressure. Optical combustion analyzer experimental setup 3 Rapid Figure shows a schematic diagram of the experimental xcellent repeatability apparatus used in this study. Table 1 shows experimental conditions. The vessel was 1 mm in diameter and 33 mm in height. The high-pressure and -temperature atmosphere was created by the combustion of premixed gas rich in oxygen. thylene and oxygen/nitrogen mixtures were introduced into the vessel. The mixture was ignited with a spark, and the pressure and temperature of the gas 1 3 data are almost same in the vessel increased due to combustion. The burned gas included Fig. Histories of injection pressure near nozzle 1% oxygen, and the temperature was estimated from an equation for a exit for 3 experimental runs gas mixture of nitrogen, carbon dioxide, and water vapor. The pressure then decreased due to heat loss, and the sample fuel was injected at a certain pressure and autoignited. The diffusion flame was then observed. Figure 3 shows an example of the pressure history. The fuel was injected once under high pressure with an oil pump unit system and heated to a kinematic viscosity of 1 mm /s. The temperature at the injection port was controlled, and the temperatures at the head and that of the gas in the vessel were monitored. The injector had one hole with a nozzle diameter of. mm. The injection pressure was set to 3 MPa, and the injection period was 9 ms. Figure shows injection pressure histories for 3 experimental runs. The injection pressure resulted in an almost constant with excellent repeatability, so that Pressure, MPa decrease Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

3 P3: Ignitibility and ombustibility of unker Fuel Oil ideal diffusion flame was realized about 1 ms after the start of injection, and the end of the injection was very sharp. The injection pressure and pressure in the vessel were determined with pressure transducers, and the nozzle lift was monitored with a gap sensor. This signal was used with a computer to detect the end of the injection. The flame was visualized with a high-speed color camera at a frame rate of frames/s. The vessel had a quartz window mm long, and three photodiodes on the wall located at the opposite side of the window allowed for the detection of light emitted by the flame. The ambient temperature at the time of injection could be easily changed by changing the ambient pressure. The ignition and extinction of the flame are often based on the emission of light or the analysis of the rate of heat release. The ignition delay and after-burning period were defined in this study using the signal captured by photodiodes as shown in Fig.5. For example, the ignition delay was defined as the period from the start of injection to the fastest timing when the signal increased in the three photodiodes. The after-burning period was defined as the period from the end of injection to the time when the signal from the three sensors returned to zero. The standard experimental conditions were Ta 1 = K (P 1 = 1.9 MPa) and Ta = K (P = 1. MPa). The method of selecting these temperatures is described in the next section. The temperature was estimated from the pressure based on the equation of an ideal gas, so that this value denotes a representative (mean) temperature. The temperature error was estimated to be ±3.7 K in total because the errors are ±3. K and ±1. K at Ta = K when the mass entered the vessel (the accuracy of the mass flow controller was ±.5%) and the pressure value (the nonlinearity estimated from calibration was ±.5%) was considered []. 3. xperimental valuation of ombustion haracteristics 3.1 valuation obtained from visualization Five of the 17 samples were selected to examine the different combustion characteristics. Table shows the general properties and Fig. shows the time series of flame images obtained with the high-speed color camera for these 5 samples. The ambient temperature was K. The timing at ms on the left side of Fig. represents the start time of injection, whereas the timing at ms on Start of injection Injection period Photodiode signal, a.u. Photodiode signal, a.u. 1 Injection period fter-burning (a) Middle (b) Lower Fig. 5 xample of photodiodes and definition of ignition delay and after-burning period nd of injection istance from nozzle tip, mm Fig. Time series of flame images obtained from high-speed color camera for five samples at Ta = K Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

4 P3: Ignitibility and ombustibility of unker Fuel Oil the right side is the end of the injection. Table General properties of the five samples selected Fuel had a stable diffusion flame, and the ignition ensity Viscosity Sulfur H O arbon Hydrogen Nitrogen delay and after-burning period were both short. Fuel had a Fuel I kg/m 3 3 mm /s /s % % % % % 15deg. 5deg. short ignition delay with a longer after-burning period. fter < < the end of injection, a luminous flame remained for a long time < The locations of flame extinction were almost the same < < Therefore, even if the ignition delay was short, the after-burning period was different for the two fuels. On the other hand, both Fuels and had long ignition delays, and the ignition location was far from the nozzle exit where the fuel was well mixed with air. The flame of Fuel was also active and proceeded upward. However, the flame of Fuel was not active and did not expand like that of Fuel. Fuel had a long after-burning period. Furthermore, Fuel had a very long ignition delay with a dark flame and a very long after-burning period. For Fuel, the flame seemed to move upward during the extinction process. Photodiode signal, a.u. Rate of heat release, MJ/s Sample No. Sample No. Injection period Middle sensor Lower sensor Rate of heat release Sample No. Sample No. Sample No. Fig. 7 Photodiode signals (middle and lower) and rate of heat release for five samples 3. valuation judged from photo-sensors and rate of heat release Figure 7 shows the middle and lower photodiode signals and rate of heat release for each of Fuels at Ta = K. Fuel was ignited only once due to bad ignitibility. In Fuels and, the ignition delays from the start of injection to the rise time of the middle photosensor were short. The peaks of premixed (initial) combustion and the rate of heat release were almost the same. The signal of Fuel decreased quickly, whereas the signal of Fuel decreased gradually. This confirmed that Fuel had a longer after-burning period. The behavior of Fuels and with long ignition delay and ignition location far from the nozzle exit was confirmed with the signals obtained from lower photodetector. fter ignition, the signal of Fuel increased rapidly with normal combustion. However, the signal from the middle detector did not increase so rapidly for Fuel. The rate of heat release of Fuel indicated a large premixed combustion after ignition with stable diffusion combustion. One of the detector signals for Fuel showed weak intensity without peaky initial combustion. Figures and 9 show the rate of heat release and the probability of ignition delay, respectively, for 3 experimental runs with Fuels and. ven if the fuel was good, the ignition delay varied from one experimental run to another due to the statistical phenomena of ignition delay. However, the variation for Fuel was smaller than that for Fuel. When the fuel was good, stable diffusion combustion could be observed, as shown in Fig. (a). When the fuel was bad, the ignition delay varied greatly, and the first peak of premixed combustion could not be observed, as shown in Fig. (b). Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

5 P3: Ignitibility and ombustibility of unker Fuel Oil Injection period (a) Good ignitibility ignitability Sample : (b) Worse ignitibility ignitability Sample : ount of number, - 1 (a) Good ignitability ignitibility 1 Sample : Minimum : 3.17 ms Maximum :.19 ms verage :.53 ms Standard deviation : (b) Worse ignitability ignitibility 1 Sample : Minimum : 5. ms Maximum : 37.7 ms verage : 1.5 ms Standard deviation : , τ, ms Fig. 9 for 3 experimental runs for good and bad fuels Fig. Rate of heat release for 3 experimental runs for good and bad fuels, τ, ms Pressure, MPa Temperature, K /T, K -1 Fig. 1 Relationship between ignition delay and temperature (rrhenius plot) (Ta =K), τ, ms (Ta 1 =K), τ, ms (Ta =K), τ, ms Gradient of rrhenius plot, ms/k -1 Fig. 11 Relationship between ignition delay and the gradient value of rrhenius plot 3.3 valuation from two ambient temperatures The ambient temperature can be determined easily because the pressure at the injection timing can be changed. Figure 1 shows an rrhenius plot of the ignition delay versus ambient temperature for five samples, Fuels. The ignition delay τ is expressed as τ =e (1/T) where is a constant. large value of means the gradient of the rrhenius plot is large, and the ignition delay becomes large at a lower ambient temperature [9]. In Fuels,, and, the value of became large at lower temperatures. t a higher ambient temperature, the difference of ignition delay was too small to judge whether the ignitibility of the fuel was good. Therefore, two lower temperatures of Ta 1 = K and Ta = K were selected in this study. Here, when the ambient temperature was changed, the ambient pressure also changed because of the experimental method. When ambient pressure decreased, the ignition delay increased because of the lower oxygen molar concentration. In this study, when the ambient Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

6 P3: Ignitibility and ombustibility of unker Fuel Oil Ignition location, mm Ta 1 =K 1 1, τ, ms Location at flame extinction, xq, mm Time at flame extinction, tq, ms Ignition location, mm Ta =K 1 1, τ, ms Location at flame extinction, xq, mm Time at flame extinction, tq, ms Fig. 1 Relationship between ignition location and ignition delay, and relationship between location and time at flame extinction (Ta 1 = K) Fig. 13 Relationship between ignition location and ignition delay, and relationship between location and time at flame extinction (Ta = K) pressure decreased, the ambient temperature decreased, too. Therefore, the difference in ignition delay appears much larger. Figure 11 shows the relationship between the ignition delay at Ta = K and that at Ta 1 = K, and the relationship between the ignition delay at Ta = K and the gradient of the rrhenius plot. The fuel with the larger gradient of the rrhenius plot at a lower temperature is judged to be the worse fuel. These two temperatures were suitable for judging the ignitibility of fuel. Figures 1 and 13 show the relationship between the ignition delay and the ignition location and the relationship between the time and location at flame extinction at Ta 1 = K and Ta = K. When the ignition delay was long, the ignition location was far from the nozzle exit. The characteristics of ignitibility can be clearly seen. When the time at flame extinction was longer, the location at flame extinction is longer at Ta 1 = K. However, the time at flame extinction was very longer, the location at flame extinction is shorter at Ta 1 = K. 3. valuation from general properties Figure 1 shows the relationships between the ignition I delay/after-burning and sulfur, and the relationship between the ignition delay/after-burning and I for the 17 samples. For a sulfur content of.5% 1.% and a I in the range, some fuels had very long ignition delays and very long after-burning periods. However, many good fuels had these same sulfur contents and I. The I does not predict the ignitibility or combustibility for recent bunker fuel oils. y investigating several fuels, Takasaki et al. reported that the fuel with lower sulfur content and a nitrogen content greater than.% caused engine damage [1]. In this study, 3 samples with different ignition delays were selected from the 17 samples, and the ratios of nitrogen, carbon, and hydrogen in the fuel were investigated. Figure 15 shows the ignition delay on a sulfur nitrogen map at Ta = K. The closed squares, inverted triangles, triangles, circles, and closed circles denote the ignition delays averaged over 1, 1,,, and less than ms, respectively. Some fuels with lower sulfur and higher nitrogen content had very long ignition delays, although there were many, τ, ms fter burning, tq, ms Sulfur, % Fig. 1 Relationship between ignition delay/after-burning and sulfur, and relationship between ignition delay/after-burning and I value Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

7 P3: Ignitibility and ombustibility of unker Fuel Oil exceptions. The nitrogen content was determined from the properties of crude oil because the removal of nitrogen is difficult [11]. However, even if the crude oil had a high sulfur content, bunker fuel oil is produced through multiple processes involving desulfurization. Therefore, it is possible that the bunker fuel oil had bad ignitibility and combustibility due to the adjustment of viscosity. Figure 1 shows the ignition delay on the sulfur carbon/hydrogen ratio map at Ta = K. Fuel with a /H ratio larger than has bad ignitibility [1, 13]. If the fuels with ignition delay longer than ms have very bad ignitibility, the /H ratio is larger than, and the sulfur content is in the range.5% 1.%. However, some fuels with short ignition delay and good ignitibility had /H ratios larger than. Figure 17 shows the ignition delay on the map of carbon/hydrogen ratio and density at Ta = K. Figure 1 shows the ignition delay on the map of carbon/hydrogen ratio and kinematic viscosity at Ta = K. When the ignition delay was longer than ms, the density was higher. However, the kinematic viscosity was not an indicator of ignition delay. Some fuels with longer ignition delay had high density and low viscosity. This is likely due to the adjustment of viscosity and sulfur content. Therefore, I based on density and viscosity was not a predictor of modern bunker fuel oil ignitibility. The aromatic components of the fuel affect the ignition delay. ring analysis was performed to analyze the molecularstructure by determining the average molecular weight from density, kinematic viscosity, and sulfur content [1]. Figure 19 shows the ignition delay on a map of carbon/hydrogen ratio and average number of aromatic rings in one molecule at Ta = K. In general, fuel that includes fewer aromatic rings is considered to be easy to autoignite. The results in this Sulfur, % Over 1ms ms~1ms ms~ms ms~ms Under ms Nitrogen, % Fig. 15 on sulfur nitrogen map (Ta = K) Sulfur, % Over 1ms ms~1ms ms~ms ms~ms Under ms arbon/hydrogen ratio, /H, - Fig. 1 on sulfur carbon/hydrogen ratio map (Ta = K) arbon % / Hydrogen %, /H Over 1ms ms~1ms ms~ms ms~ms Under ms kg/m 3 Fig. 17 on map of carbon/hydrogen ratio and density (Ta = K) arbon % / Hydrogen %, /H Over 1ms ms~1ms ms~ms ms~ms Under ms.5 mm /s Fig. 1 on map of carbon/hydrogen ratio and viscosity (Ta = K) Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

8 P3: Ignitibility and ombustibility of unker Fuel Oil study support this assumption. However, some fuels had good ignitibility but a higher number of aromatic rings. For example, Fuels,,, and had almost the same average number of aromatic rings, but the ignition delay of Fuel was different than those of Fuels,, and. Figure shows the ignition delay on a ring analysis map, which indicates a triangle map of paraffinic, aromatic, and naphthenic carbons at Ta = K. The fuels with long ignition delay had many aromatic and naphthenic carbons. In particular, the fuel with the longest ignition delay was 5% aromatic carbon, 7% naphthenic carbon, and % paraffinic carbon. In general, the fuels with low sulfur, high I, high nitrogen, high /H ratio, and high aromatic carbon had long ignition delays. This might be useful for predicting the ignitibility of fuel. However, there were many exceptions, indicating that many other causes exist for bad ignitibility and combustibility. In short, it is difficult to predict ignitibility and combustibility by analyzing the general properties of the fuel or through ring analysis. O equipment that burns bunker Over 1ms fuel oil is much better for determining the ignitibility and ms~1ms combustibility. ms~ms (Ta =K), τ, ms verage aromatic rings, - Fig. 19 on a map of carbon/hydrogen ratio and average number aromatic rings (Ta = K) 5 Paraffinic carbon, % ms~ms Under ms 7 MO 3 romatic carbon, % 3 1 Naphthenic carbon, % 5 Fig. on ring analysis map (paraffinic, aromatic, and naphthenic carbons). Summary In this study, 17 samples of bunker fuel oil with various general properties were tested using an O. ven when the ignition delays were similar, the after-burning period was not always the same. Some fuels exhibited active combustibility after ignition and some did not, even when they had similar long ignition delays. The fuels with higher I values, lower sulfur content, higher nitrogen content, higher /H ratio, more aromatic rings, and higher aromatic carbon content are often considered to have bad ignitibility and combustibility. However, some fuels in this group had short ignition delays and after-burning periods. This result indicates that many other reasons exist for bad ignitibility and combustibility. The O is very useful for determining if a fuel is good or bad because much information is obtained empirically: flame visualization, ignition delay, after-burning period, and rate of heat release. cknowledgment The authors thank Mr. Satoshi Tanaka and his colleagues at NV Petroleum Service who supplied the bunker fuel oils tested in this study. References 1) Y. Nagai, M. Haneda, Journal of the JIM, 3-1, (1999),. ) K. Takasaki, et al., Proc. of IM ongress (Hamburg), (1), 9. Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)

9 P3: Ignitibility and ombustibility of unker Fuel Oil 3). Tomita, et al., Proc. of IM ongress (Kyoto), (), No.171. ). Tomita, et al., Proc. of ISM, (5), No ). Tomita, et al., Proc. of IM ongress (Wienna), (7), No.79. ). Strøm, et al., Proc. of ISM, (5), No.5-. 7) H. Nomura, Science of Marine Fuel, (199), 1, Seizando. ) JSM, ccuracy of Measurement (Translated version), (197) JSM. 9) Y. Hamamoto et al., Transactions of JS, 5-, (199), ) K. Takasaki, et al., Proc. of IM ongress (Kyoto), (), No.. 11) T. Hayashi, Journal of the JIM, 1-1, (), 3. 1) T. Hashimoto, H. Shiihara, Journal of the JIM, -, (5), ) T. Kurosawa, H. Shiihara, Journal of the JIM, -, (7), 3. 1) M. Ogawa, Ring nalysis of Petroleum, (), 3, Seizando. Translated from Journal of the JIM Vol.3,No.5 (Original Japanese)