Numerical investigation of the application of Miller cycle and low-carbon fuels to increase diesel 1 engine efficiency and reduce emissions 2

- In this paper, validated simulations using Ricardo WAVE have been performed to investigate 9 the effect of the Miller cycle and low-carbon fuels on the performance (power, torque, BTE and BSFC) 10 and emissions of a diesel engine. The results show that the increased Miller cycle effect (larger deviation 11 of the advanced or retarded intake valve closing from the standard intake valve closing time) will 12 decrease NOx, CO and HC emissions, and slightly improve Brake Thermal Efficiency (BTE) and Brake 13 Specific Fuel Consumption (BSFC) with slight loss in engine performance and increase in soot 14 emissions. An engine running B0 (diesel with 0% Biodiesel in the blend) with a -18% Miller cycle effect 15 has a reduction in NOx of 9% and CO of 4.3% with a decrease of 1.6% in power at the rated engine 16 speed. Using low carbon fuels drastically reduces emissions with reduced BTE and increased BSFC. 17 When used in conjunction, the Miller cycle and low-carbon fuels have an improved effect on both 18 performance and emissions. The optimal results demonstrate that using B60 (60% Biodiesel in the blend) 19 and a -8% Miller effect contributes to a 1.5% improvement in power, 1.2% in BTE, 13.3% in NOx, 20 38.5% in CO, 8.9% in HC, and 33.0% in soot at a cost of 6.0% increase in BSFC. The results show that 21 it is an easy way to reduce NOx, CO, HC and soot emissions and increase the BTE of the engine by 22 combining Miller cycle and low-carbon fuels.

(NOx) and other emissions from diesel engines are significant, and often above safe levels in built-up 30 areas. In the UK, diesel consumption grew from 11 Million Tonnes of Oil Equivalent (MTOE) in 1990 31 to 27 MTOE in 2017, and 21% of the UK's greenhouse gas emissions were from road transport in 2017 32 [1]. Large vehicles such as lorries account for 76% of excess NOx emissions [2]. The issues arising from 33 NOx have led to strict legislation, with limits for NOx under EU standards Euro 5 and Euro 6 of 0.18 34 g/km and 0.08 g/km [3]. This shows a need to find adaptations for diesel engines to counter these 35 emissions problems [4]. 36 The Atkinson cycle was first developed in 1882 and was achieved by mechanical methods [5]. This 37 involved altering the effective compression stroke of the engine by either early intake (EIVC) or late 38 intake (LIVC) valve closing [6]. In both EIVC and LIVC, the compression stroke is effectively shorter 39 than the expansion stroke, which improves the Brake Thermal Efficiency (BTE) of the overall cycle [6]. 40 However, since some fuel and air escape before combustion, the power output and Brake Mean Effective 41 Pressure (BMEP) is decreased [7]. Ralph Miller adapted this cycle to include a turbocharger or 42 supercharger at the cylinder inlet [8,9]. This pushes air into the engine cylinder, leaving it at a higher 43 pressure [9]. This may reduce Brake Specific Fuel Consumption (BSFC) [10] and mitigate the power 44 loss often found with the Atkinson cycle [11]. Once the piston is beyond Bottom Dead Centre (BDC), experimentation, and found the fuel consumption was reduced by 5.6% when using the LIVC. Lin and 48 Hou [13] analysed an air-standard Miller cycle and found engine efficiency improved over the Otto cycle. 49 In addition to improving engine efficiency, the Miller cycle can be used to reduce NOx emissions [14 -50 16]. Other methods such as after treatment can reduce NOx in exhaust gases but are expensive [17]. The 51 shorter compression stroke of the Miller cycle means a lower compression ratio, with consequent 52 reduction in cylinder pressure and temperature. Lower temperature in cylinder will produce less NOx 53 since NOx forms rapidly at high temperatures especially above 1600 K [18]. This reduction in NOx 54 emissions has been shown experimentally [14, 19 -21]. Test results showed a decrease in NOx emissions 55 of 60% and in soot of 25% [21]. In another experiment, the torque, power, BTE and BSFC were improved 56 at high engine speeds and worsened at low speeds due to the charge loss and the drop in volumetric 57 efficiency, but NOx emissions reduced by 14% [20]. The difference between standard cycle and Miller 58 cycle is illustrated in Fig. 1  combustion temperature cannot quantitatively be related to a single parameter, but instead is due to 76 several coupled mechanisms which change under different conditions such as fuel or combustion 77 characteristics. Kegl [31] found that in a B100 mixture the higher injection pressure and oxygen content 78 reduced soot and CO emissions. Higher NOx emissions also arose from the advanced injection process 79 with earlier and prolonged high temperatures at combustion commencement. Others found that higher 80 proportion of biodiesels lower the exhaust temperature [24,32]. It was found that the exhaust gas 81 temperature was linked to the compression ratio. At low ratios of 18, the exhaust gas temperature for 82 biodiesel was higher than that of diesel, but as the ratio increases the exhaust gas temperature of biodiesel 83 is lower than that of diesel [32]. It is also theorised that the reduction in exhaust temperature is due to the 84 lower calorific value of biodiesel reducing the total released energy and therefore the peak temperature 85 [24, 26, 32]. 86 It is found that existing researches have not accounted for the effect of the combination of the Miller 87 cycle with low carbon fuels on the performance of diesel engine, particularly in maximising efficiency 88 and minimising NOx and other emissions. Consequently, this study will investigate the influence of 89 combining the Miller cycle and low carbon fuels on diesel engine performance and emissions. The 90 following objectives will be covered: 91 In the model the details of the flow is obtained by solving quasi-one-dimensional compressible flow 113 equations which govern the conservation of mass, momentum and energy [37,38]. A staggered mesh 114 system is used with the boundaries between volumes to solve the equations of momentum, mass and 115 energy for each volume. In explicit conservative form, the equations can be written as: 116 The general combustion equation solved in the model is: 118 A(uN2 +vO2 +wCO2 +xH2O)+ B(CcHhOoNn)→ 119 aCO+bCO2 +cH +dH2 +eH2O+ fN2 +gNO+hO+iO2 + jOH +kN (4) 120 Where, A is the mass fraction of 'Burned Air', B is the mass fraction of 'Burned Fuel'; and a, b, …, k 121 are the coefficients of the combustion products. The sum of A and B is the total mass fraction of 122 combustion products, and the ratio between them gives the relative quantities of the species. Normally 123 the entire fuel mass is burned with product mass fractions that sum to 1; the values of the coefficients a 124 to k are uniquely linked to the equilibrium equation [37,38]. 125 In terms of combustion, the Diesel Wiebe combustion model is used to obtain the rate of fuel mass 126 burned, which includes the premixed combustion, diffusion combustion and slow late combustion (tail 127 burning) in the engine cylinder [37]. The burned fuel mass fraction W based on the crank angle can be 128 calculated from the following equation. where , and are the mass fractions of the premixed, diffusion and tail combustion respectively, 134 whilst is the burn duration term determined by Equation (6).
and here refer to the crank angle and the crank angle at the start of combustion respectively, whilst 139 RPM and BRPM are the engine speed and reference speed. The mass fraction of the premixed combustion 140 can be either user-input or obtained from the ignition delay model. 141 For emissions, the NOx emissions are predicted using the Zeldovich mechanism (Equation (7) ~ (8)) 142 and the Prompt mechanism (Equation (10)  HC model assumes that the fuel trapped within the injector sac and hole volume is the major source of 160 unburned HC, and the emitted HC is proportional to the injector sac volume. The typical injector sac 161 volume is in a range of 0.3 to 1 mm 3 and about 0.2 of sac volume fuel is converted to HC [37]. 162 Soot model accounts for the soot formation and oxidation rates. The soot formation rate is based on 163 Khan-Hiroyasu-Belardini formulation, and the soot oxidation rate is based on the Nagle and Strickland-164 Constable model [37]. The soot formation rate is calculated by the equation: Where is the fuel vapor mass, and is a coefficient dependent on the fuel properties and 169 temperature. The soot oxidation rate is calculated by where is the net soot mass, is the carbon molecular weight, and is the reaction rate 174 dependent on temperature. The soot density and soot mean diameter are constants. 175 176

Model set up and procedure 177
A Volkswagen 1.9L variable-geometry turbocharger (VGT) TDI PD diesel engine is selected as the 178 test engine, which is used in a wide range of midsized cars. The engine specification is shown in Table  179 2. 180 181 The numerical model of the selected 4-cylinder engine is set up according to its specifications, as 184 shown in Fig. 2. The model is validated and then used to investigate the engine performance under the 185 proposed conditions, i.e. the designed combined Miller cycles and a number of different low-carbon 186 fuels.

187
The procedure of setting up the model is in the following steps: 188 1) Generate engine components and link them together to form a basic engine model in the WAVE; 189 2) Define geometry and boundary conditions of the engine, such as bore, stroke, intake temperature, 190 etc., and select fuels; 191 3) Validate the model using data from the engine manufacturer.  The range tested was between 1000 -4500 rpm, which is the normal operating range of the engine, 234 giving 8 data points for each case. The engine performance parameters such as Brake Power, Brake 235 Torque, Brake Thermal Efficiency (BTE), Brake Specific Fuel Consumption (BSFC) were obtained. The  The reason for the performance decrease in the LIVC Miller Cycles is that part of the intake air and 265 some injected fuel (between the starting of fuel injection to the point of intake valve closed) is pushed 266 out of the cylinder when the intake valve closes later, which is wasted and thus reduce power, torque, 267 BTE and increase BSFC. The HC emissions also decrease as the Miller Cycle effect is enhanced, but the extent of the decrease 275 is than that in NOx or CO, which indicate a weaker link between Miller Cycle effect and HC emissions. 276 The HC emissions decreased at all engine speeds, with a maximum decrease of 1.7% at 2000 rpm. effect is found on engine performance, apart from a small decrease in power and torque at 4000 rpm. At 292 the rated speed, there is almost no loss in torque or power, as shown in Table 3. In terms of BTE, there 293 is a slight loss at lower engine speeds with a crossover at 3000 rpm, above which the Miller cycle causes 294 an improvement in BTE, with a peak increase of over 2% at 4000 rpm for a -20% Miller effect. This in 295 turn causes a minor improvement in BSFC at higher engine speeds.   Fig. 7 and Table 4 show that as the Miller cycle effect increases, losses also increase with a maximum 305 power loss of 5.3% at rated speed for -25% Miller.

307
The effect of varying degrees of EIVC Miller Cycle with B0 fuel on emissions is shown in Fig. 8. 308 Compared with LIVC, the largest improvements with EIVC occur at higher engine speeds for NOx and 309 CO. The changes in NOx emissions range from -5.8% at 1000 rpm to -19.8% at 4500 rpm for -25% 310 Miller. Moreover, as the Miller cycle effect increases, the improvement in emissions becomes larger. 311 Between 0 and -5% there is an improvement under 2% in the maximum decrease of emissions, but it 312 grows to 4% between -20% and -25% 313 314 315 With EIVC, there is also an improvement in HC emissions, with a maximum decrease of 1.7% at 3500 321 rpm, the rated speed. There is roughly 0.25% improvement in emissions on average for each 5% Miller 322 increment with EIVC.

323
EIVC has a slight negative impact on the soot emissions, with the strongest impact at higher engine 324 speeds (3500 -4500 rpm), where the peak increase in emissions is 54% at 4000 rpm with a -25% 325 Miller effect. At lower engine speeds soot emissions are smaller. For example at -25% Miller effect, 326 the soot emissions are 3.7% below the baseline. 327 When finer adjustments are made to the Miller cycle effect, an optimal value is found for a pure diesel 328 engine with the maximised the emissions decrease while the minimised power and torque losses. -18% 329 Miller is found to be the optimal value, a strongly EIVC value, which gives a slight power loss at speeds 330 greater than 4000 rpm but has no noticeable impact at low engine speeds. At rated speed, there is a 1.6% 331 improvement in power and torque. 332 This optimal value gives a maximum improvement in BTE and BSFC of 2.2% at 4500 rpm, with a 333 slight decrease of no more than 1.7% at low engine speeds. At the rated speed there is a 1.5% 334 improvement in BTE and BSFC. 335 At this optimal value, the NOx emissions reduced by from 0.5% to 10 Table 5 shows the results at the rated engine speed (3500 rpm), with no Miller effect present. The 341 results in Table 4 show a general decrease in torque, power, BTE, NOx, CO, HC and soot, and an increase 342 in BSFC as the proportion of biodiesel increases. 343 344  diesel-biodiesel blends 352 Fig. 9 shows engine performance using different low-carbon fuels (diesel-biodiesel blends). The torque 353 and power curves are relatively stable at mid-range engine speeds (2000-3000) rpm regardless of different 354 fuel blends. However, the percentage of biodiesel starts to influence engine performance at either low or 355 high engine speeds, since the torque and power losses increase with increasing biodiesel fraction. The 356 maximum loss of 14.7% in power and torque is found at 1500 rpm for B100 compared to B0. Moreover, 357 the difference between B0 and B20 is small, with a maximum decrease in power of 0.5% at 3500 rpm, 358 but between B80 and B100 the difference in power increases to 3.3%. 359 The decrease of both power or torque is the most visible in B100 with the reductions of 11.0% and 360 14.5% at 1000 and 1500 rpm respectively, and 13.1% and 9.5% at 4000 and 4500 rpm. This is mainly 361 due to the lower energy content of biodiesel relative to mineral diesel, which means the less heat is 362 released during the combustion of biodiesel. As a result, the engine produces less power. 363 There is a loss in BTE across almost all fuel blends and engine speeds, and this loss increases with 364 increasing biodiesel percentage and increasing engine speeds. The BTE decreases by 0.5 -1.0% for every 365 20% increase in biodiesel fraction, and reach average 2.5% for B100 across the whole range of engine 366 speed.

367
The BSFC is also drastically affected by the fuel blends. Increasing the biodiesel fraction greatly 368 increases BSFC, by 3% to 7% per 20% biodiesel increase. Unlike the power and torque curves, the 369 increase in BSFC is generally stable across all engine speeds, which is attributed to the lower energy 370 content (lower heating value) of biodiesel. There is an average 3% fuel consumption increase for each 371 20% increase in biodiesel, with a slight increase at the highest engine speeds. For B100, the BSFC is 372 average 15.5% higher than that of pure diesel, with a peak of 19.6% at 3500 rpm. 373 The major advantage of biodiesel is the impact on emissions, as shown in Figure 9, there is decreases 374 in all emission types as the proportion of biodiesel is increased. NOx emissions are most significantly 375 reduced in the lower engine speed range, whereas the reduced CO emissions occur to higher engine 376 speeds. HC and soot emissions are evenly decreased across all engine speed speeds. 377 NOx decreases as the proportion of biodiesel is increased, and the decrease is evenly distributed across 378 all engine speeds. For example, B20 has on average 3% less NOx emissions than B0, and B60 and B80 379 have nearly 5% decrease. The largest decrease is for B100, which emits an average of 22% less NOx 380 than B0. Table 4 shows this trend at the rated engine speeds. The lower NOx emission from the low-381 carbon fuels is due to the lower heating value of the fuel.

382
CO emissions are more significantly influenced by biodiesel fraction than NOx emissions. B20 shows 383 reduced CO emissions by average 13% and by 10% at rated speed compared with B0. As the biodiesel 384 fraction increases, CO emissions continue to drop, but the extent of reduction becomes smaller. For 385 instance, the decrease from B80 to B100 is 8%, whilst that from B60 to B40 is 13%.

386
B100 experiences an average decrease in CO emissions of 51.9% across all engine speeds, and the 387 decrease at rated speed of 51.7%.

388
HC emissions are more sensitive to varying biodiesel fraction than changing Miller Cycle, and 389 emissions decrease with increasing biodiesel fraction. HC emissions are reduced evenly across all engine 390 speeds and experience a 3% decrease for each 20% increase in biodiesel fraction. The largest reduction 391 occurs to B100, which has an average decrease in HC emissions of 14.1%, and a reduction at rated power 392 of 14.6%. 393 Increasing biodiesel proportion has a noteworthy impact on soot emissions. By increasing the 394 biodiesel proportion, soot emissions can be reduced significantly. The largest reduction is at lower 395 engine speeds, with a peak value of 89.6% for B100 at 1000 rpm. The decrease in soot is not evenly 396 distributed across all engine speeds. The decrease for B100 is 89.6% at 1000 rpm and 35.9% at 4500 397 rpm. The rate of reduction decreases as the proportion of biodiesel increases. Soot decreases by 20.8% 398 between B0 and B20 and by 4.5% between B80 and B100. The average decrease over all engine speeds 399 for B100 is 55.5% with 49.3% at rated speed. 400 The reason for the emissions of CO, HC and soot is mainly because the biodiesel contains oxygen 401 atoms in it, which helps the combustion or oxidation process of the fuel, enabling more fuel to be burnt.  Table 5. The 413 results of the simulations are shown in Table 5 and Fig. 11 and 12. 414 Table 6 shows the optimal results for each fuel tested, combined with the Miller cycle used, at the rated 415 engine speed (3500 rpm). 416 417   From Figure 10 and 11, the trends of engine performance and emissions of the combined Miller cycle 447 and different fuels can be identified: 448 1) At higher content of biodiesel where there is a power loss due to lower heating value of the fuel 449 (as shown in Table 6), but use of the Miller cycle partially compensates for the loss of engine 450 performance. For lower biodiesel content and for pure diesel the Miller cycle costs a small amount of 451 engine performance, but at high biodiesel content the Miller cycle improves engine performance to a 452 certain extent and reduces emissions. For example, without the Miller cycle, every fuel blend shows a 453 decrease in power and torque. As shown in Table 5, for B7, B20, B40, and B60, using the Miller cycle 454 improves both power and torque above the baseline. By using the Miller cycle, power and torque are also 455 improved at lower engine speeds which are more adversely affected by the use of low-carbon fuels. With 456 B100 and no Miller cycle, there is a drop of 11.0% and 14.7% of power at 1000 rpm and 1500 rpm 457 respectively. When the Miller cycle is used at -7%, these losses are only reduced to 4.0% and 9.8%. 458 2) BTE is improved by an average of 3% compared with that of biodiesel without using Miller effect.

459
For all fuel blends with biodiesel fraction up to B60, there is an overall increase in BTE compared to the 460 baseline. There is still an improvement in B80 and B100 compared to biodiesel with no Miller effect, but 461 the values are below the baseline engine. 462 3) When using Miller cycle, there are noticeable reduction in BSFC compared with biodiesel with no 463 Miller cycle. The reduction varies from 1.1% for B100, to 6.2% for B60, and an average reduction across 464 other fuel blends of 2-3%. 465 4) It is also found that, as biodiesel percentage increases, the Miller effect is further restricted due to 466 the power loss. For example, by B100, the ideal Miller cycle percentage is found to be -7% (EIVC Miller 467 Cycle), compared to -18% when B0 is used. 468 5) From Table 5 and Figure 10 and Figure 11, emissions are reduced with the combination of the Miller 469 cycle and low-carbon fuels, compared to that in Table 4, Figure 8 and Figure 9. With optimal Miller 470 Cycle values, there is a minimum 2% reduction in NOx emissions relative to biodiesel with no Miller 471 effect, and more than 5% reduction with certain fuel blends such as B40. 472 Overall, the combination of Miller cycle and low-carbon fuel leads to an enormous improvement 473 compared to the baseline. For B100 with no Miller effect, there is a 19.57% reduction in NOx and 51.66% 474 in CO relative to the baseline. With the optimal Miller cycle setup this is improved to 21.56% and 475 54.31%.

476
Similar results can be seen for the other low-carbon fuels, such as B20, where there are over 100% 477 decreases in NOx and CO emissions compared to B20 without the Miller Cycle at the rated speed. Even 478 for the higher biodiesel fraction where the effect is less noticeable, e.g. B80, there is a roughly 19% 479 decrease in NOx, and 8% decrease in CO emissions from the addition of the Miller Cycle, respectively. 480