tailored surrogate fuels for the simulation of diesel ... · different engines were investigated,...

15
THIESEL 2012 Conference on Thermo- and Fluid Dynamic Processes in Direct Injection Engines Tailored surrogate fuels for the simulation of diesel engine combus- tion of novel biofuels B. Kerschgens 1 , T. Lackmann 1 , H. Pitsch 1 , A. Janssen 2 , M. Jakob 2 and S. Pischinger 2 1 Institute for Combustion Technology (ITV) – RWTH Aachen University, 52064 Aachen, Germany. E-mail: [email protected] Telephone: +(49) 241 80 97592 Fax: +(49) 241 80 92923 2 Institute for Combustion Engines (VKA) - RWTH Aachen University, 52064 Aachen, Germany. E-mail: [email protected] Telephone: +(49) 241 80 95379 Fax: +(49) 241 80 92630 Abstract. The finite nature of fossil fuel supply as well as the impact of the combustion of fossil fuels on atmospheric CO 2 levels has led to increasing research efforts in the field of renewable fuels. In this context, the cluster of excellence “Tailor Made Fuels from Biomass” at RWTH Aachen University strives to develop and apply novel, third-generation, biomass-derived fuels. Recently, several promising fuels for diesel engines have been identified, produced, and test- ed. Diesel engine experiments at the Institute for Combustion Engines (VKA in the following) con- firmed very low soot and low NO X emissions. With regard to further improvements of the combustion system, it is desirable to complement the diesel engine experiments with numerical simulations. To date, this is hindered by the lack of suitable chemical reaction mechanisms for these novel fuels. Also, in the future, the development of chemical kinetic reaction mechanisms will not be able to keep up with the number of newly proposed bio-derived fuel components. Therefore, a surrogate approach is presented here and applied in CFD simulations. Combus- tion and pollutant formation is simulated using the representative interactive flamelet (RIF) model. By inclusion of detailed reaction chemistry, ignition, combustion, and pollutant formation are described in a consistent manner. Different mixtures of iso-octane, n-heptane, toluene, ethanol, dimethylether, and potentially other components are employed to describe the combustion chemistry of the biofuels. The compositions of the surrogate fuels are compiled according to H/C ratio, oxygen content, and cetane rating of the experimental fuels. Spray, injection, and evaporation properties of the experimental fuels, as obtained from spray vessel experiments, are included in the CFD simulations. By systematic comparison of experimental and numerical results, the surrogate methodology is validated and an improved understanding of the limitations of the current surrogate is achieved. Thus, a methodology for the fast adoption of novel fuels for simulations is proposed that can be used regardless of the availability of specific chemical re- action mechanisms. Notation CA Crank Angle CFD Computational Fluid Dynamics CN Cetane Number CO Carbon Monoxide CO 2 Carbon Dioxide DBE di-buthylether DME di-methylether EGR Exhaust Gas Recirculation IMEP Indicated Mean Effective Pressure ITV Institute for Combustion Technology KH/RT Kelvin-Helmholtz/Rayleigh-Taylor

Upload: others

Post on 19-Oct-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

THIESEL 2012 Conference on Thermo- and Fluid Dynamic Processes in Direct Injection Engines

Tailored surrogate fuels for the simulation of diesel engine combus-tion of novel biofuels B. Kerschgens1, T. Lackmann1, H. Pitsch1, A. Janssen2, M. Jakob2 and S. Pischinger2

1Institute for Combustion Technology (ITV) – RWTH Aachen University, 52064 Aachen, Germany. E-mail: [email protected] Telephone: +(49) 241 80 97592 Fax: +(49) 241 80 92923

2Institute for Combustion Engines (VKA) - RWTH Aachen University, 52064 Aachen, Germany. E-mail: [email protected] Telephone: +(49) 241 80 95379 Fax: +(49) 241 80 92630 Abstract. The finite nature of fossil fuel supply as well as the impact of the combustion of fossil fuels on atmospheric CO2 levels has led to increasing research efforts in the field of renewable fuels. In this context, the cluster of excellence “Tailor Made Fuels from Biomass” at RWTH Aachen University strives to develop and apply novel, third-generation, biomass-derived fuels.

Recently, several promising fuels for diesel engines have been identified, produced, and test-ed. Diesel engine experiments at the Institute for Combustion Engines (VKA in the following) con-firmed very low soot and low NOX emissions. With regard to further improvements of the combustion system, it is desirable to complement the diesel engine experiments with numerical simulations. To date, this is hindered by the lack of suitable chemical reaction mechanisms for these novel fuels. Also, in the future, the development of chemical kinetic reaction mechanisms will not be able to keep up with the number of newly proposed bio-derived fuel components.

Therefore, a surrogate approach is presented here and applied in CFD simulations. Combus-tion and pollutant formation is simulated using the representative interactive flamelet (RIF) model. By inclusion of detailed reaction chemistry, ignition, combustion, and pollutant formation are described in a consistent manner. Different mixtures of iso-octane, n-heptane, toluene, ethanol, dimethylether, and potentially other components are employed to describe the combustion chemistry of the biofuels. The compositions of the surrogate fuels are compiled according to H/C ratio, oxygen content, and cetane rating of the experimental fuels.

Spray, injection, and evaporation properties of the experimental fuels, as obtained from spray vessel experiments, are included in the CFD simulations. By systematic comparison of experimental and numerical results, the surrogate methodology is validated and an improved understanding of the limitations of the current surrogate is achieved. Thus, a methodology for the fast adoption of novel fuels for simulations is proposed that can be used regardless of the availability of specific chemical re-action mechanisms. Notation CA Crank Angle CFD Computational Fluid Dynamics CN Cetane Number CO Carbon Monoxide CO2 Carbon Dioxide DBE di-buthylether DME di-methylether EGR Exhaust Gas Recirculation IMEP Indicated Mean Effective Pressure ITV Institute for Combustion Technology KH/RT Kelvin-Helmholtz/Rayleigh-Taylor

Page 2: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

2 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

2-MTHF 2-methyltetrahydrofuran NOX Nitrogen Oxides NEDC New European Driving Cycle PDF Probability Density Function RCM Rapid Compression Machine RIF Representative Interactive Flamelet RWTH Rheinisch Westfälische Technische Hochschule Aachen SOI Start of Injection TDC Top Dead Center TMFB Tailor Made Fuels from Biomass VKA Institute for Combustion Engines

1. Introduction

The needs of future societies for continuous mobility necessitate the search for alternatives for fossil energy sources. One attractive option is the production and use of fuels from biomass, as they have the potential to considerably reduce CO2-emissions as well as pollutant emissions such as soot or nitrogen oxides. The holistic development of these fuels as a combination of biomass processing and combustion technology has proven to be a research area with a high need for interdisciplinary co-operation between natural and engineering sciences. In this context, the Cluster of Excellence "Tailor-Made Fuels from Biomass" (TMFB) at RWTH Aachen University was established in 2007 as part of the Excellence Initiative by the German Research Foundation in order to develop new, biomass-based, synthetic fuels for mobile applications.

Several novel biofuels for diesel engine combustion were identified and produced in the TMFB framework. Compared to standard diesel fuel, these fuels have an increased oxygen content and low-er cetane rating. Diesel engine experiments utilizing these fuels confirmed very low soot and NOX emissions [19 - 22]. As a drawback, increased emissions of unburned hydrocarbons and CO, espe-cially in the low load operating range, have been observed.

To further improve the combustion system and assess emissions formation mechanisms in more detail, it is desirable to complement the diesel engine experiments with numerical simulations. Here, such simulations are perfomed using the Representative Interactive Flamelet (RIF) model [4, 32]. In many studies concerning the modelling of compression ignition in internal combustion engines, the RIF model has been applied successfully. A general review of the model and its applications can be found in [3]. Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800 cm3/cylinder) engines (see also [2, 4, 12, 13, 36, 40]).

By the use of a detailed chemical reaction mechanism, the RIF model inherently accounts for low and high temperature auto-ignition, heat release, and pollutant formation. However, despite the fo-cused research on reaction chemistry within the TMFB group and elsewhere [5, 6, 14, 15, 38, 39], chemical reaction mechanisms of novel biofuels are not readily available. Even when such mecha-nisms are developed for the presently considered fuels, the future development of detailed chemical kinetics will not be able to keep up with the number of newly proposed bio-derived fuel components.

In the following, a surrogate approach is therefore proposed to overcome this problem, and to suggest a methodology that will be applicable to future, yet unknown fuels. This approach is exempli-fied for 2-methyltetrahydrofuran (2-MTHF in the following), a cellulose-derived fuel with a furanic struc-ture, and for a blend of 70% 2-MTHF and 30% di-buthylether (DBE in the following) by liquid volume.

This paper is arranged as follows: The first section shortly describes the combustion modelling approach employed in the current investigation. After this, the engine test bench, the investigated op-erating points, and the corresponding numerical setup, including the composition of the surrogate fuels and the chemical reaction mechanism, are presented. Spray vessel experiments at diesel engine like conditions are compared to simulations, confirming the correct representation of the 2-MTHF spray for the engine simulations. The results section contains an assessment of the proposed surrogate mech-anism compared to experimentally determined ignition delay times of 2-MTHF, and the comparison of the numerical simulations to the engine experiments, including pollutants. Finally, the conclusions and major findings from the study are summarized and an outlook on future work is given.

Page 3: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 3

2. Mathematical Model

2.1 CFD Code

The CFD code used in this work is AC-FluX (formerly known as GMTEC), a flow solver based on Finite Volume methods [9] that employs unstructured, mostly hexahedral meshes. AC-FluX solves the partial differential equations for continuity and momentum, an equation for the total enthalpy (in-cluding heats of formation), and two equations to account for the turbulence (k-epsilon-model). AC-FluX is documented by Khalighi et al. [24] and by Ewald et al. [8].

The liquid phase is modeled using the discrete droplet model (DDM), which describes the spray using a Monte-Carlo method. Since the spray consists of a large number of droplets, only the behavior of a representative subset of all droplets (called parcels) is calculated in detail. The gas phase and the liquid phase are coupled through source terms in the governing equations for the gas phase. A detailed description of the spray modeling in AC-FluX can be found in Spiekermann et al. [37].

For the physical properties of the fuel, toluene has been used, as the data base for 2-MTHF is quite incomplete. The most relevant properties of toluene like molecular weight, heat of vaporization, critical pressure, and critical temperature, are very similar to 2-MTHF. The justification of this assump-tion is proven by comparison of CFD simulations using toluene to spray vessel experiments of 2-MTHF below (section 3.2)

Concerning the 2-MTHF/DBE blend, a single-component vaporization model is used and the fuel composition is assumed uniform in the liquid phase, which is consistent with the typical modeling approach for other complex fuels. Although, for instance, diesel fuel contains hundreds of species with a wide range of boiling temperatures, no separation of these components is usually observed in evap-orating diesel sprays. In a recent study, Hottenbach et al. [17] optically investigated a diesel-like spray of an iso-octane/ethanol mixture, where no separation of these components could be found, which jus-tifies the assumption used here.

2.2 The laminar Flamelet concept

The laminar flamelet concept views a turbulent flame as an ensemble of flamelet structures at-tached to the instantaneous position of the flame surface, which itself is corrugated by the turbulent flow field. The assumption of combustion occurring in thin, locally one-dimensional layers was intro-duced by Williams [41]. Flamelet equations based on the mixture fraction as independent parameter and using a scalar dissipation rate to describe the turbulent mixing process were derived by Peters [30, 31]. Reviews on the flamelet concept can be found in [30 - 33].

An asymptotic analysis of the species transport equations yields the flamelet equations for the mass fraction of species j as

02 2

2

=−∂

∂−

ρωχ jjj

Z

Yt

Y &, (1)

where jω& denotes the chemical source term for species j , which is typically evaluated using detailed chemistry calculations. In the derivation of Eq. 1, the assumption of unity Lewis numbers and equal diffusivities for all species has been made.

2.3 RIF Model

Of fundamental importance for the RIF model is the separation of the treatment of physical and chemical processes, thereby decoupling the time scales and spatial discretization used. This al-lows for solving the chemistry, which is extremely nonlinear especially during ignition, with time steps much smaller than the CFD time steps. Additionally, the discretization of the flame structure in phase space is independent of the spatial discretization used in the CFD simulation and the flame structure is accessible for detailed analysis.

In Fig. 1, the schematic flow diagram shows the interaction of the computational fluid dynam-ics and the flamelet code. At every time step, the CFD code passes the scalar dissipation rate condi-

Page 4: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

4 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

tioned on stoichiometric mixture, stχ̂ and the mean pressure p̂ to the flamelet code. Based on this in-formation, a presumed shape for the dissipation rate, and the given initial and boundary conditions, the flamelet code solves Eq. 1 in addition to an energy equation in similar form with time steps that may be much smaller. The solution yields species profiles for every species contained in the chemical reaction mechanism. The species profiles as function of mixture fraction are convoluted with a pre-sumed PDF of mixture fraction, typically taken to be a β-PDF [12] scaled by the mean and the vari-ance of the mixture fraction in a particular cell, to yield the mean species mass fractions for each CFD cell. This yields the three-dimensional species distribution, heat release, and temperature. The scalar dissipation rate considers the effect of turbulence on the flame chemistry regarding mixing and flame stretch.

3. Experimental and Numerical Setup

3.1 Diesel engine test bench

The single-cylinder engine used for the diesel engine tests has a swept volume of 0.39 l and was designed for low emission levels while at the same time featuring high fuel efficiency. A compres-sion ratio of 15:1 was selected in order to keep the NOX emissions low in spite of the increased charge density, following typical EURO 6 development strategies. The combustion system reaches a specific output of 80 kW/l at maximum peak firing pressures of 220 bar. A common rail system with a maxi-mum fuel injection pressure of 2000 bar is used as injection system. To optimize the flow characteris-tics, one intake port was designed as a filling port, the second one as a classic swirl port. Creating charge movement was supported by seat swirl chamfers on both intake valves. The combustion chamber geometry was designed with a conventional recess shape, which was further optimized to-gether with the nozzle geometry (8-hole, ks = 1.5) in order to maximize air utilization. The low com-pression ratio of 15:1, early injection and high injection pressures as well as improved EGR cooling make very low particulate emissions possible, and as a result the research engine meets the Euro 6 standard. Table 1 shows a summary of the parameters of the test engine used. Additional information on the single-cylinder research engine can be found in earlier publications [1, 25, 26].

Four load points were selected, three of which are within the NEDC range for an inertia weight class of 1590 kg. The fourth load point is of interest for future downsizing concepts. Table 2 shows the respective calibration.

Fig. 1. Structure of the Representative Interactive Flamelet concept

Page 5: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 5

Table 1. Single-cylinder engine configuration

Unit Single-cylinder engine

Benchmark - Euro 6

Displacement cm³ 390

Stroke mm 88.3

Bore diameter mm 75

Compression ratio - 15

Valves per cylinder - 4

Maximum peak pressure bar 220

Fuel injection system specifications: - Bosch Piezo Common Rail System

Maximum injection pressure bar 2000

Hydraulic Flow Rate (HFR) cm³/30s 310 at 100 bar

Nozzle hole diameter µm 109

Number of spray holes - 8

Spray Cone Angle ° 153

Table 2. Engine Calibration

Load point Center of Combus-

tion [°CA after TDC]

Pressure Rail [bar]

Boost [bar, absolute]

Exhaust manifold [bar, absolute]

1. n = 1500 min-1, IMEP = 4.3 bar

6.6@ 0.5 g/kWh ISNOx

720 1.07 1.13

2. n = 1500 min-1, IMEP = 6.8 bar

5.8@ 0.5 g/kWh ISNOx

900 1.5 1.6

3. n = 2280 min-1, IMEP = 9.4 bar

9.2@ 0.5 g/kWh ISNOx

1400 2.29 2.39

4. n = 2400 min-1, IMEP = 14.8 bar

10.8@ 0.3 g/kWh ISNOx 1800 2.6 2.8

All fuels were analyzed with a single injection and at a constant center of combustion, which was chosen differently for the respective load points (see Tab. 2), whereby in each case the start of in-jection was adjusted accordingly. The tolerance for the center of combustion is +/- 0.1 °CA. The con-stant ISNOx level was obtained by adjusting the EGR rate accordingly. The other calibration parame-ters such as intake manifold pressure, fuel injection pressure, and charge air temperature had been optimized in earlier studies for a realistic 4-cylinder engine with a two-stage boosting device, all in compliance with the Euro 6 standard [27]. Some details on the engine settings resulting from the above described strategy are given in Table 3.

Page 6: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

6 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

Table 3. Engine setting details

Load

point

2-MTHF 70% 2-MTHF + 30% DBE

% EGR % inlet oxygen SOI [°CA after TDC] % EGR % inlet oxygen SOI [°CA after TDC]

1. No ignition obtained

32.4 18.2 -17.65 2. 49.5 14.3 -13.46 3. 47.4 16.4 -11 43.4 16.9 -5.31 4. 39.8 15.5 -6.6 36.8 16.4 -5.9

3.2 Spray Vessel

Shadowgraphic investigations of a 2-MTHF spray have been conducted in a high-pressure chamber, capable of pressures up to 150 bar and temperatures up to 1000 K. Details on the setup can be found in [18]. The chamber is equipped with the same injection system as the diesel engine de-scribed above, ensuring comparability of the results. In operation, a continuous air volume flow of 50mn

3/h is electrically heated before entering the measurement volume. Boundary conditions compa-rable to full- and part-load diesel engine operation conditions can be generated inside the measure-ment volume. The optical access to this volume for shadowgraphic measurements, which is a com-mon state-of-the art measurement technique, is achieved by two opposing quarz-glass windows mounted perpendicularly to the injector holder. For illumination, broadband light is focussed on a translucent glass generating a diffusive point light source. This light source is positioned in the focal point of a convex mirror. The parallelized light, which is reflected by the mirror, is transmitting through the measurement volume and deviated as a function of density gradients. Therefore, density gradients as well as the shadow of the solid spray core are visible in the detected image. Consequently, the an-gle, penetration, distribution, and temporal development of the liquid and gaseous spray phase can be investigated.

The 2-MTHF spray was investigated at a temperature of 800 K, back pressure of 50 bar, and rail pressure of 900 bar, corresponding to the engine conditions for load point 2. CFD calculations for these conditions have been performed and the parameters of the KH/RT breakup model were adjust-ed accordingly. The results for liquid and gaseous penetration are compared to the measurements in Fig. 2, showing good agreement and thus confirming the approximation of the physical spray proper-ties of 2-MTHF by toluene.

Fig. 2. Comparison of liquid and gaseous spray penetration from simulation (dashed lines) and experiment (solid lines) for 2-MTHF

Page 7: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 7

3.3 Numerical Setup

Computations started from intake valve closure (IVC) at −134.6 deg CA aTDC and ended at exhaust valve opening (EVO) at 120 deg CA aTDC. The starting solution at IVC was initialized with pressure and temperature taken from the experiments. The internal EGR rate was assumed to be 5 %. The EGR composition regarding oxygen, CO, NOX, CO2, and water vapor content was initialized according to the experimental measurements (see Tab. 3). The velocity field was initialized with a swirl number of 1.159, according to the experiments.

The simulation used a sector grid representing 1/8th of the combustion chamber, thereby tak-ing advantage of the axial symmetry with respect to the placement of the nozzle holes. The mesh size was 40224 cells at top dead center (TDC). A sensitivity study revealed that this was a sufficient resolu-tion for all cases. To account for the compression of the grid cells along the cylinder axis due to the piston movement, a cell layer removal technique was applied in the cylinder region throughout the simulations. The CPU time for a complete simulation is about 36 h on a single processor. The wall temperatures were set to 440 K, based on experimental experience, and held constant during the simulations.

The combustion chemistry is discretized on 101 grid points in Z-direction.

3.4 Reaction Chemistry

3.4.1 Chemical Reaction Mechanism

A chemical reaction mechanism for a mixture of n-heptane, iso-octane, toluene, ethanol, and dimethylether is used for the simulations.

A detailed chemical kinetic mechanism for DME (LLNL) was taken from Fischer et al. [10]. The mechanism consists of 79 species and 351 reversible reactions and was validated extensively over a wide range of temperatures, pressures and equivalence ratios [7, 23]. Using a multi-stage reduction method proposed by Pepiot-Desjardins and Pitsch [29], the detailed mechanism was reduced to a skeletal mechanism composed of 33 species and 64 forward and reverse reactions. The reduced mechanism was then combined with a consistent chemical kinetic mechanism for n-heptane, iso-octane, toluene, and ethanol by Pitsch [34] that consists of 631 reactions among 139 chemical spe-cies. A submechanism from [16], which accounts for the thermal, prompt, and nitrous oxide contribu-tions to NOx formation and for NOx reburn by hydrocarbon radicals and amines is added, leading to a total number of 910 reactions and 170 chemical species. The resulting mechanism accounts for low- and high temperature auto-ignition, heat release, and pollutant formation.

Due to the fact that the soot emissions measured were very low for all cases, no soot model was used for the simulations.

3.4.2 Composition of Surrogate Fuel

As stated before, the composition of the surrogate fuel was chosen to satisfy the cetane num-ber, the oxygen content, and the hydrogen/carbon ratio of the experimental fuels. For Surrogate A, matching the H/C ratio of the experimental fuel yields a rather high toluene content of almost 40 %. To reduce the aromaticity of the surrogate, the H/C requirement was relaxed to a H/C ratio of 2.3 for sur-rogate B, while keeping oxygen content and cetane rating constant (see Tab. 4). An additional uncer-tainty is in the cetane numbers of both the experimental fuels and the surrogate mixtures. The cetane ratings of the surrogates are approximated by a linear interpolation of the cetane numbers of the indi-vidual components.

Zero-dimensional ignition-delay time calculations for the surrogates, using the FlameMaster Code [43], are compared to experimental 2-MTHF measurements from [5, 15], which will be discussed in the results section below.

Based on the experience gained from the 2-MTHF surrogates, a surrogate C is proposed for the 2-MTHF/DBE blend (see Tab. 4). Unfortunately, no ignition delay time measurements were availa-ble for this blend, limiting the assessment of the surrogate to the engine cases.

Page 8: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

8 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

Table 4. Properties of experimental fuels and fuel surrogates

2-MTHF Surrogate A Surrogate B 70% 2-MTHF + 30% DBE Surrogate C

Compound formula [%-weight] C5H10O

6.4 % n-heptane39.6 % toluene 18 % ethanol 36 % DME

24 % n-heptane19 % toluene

39.5 % ethanol17.5 % DME

27.7% C8H18O 72.3% C5H10O

15.3% n-heptane32.4% toluene 13.5% ethanol 38.8% DME

H/C ratio 2 1.98 2.32 2.1 2

Oxygen content [%-weight] 18.6 18.6 19.1 16.1 16.6

Cetane number 20 – 30 27 27 25 - 40 33

4 Results and Discussion

4.1 Ignition delay time validation for 2-MTHF surrogates

In Fig. 3, experimental data from [5, 15] for 2-MTHF at stoichiometric air/fuel mixture and 20 bar pressure are compared to results from homogeneous reactor simulations for surrogates A and B.

For surrogate A, the calculated ignition delay times at high and intermediate temperatures (above ~800 K) match the experimental results quite well. For the lower temperatures, the ignition de-lay times calculated from surrogate A are clearly too short, and also the slope as obtained from the calculations is flatter than the slope from the RCM experiments.

The calculated ignition delay times at high and at intermediate temperatures for surrogate B are quite similar to those from surrogate A, matching the experimental results quite well. Despite the same cetane number for both surrogates, surrogate B fits the experimental RCM data for lower tem-peratures better than surrogate A. Nevertheless, the ignition delay times at low temperatures from sur-rogate B are too short and the slope is too flat. Also with other combinations of the given surrogate components (not shown here), the low temperature slope of 2-MTHF, as shown by the RCM data, could not be matched. Here, new surrogate components, better suited to the furanic structure of 2-MTHF, are desirable.

Fig. 3. Ignition delay time over inverse of temperature for 2-MTHF. Stoichiometric conditions at 20 bar. RCM experiments from [5] (red plus symbols), shock tube experiments from [15] (green crosses), and re-

sults from surrogate A (blue line) and surrogate B (black line).

Page 9: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 9

Fig. 4. Pressure traces for 2-MTHF. Load points 3 and 4 from engine experiment (red, solid line), simula-tion with surrogate A (blue, dashed line), and surrogate B (black, dotted line)

In summary, neither surrogate matches the ignition behavior of 2-MTHF over the whole tem-perature range studied. Both surrogates describe the high to intermediate temperature range fairly well. Surrogate B shows better agreement to low to intermediate temperature experiments, while the calculated ignition delay times in this region are still distinctly shorter than the measured ones. Appli-cation of the surrogates to diesel engine simulations is discussed in the next subsection.

4.2 Comparison of experimental and numerical results

Figure 4 compares the pressure traces for 2-MTHF from the diesel engine experiment and the numerical results for surrogates A and B. In the experiments, 2-MTHF did not ignite at the low load points 1 and 2, so here only the load points 3 and 4 (see Tab. 2) are shown.

For both surrogates, ignition timing, peak pressure, and burn rate from experiment and simula-tion match acceptably. The first stage ignition and the corresponding low temperature heat release are apparently overestimated. This corresponds to the too fast low temperature chemistry, observed in Fig. 3.

As observed from Fig. 4, Surrogate B works better as surrogate for 2-MTHF than surrogate A, confirming the relaxation of the H/C ratio criterion (see Tab. 4). The general ignition timing seems to be captured quite well by the cetane number, although here the cetane number does not capture the differences in the low temperature range.

Figure 5 compares the pressure traces for the 2-MTHF/DBE blend from the diesel engine ex-periments and the numerical results for surrogate C (see Tab. 4). In the experiment, 30% DBE was blended to the 2-MTHF in order to enable stable operation in the low load points 1 and 2.

Simulations using surrogate C match the ignition timings of the experiments fairly well at all load points. The prediction quality for peak pressures and burn rates is rather weak, except for load point 2 (upper right part of Fig. 5). As for the other surrogates, the first stage ignition is distinctly over predicted. The overall simulation quality is clearly worse than for pure 2-MTHF, emphasizing the im-portance of kinetic experimental data for the surrogate validation.

Page 10: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

10 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

4.3 Assessment of pollutant formation

From the preceding subsection it is evident that only the results for surrogates A and B have a quality that enables assessment of pollutant formation. Figure 6 shows the emissions at exhaust valve opening for load points 3 and 4 from operation with pure 2-MTHF and from simulations with surrogate A and surrogate B. Regarding CO emissions, the simulations overestimate the experimental results. While the trend is clearly captured by surrogate B, surrogate A does not capture the trend. NOX emis-sions are underestimated by both surrogates, while the trend is captured well. Again, the performance of surrogate B is slightly better.

Fig. 6. CO (left) and NOX (right) emissions from experiment (red) and simulation with surrogate A (blue) and surrogate B (black) for 2-MTHF. Load points 3 and 4

Fig. 5. Pressure traces for 2-MTHF/DBE blend. Load points 1 to 4 (from left to right, top to bottom) from en-gine experiment (red, solid line) and from simulation with surrogate C (green, dashed line)

Page 11: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 11

4.4 Ignition behavior and flame structure

In the following, the flamelet solution is analyzed with regard to ignition, yielding additional in-formation about the low temperature chemistry of Surrogates A and B. In addition, the flame structure is analyzed with regard to pollutant formation, which yields explanations about the pollutant formation pathways and the differences between the two load points.

As described above, the chemistry is solved in flamelet space. In the flamelet, a mixture frac-tion value of Z = 0 corresponds to pure oxidizer, while a value of Z = 1 corresponds to pure fuel. The mixture fraction can be seen as a normalized equivalence ratio, ranging from zero to one. The stoichi-ometric mixture fraction depends not only on the fuel, but also on the oxygen concentration of the oxi-dizer and thus on the EGR rate. For the given boundary conditions, Zst is 0.06 for load point 3 and 0.057 for load point 4. The species mass fractions for each CFD cell are obtained by the integration of a β-PDF [12], scaled by the mean and the variance of the mixture fraction in the cell.

Figure 7 shows flamelet profiles for load point 3 at four times, chosen to represent different stages during the ignition process. For better analysis of the graphs, only the range from Z = 0 to Z = 0.3 is shown, while computations included the whole range from zero to one. The figures show the mass weighted probability density function (PDF) of the mixture fraction for the whole CFD domain and the temperature and Keto profiles as resulting from the simulations with surrogates A and B. Here Keto abbreviates ketohydroperoxide species, which play a major role for low temperature chain branching reactions leading to only small increases in temperature (i.e. first stage ignition). The pres-ence of ketohydroperoxide species is an indicator for ongoing low-temperature chemistry.

At 5 deg. CA before TDC (upper left of Fig. 7), the PDF shows a range of mixture fraction from zero to 0.15 indicating a stratified mixture, and a peak at Z = 0, which indicates the presence of pure, unmixed oxidizer in the cylinder. The temperature has not yet increased and only negligibly small mass fractions of Keto-components are observed. This corresponds to the pressure trace (left part of

Fig. 7. Flamelet solution profiles for load point 3 at different times. 5 deg. CA before TDC (upper left), TDC (upper right), 5 deg. CA after TDC (lower left), and 10 deg. CA after TDC (lower right). PDF (red, solid line), temperature for surrogate A (blue, dashed line) and B (black, dashed line), and Keto mass fraction for surro-

gate A (turquois, dotted line) and B (grey, dotted line). Stoichiometric mixture fraction (green, dash-dotted line) is Zst = 0.06

Page 12: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

12 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

Fig. 8. Flamelet solution profile for load point 3. PDF (red, solid line), NOX mass fraction from surrogate A (blue, dashed line) and B (black, dashed line), CO mass fraction from surrogate A (turquois, dotted line) and

B (grey, dotted line). Left side: at 10 deg CA after TDC, right side: at 50 deg CA after TDC. Stoichiometric mixture fraction (green, dashed-dotted line) is Zst = 0.06

Fig. 4), where at this time no observable ignition has taken place. At TDC (upper right of Fig. 7), the pressure trace (Fig. 4) indicates a first ignition stage for the simulations. Corresponding to this, the temperature profiles indicate an only small temperature rise with temperatures still below 1000 K. Substantial amounts of Keto-species are present. Keto concentrations and temperatures are slightly higher for surrogate A. At 5 deg. CA after TDC (lower left of Fig. 7), the temperature profiles indicate second stage ignition yielding high temperatures in the rich part of the flamelet (Z>Zst). Due to homog-enisation, only very small parts of the mixture are still rich, as indicated by the PDF. Keto species are only present in very rich parts of the flamelet, where second stage ignition has not yet taken place and temperatures are below 1000 K (Z > 0.15 for surrogate B, Z > 0.2 for surrogate A). At 10 deg. CA after TDC (lower right of Fig. 7), the fully ignited mixture fraction range is broadened towards the richer and the leaner side, as depicted by the substantially increased temperature. The steep temperature gradi-ent, especially at the lean side, corresponds to a comparably low dissipation rate and a lean premixed flame. At lean mixtures, PDF and high temperature overlap, resulting in a strong pressure rise in the cylinder, as can be observed in the left part of Fig. 4. The higher H/C ratio of surrogate B leads to a slightly higher maximum temperature at Zst and also a slightly faster temperature rise towards lean mixtures, resulting in the differences in peak pressure observed in Fig. 4.

The findings made in the context of Fig. 7 are concordant to the findings from Figs. 3 and 4 showing too strong low temperature chemistry for both surrogates. Analysis of the flamelet solution for load point 4 yields essentially the same information, but is not shown here for brevity. When compar-ing Figs. 3 and 7 however, it should be noted that for Fig. 3 the conditions are stoichiometric at 20 bar pressure, while the engine experiments and simulations are undertaken at globally lean conditions at a pressure greater than 50 bar, inhibiting one to one comparison of temperatures.

Figures 8 and 9 show the CO mass fraction, the NOX mass fraction (scaled by a factor of 100 to fit the axis), and the PDF. For better analysis of the graphs, only the range from Z = 0 to Z = 0.1 is shown, although computations included the complete range from zero to one. The left sides of Figs. 8 and 9 show the flamelet profiles at 10 and 16 deg CA after TDC, respectively, while the right sides show the profiles at 50 deg CA after TDC. The timings where chosen to illustrate the flame structure just after the main pressure rise (compare Fig. 4), and after end of combustion.

In the left part of Fig. 8, most of the mixture fraction is distributed around a mean value of Z = 0.03, while no mixture reaches beyond Zst. The mixture is lean everywhere in the cylinder, indicating sufficient premixing. As expected, the CO mass fraction sharply rises in the rich region (right of Zst), and the NOX mass fraction peaks at slightly lean conditions (directly left of Zst). Another CO peak is observed at very lean conditions (Z = 0.025), indicating yet unburned lean and presumably cold mix-ture.

At 50 deg CA after TDC (right part of Fig. 8), the PDF indicates a more homogeneous mixture. The peak at Z = 0 is distinctly lower, while the PDF distribution in general is narrower. A fraction of the CO in the lean region has burned, while a substantial amount still remains, leading to the CO emis-sions observed. There is no overlap of the PDF and the CO profile for Z > 0.05, indicating that CO from rich regions does not contribute to the engine out emissions. The pollutant profiles from surro-gates A and B do not differ much at this load point. The observations regarding CO emissions made here are in agreement with the results of Musculus et al. [28], where it was shown experimentally that

Page 13: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 13

Fig. 9. Flamelet solution profile for load point 4. PDF (red, solid line), NOX mass fraction fraction from surro-gate A (blue, dashed line) and B (black, dashed line), CO mass fraction from surrogate A (turquois, dotted line) and B (grey, dotted line). Left side: at 16 deg CA after TDC, right side: at 50 deg CA after TDC. Stoi-

chiometric mixture fraction (green, dash-dotteded line) is Zst = 0.057

unburned hydrocarbon emissions in a low temperature combustion diesel engine originated from very lean regions, were the reaction rates are too low to complete combustion. For the later time in the cy-cle (right part of Fig. 8), the NOX mass fraction is, due to diffusive effects, more evenly distributed.

For load point 4 (Fig. 9), the flamelet profiles are quite similar to those of Fig. 8. However, the increased average mixture fraction leads to a higher average value of the mixture fraction PDF and less unmixed oxidizer, especially after combustion (right part of Fig. 9). The profiles for CO and NOX are slightly different from those observed in Fig. 8. For surrogate B, the main difference is the focus of the PDF at higher values of Z, which yields less contribution from the very lean region and thus less CO emissions, and at the same time a higher contribution from regions close to stoichiometric with higher NOX mass fraction. For surrogate A, the lean CO profile is shifted to less lean regions, which results in a stronger overlap of PDF and CO profile and thus in too high engine-out emissions from simulation with surrogate A (compare Fig. 6).

Conclusions

A surrogate approach to describe the combustion chemistry of novel biofuels in a diesel en-gine has been described and tested.

The surrogate fuels have been compiled according to cetane number, oxygen content, and H/C ratio of the biofuels. Ignition delay time calculations and diesel engine simulations, both using de-tailed reaction chemistry of surrogate fuels, have been compared to the respective experiments for 2-MTHF. The ignition delay time calculations and the analysis of the engine simulation results revealed limitations of the surrogate approach regarding the low temperature behavior of the surrogate. In the future, additional surrogate components will be employed to address this issue. For the engine simula-tions of 2-MTHF, satisfactory results regarding pressure traces and pollutant emissions were obtained. This confirmed the proposed approach and the methodology to define the surrogate composition, and allowed for better insight into the pollutant formation processes in the engine.

The assessment of a surrogate for a 2-MTHF/DBE blend was limited to engine simulations, as no ignition delay time measurements were available. The simulation quality was distinctly lower than for the pure 2-MTHF surrogates, showing the importance of kinetic experiments for the surrogate vali-dation and indicating the limitations of the cetane number as index for the ignition behavior.

Acknowledgement

This work has been performed within the Cluster of Excellence “Tailor-Made Fuels from Bio-mass”, which is funded by the Excellence Initiative of the German federal and state governments to promote science and research at German universities.

Page 14: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

14 B. Kerschgens, T. Lackmann, H. Pitsch, A. Janssen, M. Jakob, S. Pischinger

References [1] Adolph, D.; Busch, H.; Pischinger, S.; Kolbeck, A.; Lamping, M.; Körfer, T. MTZ, 2008, vol. 69, pp. 42-50. [2] Barths, H., Hasse, C., Bikas, G., and Peters, N.; Simulation of Combustion in DI Diesel Engines using an Eu-

lerian Particle Flamelet Model. Proc. Combust. Inst., 28:1161-1168, 2000. [3] Barths, H., Hasse, C., and Peters, N.; Computational fluid dynamics modelling of non-premixed combustion

in direct injection diesel engines. International Journal of Engine Research, 1:249-267, 2000. [4] Barths, H., Pitsch, H., and Peters, N.; 3D Simulation of DI Diesel Combustion and Pollutant Formation Using

a Two-Component Reference Fuel. Oil and Gas Science and Technology, Rev. IFP 54-2:233-244, 1999. [5] Brassat, A., Thewes, M., Müther, M., Pischinger, S., Lee, C., Fernandes, R.X., Olivier, H., Uygun, Y.; Analy-

sis of the Effects of Certain Alcohol and Furan-Based Biofuels on Controlled Auto Ignition. SAE Paper No. 2012-01-1135

[6] Beeckmann, J.; Cai, L.; Röhl, O.; Pitsch, H.; Peters, N.: A Reduced Kinetic Reaction Mechanism for the Au-toignition of Dimethyl Ether. SAE Paper 2010-01-2108, 2010

[7] Curran, H. J., S. L. Fischer, and F. L. Dryer, "The Reaction Kinetics of Dimethyl Ether. II: Low-Temperature Pyrolysis and Oxidation in Flow Reactors,"Int. J. Chem. Kinet. 32: 741–759, 2000

[8] Ewald, J., Freikamp, F., Paczko, G., Weber, J., Haworth, D. C., and Peters, N.; GMTEC: GMTEC Develop-er's Manual.Technical report, Advanced Combustion GmbH, 2003.

[9] Ferziger, J. H. and Perić, M.. Computational Methods for Fluid Dynamics. Springer, Berlin, 2002. [10] Fischer, S. L., F. L. Dryer, and H. J. Curran, "The Reaction Kinetics of Dimethyl Ether. I: High-Temperature

Pyrolysis and Oxidation in Flow Reactors," Int. J. Chem. Kinet. 32: 713–740, 2000. Lawrence Livermore Na-tional Laboratory, Livermore, CA, UCRL-JC-239461

[11] Girimaji, S. S.. On the Modelling of Scalar Diffusion in Isotropic Turbulence. Physics of Fluids A, 4:2529-2537, 1992.

[12] Hasse, C., Barths, H., and Peters, N., “Modelling the Effect of Split Injections in Diesel Engines Using Repre-sentative Interactive Flamelets,” SAE Technical Paper 1999-01-3547, 1999, doi:10.4271/1999-01-3547

[13] Hergart, C., Barths, H., and Peters, N., “Modeling the Combustion in a Small-Bore Diesel Engine Using a Method Based on Representative Interactive Flamelets,” SAE Technical Paper 1999-01-3550, 1999, doi: 10.4271/1999-01-3550.

[14] Heufer, K.A.; Fernandes, R.X.; Olivier, H.; Beeckmann, J.; Röhl, O.; Peters, N.: Shock tube investigations of ignition delays of n-butanol at elevated pressures between 770K and 1250K. P. Combust. Inst., vol. 33, pp. 359-366, 2010

[15] Heufer K.A., Olivier H. Medvedev S.P., Khomik S.V.; Optical Investigation of Shock Induced Ignition of Differ-ent Biofuels; 23rd ICDERS July 24-29, 2011 Irvine, USA

[16] Hewson, H. C. and Bollig, M.. Reduced Mechanisms for NOx Emissions from Hydrocarbon Diffusion Flames. Symposium (International) on Combustion, 26:2171-2180, 1996.

[17] Hottenbach, P., Brands, T., and Grünefeld, G., "An Experimental Investigation on the Evaporation Character-istics of a Two-Component Fuel in Diesel-Like Sprays," SAE Int. J. Engines 4(1):800-812, 2011, doi:10.4271/2011-01-0688.

[18] Jakob, M.; Pischinger, S.; Adomeit,P.; Kolbeck, A.; “Glow-plug Ignition of Ethanol Fuels under Diesel Engine Conditions” SAE Technical Paper 2011-01-1391

[19] Janssen, A.; Muether, M.; Pischinger, S.; Kolbeck, A.; Lamping, M.; Koerfer, T.: Tailor-Made Fuels for Future Advanced Diesel Combustion Engines. SAE Paper 2009-01-1811, 2009

[20] Janssen, A.; Muether, M.; Pischinger, S.; Kolbeck, A.; Lamping, M.: Tailor-Made Fuels: The Potential of Oxy-gen Content in Fuels for Advanced Diesel Combustion Systems. SAE Paper 2009-01-2765, 2009

[21] Janssen, A.; Muether, M.; Pischinger, S.: Potential of Cellulose-Derived Biofuels for Soot Free Diesel Com-bustion. SAE Int. J. Fuels Lubr., vol. 3 (1), pp. 70-84, 2010

[22] Janssen, A.; Jakob, M.; Müther, M.; Pischinger, S.; Klankermayer, J.; Leitner, W.: Tailor-made Fuels from Bi-omass – Potential of Biogenic Fuels for Reducing Emissions; MTZ, 2010, vol. 71, pp. 922-928.

[23] Kaiser, E. W., T. J. Wallington, M. D. Hurley, J. Platz, H. J. Curran, W. J. Pitz, and C. K. Westbrook, “Experi-mental and Modeling Study of Premixed Atmospheric-Pressure Dimethyl Ether-Air Flames,” Journal of Physi-cal Chemistry A 104, No. 35, 8194-8206 (2000)

[24] Khalighi, B., El Tahry, S., Haworth, D., and Huebler, M., “Computation and Measurement of Flow and Com-bustion in a Four-Valve Engine with Intake Variations,” SAE Technical Paper 950287, 1995, doi:10.4271/950287

[25] Lamping, M.; Kolbeck, A.; Körfer, T.; Adolph, D.; Busch, H.; Pischinger, S. MTZ, 2008, vol. 69, pp. 24-31. [26] Lamping, M.; Janssen, A.; Kolbeck, A.; Schernus, C.; Wohlberg, R.; Wedowski, S.; Körfer, T. Application of

increased power density for future Diesel engines - a requirement for downsized powertrains; Fisita, Buda-pest, 2010.

[27] Muether, M., Lamping, M., Kolbeck, A., Cracknell, R. et al., "Advanced Combustion for Low Emissions and High Efficiency Part 1: Impact of Engine Hardware on HCCI Combustion," SAE Technical Paper 2008-01-2405, 2008, doi:10.4271/2008-01-2405

Page 15: Tailored surrogate fuels for the simulation of diesel ... · Different engines were investigated, ranging from small displacement volumes (300 cm3/cylinder) to marine diesel (4800

Tailored surrogate fuels for the simulation of diesel engine combustion of novel biofuels 15

[28] Musculus, M., Lachaux, T., Pickett, L., and Idicheria, C.,“End-of-Injection Over-Mixing and Unburned Hydro-carbon Emissions in Low-Temperature-Combustion Diesel Engines,” SAE Technical Paper 2007-01-0907, 2007, doi: 10.4271/2007-01-0907.

[29] Pepiot-Desjardins, P.; Pitsch, H.; An efficient error-propagation-based reduction method for large chemical kinetic mechanisms Original Research Article; Combustion and Flame, Volume 154, Issues 1–2, July 2008, Pages 67-81

[30] Peters, N.. Local quenching of diffusion flamelets and non-premixed turbulent combustion. Western States Section of the Combustion Institute, paper WS 80-4, Spring Meeting, Irvine CA, 1980.

[31] Peters, N.. Laminar Diffusion Flamelet Models in Non-Premixed Turbulent Combustion. Prog. Energy Com-bust. Sci., 10:319-339, 1984.

[32] Peters, N.. Laminar flamelet concepts in turbulent combustion. Symposium (International) on Combustion, 21:1231-1250, 1986.

[33] Peters, N.. Turbulent Combustion. Cambridge University Press, Cambridge, 2000. [34] Pitsch, H.. Private communication. [35] Pitsch, H, “Entwicklung eines Programmpaketes zur Berechnung eindimensionaler Flammen am Beispiel

einer Gegenstromdiffusionsflamme”, Master’s thesis, RWTH Aachen, Germany, 1993 [36] Pitsch, H., Barths, H., and Peters, N., “Three-Dimensional Modeling of NOx and Soot Formation in DIDiesel

Engines Using Detailed Chemistry Based on the Interactive Flamelet Approach,” SAE Technical Paper 962057, 1996, doi:10.4271/962057.

[37] Spiekermann, P., Jerzembeck, S., Felsch, C., Vogel, S., Gauding, M., and Peters, N.. Experimental Data and Numerical Simulation of Common-Rail Ethanol Sprays at Diesel Engine-Like Conditions. Atomization and Sprays, 19:357-386, 2009.

[38] Vranckx, S.; Heufer, K.A.; Lee, C.; Olivier, H.; Schill, L.; Kopp, W.A.; Leonhard, K.; Taatjes, C.A.; Fernandes, R.X.: Role of Peroxy Chemistry in the High Pressure Ignition of n-Butanol - Experiments and Detailed Kinetic Modeling. Combust. Flame, vol. 158, pp. 1444-1455, 2011

[39] Vranckx, S.; Lee, C.; Fernandes, R.X.: Auto-ignition kinetics of biomass derived alternative fuels for advanced combustion. SAE Paper 2011-01-1780

[40] Weber, J., Peters, N., Diwakar, R., Siewert, R. et al., “Simulation of the Low-Temperature Combustion in a Heavy Duty Diesel Engine,” SAE Technical Paper 2007-01-0904, 2007, doi:10.4271/2007-01-0904.

[41] Williams, F.A.. Recent advances in theoretical descriptions of turbulent diffusion flames. Turbulent Mixing in Nonreactive and Reactive Flows, Plenum Press, New York: 189-208, 1975.