1 Introduction

Reforming of hydrocarbons to a hydrogen- (H2) and carbon monoxide (CO)-rich reformate gas or synthesis gas is a well-established technology in the chemical industry. However, in mobile applications with strong restrictions, for example in terms of packaging, system complexity, and weight, the industrial fuel reformer technologies need to be scaled down and adapted. Further challenging requirements for automotive applications in comparison to stationary systems are a long lifetime without extensive maintenance, low costs, no or minimum refueling of second working fluid (e.g., water), highly dynamic operation, and a high number of cold starts.

Automotive fuel reforming could be used in different application areas. One possibility is the combustion of H2 and CO either thermally in a combustion engine or electrochemically in a fuel cell for vehicle propulsion. Besides, it can be used for electric power generation in an auxiliary power unit (APU), or H2 can be added to gasoline- or diesel-fueled combustion engines for enhancement of pollutant formation and fuel consumption [1–8]. Finally, it might be applied for after treatment issues as it is done in this investigation [9–18]. The diesel fuel reformer that is discussed in this paper is part of a combined Lean NOX Trap (LNT) + selective catalytic reduction (SCR) diesel exhaust after treatment system that was described in detail by the authors in a previous paper [19].

In general, in the reforming process, hydrocarbons get converted to a H2- and CO-rich synthesis gas by oxidation. The different reforming processes and reactions can be classified by the oxidizing agent (O2, H2O, CO2). An additional characterization consists of the energetic process and reaction behavior (endothermic or exothermic).

The catalyzed partial oxidation (CPOx) is an exothermic reforming reaction and is based on molecular O2. The strong understoichiometric reaction (air/fuel ratio λ ≈ 0.34) is usually adiabatically operated at 700–1,000 °C [20]. The λ operation range is narrow. Low λ values (<0.34) are critical due to soot and methane formation. High λ values (>0.39) are critical due to exothermic temperature increase. In comparison to the endothermic reforming processes, the complexity of CPOx is low. On the one hand, the exothermic character is advantageous. No complex energy supply is necessary, and the process can be operated very dynamically. However, the high system temperature requires thermal stable catalysts and a very accurate operation control. In addition, an air supply is necessary.

By the endothermic steam reforming (SR), the hydrocarbon conversion takes place with H2O and is operated at 750–900 °C [20]. Considerable energy needs to be supplied to operate the reaction and to vaporize H2O. Therefore, SR is a complex technology for automotive application. The reaction enthalpy needs to be generated, for example by fuel combustion, and transferred to the reactants or to the catalyst by heat exchangers. This results in a low dynamic load change. Additionally, a closed H2O circulation, including for example tank and vaporizer, is necessary. This leads to a high system complexity, package, and weight.

A second endothermic reforming process is the CO2 or dry reforming (DR) that takes place with CO2. To avoid soot formation, DR needs to be operated >900 °C [20]. In comparison to SR, DR has less technical relevance up to now. Nevertheless, it is a possible reaction pathway for reforming with CO2-containing exhaust gas or of CO2-containing natural gas and bio gas.

A combination of exothermic and endothermic reactions is advantageous for the reforming process control. The autothermal reforming (ATR) is in ideal case energetically neutral. Oxidative steam reforming (net endothermic) and steam-supported partial oxidation (net exothermic) are intermediate stages. Fuel reforming with variable exhaust gas can represent all energetic stages. In autothermal and net exothermic reactors, the required reaction enthalpy of the endothermic reactions can be supplied by the exothermic reactions inside the reactor without complex heat exchange components. Hence, the system can be designed compactly and operated dynamically. Additionally, the critical high reaction temperature of the CPOx can be reduced and damped by endothermic heat consumption. Pure combination of SR and CPOx still requires water and air supply. In contrast, using O2-, H2O-, and CO2-containing exhaust gas from a combustion engine offers potential to reduce the supply systems of additional working fluids. Summarizing, exhaust gas-based reforming offers potential to design a compact and dynamic reactor with advantages in terms of thermal behavior as well as energetic efficiency.

2 Description of Reformer Concept and Layout

The outline on-board fuel reformer as part of the combined LNT + SCR after treatment system is predominantly operated by exhaust gas, which is taken downstream diesel particulate filter (DPF) (the LNT + SCR combination is positioned downstream DPF, detailed system description in [19]). In general, three different concepts for reactant mixing upstream reforming catalyst can be distinguished. In the first type, the reactant mixing processes have no thermal interaction with the hot reforming zone (for example, the CPOx reactor in [21]). The second type is a coaxial design with reforming catalyst in the inner part and flow of fresh air or exhaust gas around the catalyst. By this, the reactant mixing processes are thermally interacting with the hot reforming zone (for example, the CPOx reactor in [15, 22]). The third type is also a coaxial design but with a ring-shaped reforming catalyst and flow of fresh air or exhaust gas along the inner walls of the catalyst. This concept also offers high potential of thermal interaction of the reactant mixing processes and the hot reforming zone. Similar concepts are described in [23] and [24]. The developed on-board fuel reformer is a coaxial design according type 3 and shown in detail in Fig. 1. Diesel is injected into the reformer mixing chamber by a serial air-assisted, low-pressure injector (5 bar fuel injection pressure, 1 bar air pressure) at a significant understoichiometric air/fuel ratio (λ Ref = 0.35–0.80) and is catalyzed to a H2- and CO-rich synthesis gas.

Fig. 1
figure 1

CAD design of the fuel reformer

Full size image

The reaction enthalpy, which is released in the ring-shaped reformer catalyst, supports fuel vaporization via internal heat transfer. The reforming reaction process can be described by superposition of exothermic CPOx by the residual oxygen in the exhaust gas as well as the secondary air and endothermic SR and DR which occurs via H2O and CO2 introduced by the exhaust gas. Moreover, the reforming products are temperature dependently in chemical equilibrium, among others mainly according to the water gas shift (WGS) and methanation reaction.

The electric heating device is only activated during cold start. As soon as the reformer catalyst is heated up to a temperature of 250–300 °C, the heating device is switched off and the proceeding of heating up takes place by lean oxidation of small diesel injections on the reformer catalyst. Because of the outline position, the reformer does not cool down fast during low engine load operation. Hence, the electrical-supported reformer cold start is only necessary after long break intervals with engine off.

3 Investigation Methodology

3.1 Theoretical

3.1.1 Chemical Equilibrium Calculations

The theoretical study of the thermochemistry of the exhaust gas reforming was done with a commercially available chemical reaction and equilibrium software [25]. The applied diesel fuel was defined according the composition from [26] and implemented in the software. Missing fuel species in the base library were implemented manually with data sets from [27]. The applied diesel fuel composition results in a theoretical hydrocarbon species C12.93H24.40. The exhaust gas composition (residual O2, H2O, CO2, N2) and the corresponding air/fuel ratio λ were calculated according to the reaction equations. The equilibrium determination was conducted for the following gas matrix: diesel (all single species from [26]), H2O, O2, H2, CO2, CO, CH4, and C.

3.1.2 Definition of Reforming Parameters

Due to the strong λ dependence of the exhaust gas composition of the real engine, the energetic reforming condition (exothermic for high residual oxygen in very lean exhaust gas, endothermic for λ close to 1) is strongly variable. In order to describe the reforming condition, three standardized parameters were defined beside conventional parameters like temperature and space velocity. The first parameter r Exhaust is the ratio of the exhaust gas molar flow to the sum of exhaust gas and secondary air molar flows in the reformer (Eq. 1). This parameter is an indicator for the secondary air portion in the reformer inlet gas flows. The second parameter λ of the reformate gas and of the input gas flows is calculated according to Eq. 2 (O is number of oxygen atoms in all exhaust components, C is number of carbon atoms in all exhaust components, H is number of hydrogen atoms in all exhaust components). The third parameter r POx is the main indicator of the exhaust gas reforming process (Eq. 3). It describes the fraction of the molar oxygen source to the total amount of all oxygen sources for the reforming reactions (O2, H2O, and CO2). r POx = 1 is pure CPOx (maximum exothermic) and r POx = 0 is pure SR + DR (maximum endothermic). It can either be calculated by the molar flows or by the reforming reaction characterization parameters S/C, C/C, and O/C (Eqs. 4–6) that are derived from the stoichiometric coefficients of the reforming reaction equations (Eqs. 7–9). The theoretical and experimental results are discussed based on these parameters.

$$ {r}_{\mathrm{Exhaust}}=\frac{{\overset{\cdot }{n}}_{\mathrm{Exhaust}}}{{\overset{\cdot }{n}}_{\mathrm{Exhaust}}+{\overset{\cdot }{n}}_{\mathrm{Air}}} $$
(1)
$$ \lambda =\frac{{\displaystyle \sum {n}_O}}{2\cdot {\displaystyle \sum {n}_C}+0.5\cdot {\displaystyle \sum {n}_H}} $$
(2)
$$ {r}_{POx}=\frac{2\cdot {\overset{\cdot }{n}}_{O_2}}{2\cdot {\overset{\cdot }{n}}_{O_2}+{\overset{\cdot }{n}}_{H_2O}+{\overset{\cdot }{n}}_{C{O}_2}}\kern.3em or\kern.3em \frac{\mathit{\mathsf{O}}/\mathit{\mathsf{C}}}{\mathit{\mathsf{O}}/\mathit{\mathsf{C}}+\mathit{\mathsf{S}}/\mathit{\mathsf{C}}+\mathit{\mathsf{C}}/\mathit{\mathsf{C}}} $$
(3)
$$ \mathit{\mathsf{S}}/\mathit{\mathsf{C}}=\frac{{\overset{\cdot }{n}}_{H_2O}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(4)
$$ \mathit{\mathsf{C}}/\mathit{\mathsf{C}}=\frac{{\overset{\cdot }{n}}_{C{O}_2}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(5)
$$ \mathit{\mathsf{O}}/\mathit{\mathsf{C}}=\frac{2\cdot {\overset{\cdot }{n}}_{O_2}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(6)
$$ {C}_x{H}_y+x{H}_2O\rightleftarrows xCO+\left(\mathit{\mathsf{x}}+\frac{\mathit{\mathsf{y}}}{\mathit{\mathsf{2}}}\right){H}_2\kern3em {\varDelta}_R{\mathrm{H}}_{298}^0=+1,911\frac{kJ}{mo{l}_{{\mathrm{C}}_{12.93}}{{}_{\mathrm{H}}}_{{}_{24.40}}} $$
(7)
$$ {C}_x{H}_y+xC{O}_2\rightleftarrows 2xCO+\frac{y}{2}{H}_2\kern3em {\varDelta}_R{H}_{298}^0=+2,442\frac{kJ}{mo{l}_{C_{12.93}}{{}_H}_{{}_{24.40}}} $$
(8)
$$ {C}_x{H}_y+\frac{1}{2}x{O}_2\rightleftarrows xCO+\frac{y}{2}{H}_2\kern3em {\varDelta}_R{H}_{298}^0=-1,215\frac{kJ}{mo{l}_{C_{12.93}}{{}_H}_{{}_{24.40}}} $$
(9)

3.2 Experimental

3.2.1 Catalyst Preparation

The reformer catalyst is ring-shaped and based on ceramic substrate with a PGM load of 60 g/ft3 (0/0/1 Pt/Pd/Rh) and a volume of 0.5 L (company CCD GbR). The cell density of the reforming substrate is 600 cpsi. The reforming catalyst was not initially thermally aged but did not show any degradation in reference points during the experimental investigation period.

3.2.2 Engine Test Bench Experiments

The full-scale experiments for investigating the performance of the fuel reformer were conducted with a diesel engine on a transient engine test bench. The control logic of the reformer was running on a rapid control prototyping system (ES1000 Fa. ETAS). Communication between engine and reformer control was done via ECU bypass. The emissions were measured by different analyzers (chemiluminescence detection (CLD): NOX, NO; non-dispersive infrared spectrometry (NDIR): CO, CO2; Magnos O2 detection: O2; flame ionization detection (FID): total hydrocarbons (THC); Fourier transformation infrared spectrometry (FTIR): NO, NO2, CO, CO2, H2O, NH3, N2O, CH4, SO2; process mass spectrometer (PrMS): H2, H2S, SO2, COS, CS2) and sensors (NOX, λ) at various positions (engine out, downstream of the DPF, end of pipe). CLD, NDIR, FID, and Magnos were combined in one exhaust analyzer system (AMA). The PrMS is an analytical unit constructed by V&F including the TWIN-MS (CIMS with soft ionization by source gases: Xe, Kr, Hg) and the H-Sense (EI with ionization by ion beam), which allows to detect all data parallel in lifetime with a very high dynamic (30 ms for each substance). The gas composition of the outline exhaust gas fuel reformer was not measured directly in the reformate gas but calculated based on emission measurements in the full exhaust gas flow and the according dilution factor. The diluted measurement was chosen in order to avoid HC concentrations on high level and avoid measurement device errors. The thermocouples in the reformer catalyst have a diameter of 0.5 mm. The total after treatment system was setup very modularly. For the reformer measurements, the LNT (original position in the upper exhaust bypass) and the SCR (original position downstream the mixer unit) catalyst units were replaced by empty exhaust pipes. A schematic overview of the experimental setup on the engine test bench is given in Fig. 2.

Fig. 2
figure 2

Schematic experimental setup on engine test bench: overview of exhaust gas analyzers, sensors, actuators, and rapid prototyping control

Full size image

4 Results and Discussion

4.1 Theoretical Study of Thermochemistry

A theoretical study of the thermochemistry was conducted in order to investigate the boundary conditions of the fuel reforming on base of highly variable exhaust gas composition. Furthermore, the definition of the control strategy is based on these considerations.

The core parameter of the exhaust gas reforming is the variable exhaust gas composition. For low λ Engine close to 1 with very low residual O2, e.g., during high engine load, the reaction conditions in a purely exhaust gas-based reformer are very endothermic (r POx close to 0). The reformer operation under these conditions requires either high exhaust temperatures or a very high λ Ref in order to supply the necessary reaction enthalpy to the reformer. In contrast, during low engine load or in extreme case during engine overrun with high residual O2 and only minor H2O and CO2, the reaction conditions are very exothermic (r POx close to 1). Under these conditions, the reformer operation is critical in case of high exhaust gas temperatures. Figure 3 shows the theoretically required λ Ref for an adiabatic exhaust gas reformer in order to achieve, for example, a reformer temperature of 900 °C as function of λ Engine for two different T Exhaust. As an example, the left diagram shows an outline reformer with low T Exhaust, whereas the right diagram shows an in-line reformer with high T Exhaust. Additionally, the reductant formation yield (Y Red) as an important indicator in terms of energetic efficiency is shown. Y Red is defined as ratio of oxygen consumption potential of the reducing agent available in the exhaust gas relative to the additional fuel introduced to reach rich operation. It can generally be calculated by Eq. 10.

Fig. 3
figure 3

Theoretically required λ Ref for adiabatic exhaust gas reforming in order to achieve a reformer temperature of 900 °C (as example) and the corresponding reductant yield Y Red in chemical equilibrium as function of λ Engine: a T Exhaust = 100 °C, b T Exhaust = 600 °C; zone II: useable λ operation range, zone I, III: not useable λ operation range (T Diesel = 25 °C)

Full size image
$$ {Y}_{\mathrm{Re}d}=\frac{{\overset{\cdot }{n}}_{H_2}+{\overset{\cdot }{n}}_{CO}+\left(2x+\frac{y}{2}\right)\cdot {\overset{\cdot }{n}}_{C_x{H}_y}-2\cdot {\overset{\cdot }{n}}_{O_2}-{\overset{\cdot }{n}}_{NO}-2\cdot {\overset{\cdot }{n}}_{N{O}_2}}{\left(2x+\frac{y}{2}\right)\cdot \overset{\cdot }{n_{C_x{H}_y, Diesel}}} $$
(10)

For low λ Engine, λ Ref needs to be enleaned up to 1 in order to achieve the target reformer temperature. Y Red is zero and hence this operation range is, besides heating up, not practically useful (zone I). The higher the λ Engine, the lower must be λ Ref to keep the target reformer temperature. For high T Exhaust and λ Engine, the required λ Ref reaches the minimum (λ Ref,min). Below λ Ref,min, soot formation and considerable HC slip will occur. Thus, this operation range is not useful (zone III). Summarizing, high T Exhaust are advantageous for low λ Engine. However, the useable λ Engine operation range is limited. An outline positioned reformer in combination with a secondary air supply for low λ Engine offers the best potential.

The complete operating limits and the possible λ Ref operation range for adiabatic exhaust gas reforming assuming chemical equilibrium as function of r POx are shown in Fig. 4. The operating range is defined by four limits ((1) maximum T Ref, (2) start of soot formation, (3) minimum T Ref for minimum reaction rate and avoidance of methanation, (4) λ Ref,min).

Fig. 4
figure 4

Possible λ Ref operation range for adiabatic exhaust gas reforming in chemical equilibrium as function of r POx (λ Engine as example for r Exhaust = 0.8) (r Exhaust = 0.8, T Exhaust = 250 °C, T Air = 25 °C, T Diesel = 25 °C)

Full size image

For decreasing r POx, λ Ref needs to be increased to keep the reformer within the limits. The λ Ref operation range gets wider and is less sensitive to inaccuracies in the reformer control. Moreover, the soot formation limit is less critical. In contrast, for high r POx, the operation range is very narrow and close to the soot formation and stoichiometric limits. Hence, low r POx are advantageous.

4.2 Steady-State Reforming Measurements

The reforming operation was investigated in terms of maximizing the system efficiency and minimizing HC and CH4 emission while complying with the maximum temperature limit. For the efficiency evaluation, two parameters were assessed. The reductant formation yield Y Red (due to low activity, HC is not considered for this evaluation) is most relevant for LNT regeneration. The ratio of the lower heating values of the products and the injected diesel fuel (heating value ratio (HVR)) is only relevant for heating up processes, for example for DPF regeneration or LNT heat-up but less relevant for LNT regeneration (Eq. 11). The C-equivalent HC conversion is calculated by Eq. 12.

$$ HVR=\frac{H_{H_2}\cdot {\overset{\cdot }{m}}_{H_{2,}\mathrm{R}\mathrm{e}f}+{H}_{CO}\cdot {\overset{\cdot }{m}}_{CO,\mathrm{R}\mathrm{e}f}+{H}_{HC}\cdot {\overset{\cdot }{m}}_{HC,\operatorname{Ref}}}{H_{Diesel}\cdot {\overset{\cdot }{m}}_{Diesel}} $$
(11)
$$ {\eta}_{HC}=1-\frac{x\cdot {\overset{\cdot }{n}}_{H{C}_{X,Ref}}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(12)

The influence of λ Ref on the reformate gas composition, temperature, and the efficiency indicators is shown in Fig. 5 (as example for λ Engine = 2.05, r POx = 0.70, r Exhaust = 0.8, SVRef = 9,000 h−1, and T Exhaust ≈ 200 °C).

Fig. 5
figure 5

Measured reformate gas composition and reforming catalyst temperature as function of λ Ref (λ Engine = 2.05, r POx = 0.70, r Exhaust = 0.8, SVRef = 9,000 h−1, T Exhaust ≈ 200 °C)

Full size image

With increasing λ Ref, the HC conversion is steadily increasing and reaches full conversion for λ Ref = 0.55. Maximum λ Ref is defined by the maximum temperature. The methane emission for low λ Ref can be explained either by incomplete HC conversion or by the methanation reaction due to low temperatures. The reductant yield reaches a maximum at λ Ref = 0.45. Y Red decreases for lower λ Ref because of incomplete HC conversion and for higher λ Ref due to total oxidation reactions to the products H2O and CO2. For this example, the optimum λ Ref is 0.525 as best compromise of minimum non-methane hydrocarbons (NMHC) and CH4 emissions and maximum Y Red while complying with the maximum temperature limit.

The influence of r POx on the reformate gas composition, temperature, and the efficiency indicators is shown in Fig. 6 (as example for r Exhaust = 0.8, SVRef = 9,000 h−1, and T Exhaust ≈ 200 °C). λ Engine is variable. λ Ref needs to be adapted to keep the other boundary conditions constant and the HC emission low.

Fig. 6
figure 6

Measured reformate gas composition and reforming catalyst temperature as function of r POx (λ Engine variable, λ Ref variable, r Exhaust = 0.8, SVRef = 9,000 h−1, T Exhaust ≈ 200 °C)

Full size image

For high λ Engine and hence high r POx with predominantly exothermic conditions, λ Ref needs to be low in order to comply with the maximum temperature. However, this results in increased HC emissions. For lower λ Engine and hence lower r POx, λ Ref needs to be enleaned to avoid reformer cool down and HC emission. For low r POx, even for lower temperatures, the HC emission is zero. The optimum operation point with a maximum Y Red is r POx = 0.70. Y Red decreases for lower r POx because of increased λ Ref and for higher r POx due to incomplete HC conversion. For very low r POx with high H2O concentrations, H2 increases caused by WGS on the reformer outlet.

Finally, the influence of the reformer space velocity SVRef on the reformate gas composition, temperature, and the efficiency indicators is shown in Fig. 7 (as example for r POx = 0.70 and T Exhaust ≈ 200 °C). λ Ref and r Exhaust need to be adapted to keep the other boundary conditions constant and the HC emission low.

Fig. 7
figure 7

Measured reformate gas composition and reforming catalyst temperature as function of SVRef (λ Engine variable, λ Ref variable, r Exhaust variable, r POx = 0.70, T Exhaust ≈ 200 °C)

Full size image

The relative heat losses increase with decreasing SVRef. To avoid lower reformer temperature and HC emissions, λ Ref needs to be increased. This results in lower Y Red. At high SVRef, HC emissions cannot be avoided for this operation point (r POx = 0.70). The higher temperature difference between substrate inlet and outlet is an indication for different axial reaction zones (the fast CPOx occurs on the inlet, the slower SR and DR occur on the outlet). To keep the temperature of the substrate inlet below the limit, low λ Ref is necessary and results in HC emissions. For lower r POx (not shown), higher SVRef can be managed without HC emission. For this operation point, the optimum SVRef is ≈10,000 h−1 with λ Ref = 0.5. This represents a thermal power of 10 kWth (considering the lower heating value of the injected diesel).

Summarizing, for all optimized reforming operation points, the reformate gas composition and the efficiency indicators are shown as function of λ Ref in Fig. 8. The left diagram shows a comparison of the H2 and CO molar concentrations achieved under steady-state conditions with those attainable at stoichiometric conditions. The concentrations are close to the maximum yield and demonstrate high fuel conversion. In addition, the rich exhaust gas composition of a rich diesel engine operation is shown. Compared to the rich diesel engine operation, substantially higher concentrations can be achieved by the reformer. The right diagram shows a comparison of the achieved HVR as well as the reductant yield Y Red in contrast to corresponding theoretically attainable pure CPOx values. At lean engine operation and consequently predominantly exothermic reactant composition, due to thermal reasons, a low λ Ref must be set. Due to the high r POx for these conditions, HVR and Y Red are close to the CPOx value. With decreasing engine lambda and consequently increasing endothermic reactant composition, λ Ref has, on the one hand, to be enleaned to support endothermic reaction enthalpy. On the other hand, it offers high reformate quality. In comparison to pure CPOx, exhaust gas reforming shows significant benefits for HVR and Y Red due to the partly conversion of thermal energy of the exothermic CPOx into chemical energy by the endothermic reactions. This helps reducing thermal stress and aging. However, HVR is reached due to partial oxidation of fuel with molecular oxygen a maximum of 85 % at λ Ref = 0.35. Consequently, for DPF regeneration in comparison with late engine internal injection or engine external fuel injection with ideal HVR = 1, external fuel reforming does not represent an alternative.

Fig. 8
figure 8

Reformer output as function of reformer air/fuel ratio for optimized reformer operation: a Measured and theoretical H2 and CO molar concentrations in reformate. b Measured and theoretical CPOx values for ratio of heating values (HVR) and reductant yield (Y Red, only H2 + CO considered) of the reformer, engine air/fuel ratio of the exhaust gas supply to the reformer

Full size image

A final comparison of the reductant yield Y Red for the exhaust gas fuel reformer and an engine-rich operation is shown in Fig. 9. Y Red for the engine operation in case of non-available measurement of all exhaust gas species is calculated by Eq. 13. The theoretical optimum reductant yield is represented by the direct use of fuel for the LNT regeneration. In this case, the reductant yield is 100 %. However, this approach is usually hardly practicable.

Fig. 9
figure 9

Reductant formation yield of additional fuel for LNT enrichment by fuel reforming in comparison to Diesel engine enrichment of various engine calibrations: a as function of A/F ratio of the engine during lean operation and b as function of the engine’s brake mean effective pressure (BMEP)

Full size image
$$ {Y}_{\mathrm{Red},\mathrm{engine}}=\frac{{\overset{\cdot }{m}}_{\mathrm{engine}\kern0.5em \mathrm{fuel},\kern0.5em \mathrm{rich}\kern0.5em \mathrm{mode}}\cdot \left(1-{\lambda}_{\mathrm{engine},\kern0.5em \mathrm{rich}\kern0.5em \mathrm{mode}}\right)}{{\overset{\cdot }{m}}_{\mathrm{engine}\kern0.5em \mathrm{fuel},\kern0.5em \mathrm{rich}\kern0.5em \mathrm{mode}}-{\overset{\cdot }{m}}_{\mathrm{engine}\kern0.5em \mathrm{fuel},\kern0.5em lean\kern0.5em \mathrm{mode}}} $$
(13)

The conventional LNT regeneration done by engine internal enrichment is mainly achieved by throttling and hence intake oxygen reduction as well as applying early and late post-injections for excess HC, CO, and H2 concentrations. However, the main portion of the additionally injected fuel is consumed for compensating combustion efficiency losses caused by engine measures to reduce the typical lean diesel air-to-fuel ratio towards λ = 1. As shown in Fig. 9, only 10–45 % of the additionally expended fuel is available as reducing agent in the exhaust gas. The scatter band in Fig. 9 includes various current diesel engine calibrations. The reductant yield is slightly increasing with increasing BMEP or decreasing A/F ratio during normal lean operation mode. In contrast to the diesel engine enrichment, the reductant yield of the fuel reformer reaches a significantly higher level of about 50–60 %, as shown in Fig. 9. The exhaust gas fuel reformer operation was optimized in terms of low HC emission and high efficiency. The reductant loss in the reformer is caused by the partial oxidation of the fuel with oxygen from lean exhaust gas or secondary air. Nevertheless, the reformer shows significantly higher efficiency in comparison to the conventional approach by engine internal-rich operation. In contrast to the engine-rich operation, the reformer operation is also possible for zero engine loads (idle, overrun). The final fuel consumption penalty in the LNT system in the vehicle is strongly dependent on the LNT operation strategy and NOX emission and is described in detail in a previous paper [19].

The HC emissions are summarized in Fig. 10. The left diagram shows the O-equivalent HC share in the sum of all reductants r HC, O-equivalent (Eq. 14) by the exhaust gas fuel reformer in comparison to diesel engine-rich operation. This parameter is based on HC species C3H6 due to calibration of the FID by C3H6. The right diagram shows the C-equivalent NMHC and CH4 yields (Eqs. 15 and 16) achieved under steady-state reforming conditions.

Fig. 10
figure 10

HC emissions in rich feed gas for LNT regeneration by fuel reforming in comparison to diesel engine enrichment of various engine calibrations: a O-equivalent HC share in total reductants as function of the engine’s brake mean effective pressure (BMEP). b NMHC and CH4 yield as function of λ Ref

Full size image
$$ {r}_{\mathrm{HC},O\hbox{-} \mathrm{equivalent}}=\frac{9\cdot {\overset{\cdot }{n}}_{C_3{H}_6}}{9\cdot {\overset{\cdot }{n}}_{C_3{H}_6}+{\overset{\cdot }{n}}_{H_2}+{\overset{\cdot }{n}}_{CO}} $$
(14)
$$ {Y}_{C{H}_4}=\frac{{\overset{\cdot }{n}}_{C{H}_4}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(15)
$$ {Y}_{NMHC}=\frac{x\cdot {\overset{\cdot }{n}}_{NMH{C}_x}}{12.93\cdot {\overset{\cdot }{n}}_{Diesel}} $$
(16)

For the investigated engine-rich calibrations, the HC share in the reductants was 15–40 % for low engine loads and up to 55 % for high engine loads. The reformer shows relevant HC emissions up to 12 % only during low engine loads with high r POx and hence low λ Ref setting. The HC emissions are strongly dependent on λ Ref as summarized in the right diagram by Y NMHC and Y CH4. Above λ Ref = 0.5, the HC yields are below 1 %. For optimized reformer control at low λ Ref close to λ Ref,min, for example at conditions with high λ Engine and hence high r POx, HC yields for CH4 and NMHC of 5–7 % have been achieved.

4.3 WGS Measurements

In addition to the steady-state reforming investigations, the potential of a water gas shift unit positioned downstream the reformer was evaluated. The general intention is the minimization of the CO share in the reformate gas and increasing the H2 share due to the lower activity of CO, for example for the LNT regeneration.

The steady-state potential of a passively operated WGS, i.e., without H2O dosing, is strongly dependent on the reforming conditions, especially on the H2O concentration in the reformate gas. For high λ Engine and hence high r POx, the H2O concentration in the reformate gas is very low. Under these conditions, almost no WGS is possible. With decreasing r POx and increasing H2O concentration, steady-state WGS potential increases and reaches theoretically, for example 85 % at r POx = 0.45 (85 % of CO after the reforming can be shifted to H2, 82 % were measured). For transient operation conditions, for example in an after treatment application, the Pt/CeO2-based WGS shows further disadvantage. During reforming start, H2 and CO get converted to H2O and CO2 by reducing the oxygen storage of WGS (due to CeO2). This represents a considerable loss in the overall efficiency. On the opposite side, one advantage of this oxygen function on the WGS could be the conversion of a potential HC peak in the reformate gas during reformer start.

5 Summary and Conclusions

For the application area of LNT operation as part of a diesel exhaust after treatment system, a fuel reformer was designed and theoretically and experimentally investigated. As the most suitable design, a coaxial design with a ring-shaped reformer catalyst was chosen. The reformer is positioned outline to enable a very flexible reformer operation. The fuel reforming concept is predominantly based on engine exhaust gas to avoid complex system architectures and to take advantage of the “free of charge” existing exhaust gas.

By the theoretical and experimental investigations, the strong dependence of the reformer operation on the exhaust gas condition, i.e., of the O2, H2O, and CO2 concentrations, could be demonstrated. With steady-state measurements, optimum reformer control settings could be defined in order to find both high system efficiency and low HC slip. In comparison to pure CPOx, exhaust gas reforming shows high benefits in terms of efficiency and thermal behavior, especially with regard to the critical maximum temperature. It is caused by partly conversion of thermal energy of the exothermic CPOx into chemical energy by the endothermic reforming reactions. This helps reducing thermal stress and aging.

In comparison to conventional engine-rich operation for LNT regeneration, the reformer shows significant advantages in terms of system efficiency and HC share in the reductants. With the reformer, a reductant yield of above 50 % could be achieved, whereas the engine-rich operation shows 10–25 % up to medium engine load and up to 45 % for high engine load.

A passively operated water gas shift unit, i.e., without secondary H2O supply, positioned downstream the reformer was experimentally tested to reduce the CO share in the reformate gas. However, a WGS benefit could only be achieved for low λ Engine with thereupon high H2O concentrations in the reformate gas. In transient operation with discontinuous lean and rich operation, the WGS is not suitable due to reformate oxidation by the oxygen storage capacity.

Summarizing, the outline positioned, coaxial designed exhaust gas reforming concept shows beneficial performance in steady-state operation. However, highly transient operation in a vehicle and especially the system cold start need to be further investigated and optimized.