The present work focuses on the further development of the reverse-fow reactor(RFR) concept for the highly endothermic methane steam reforming (MSR) reaction proposed by B.Glöckleret al.in【1,2】. TheRFRis a heat integrated reactor concept that is characterized by the use of the thermal mass of the catalytic packed bed as energy reservoir. In this kind of reactor, regenerative heat recovery is achieved by periodically reversing the direction of the flow. Hence,the energy stored in the outlet section of the packed bed during one period is used to preheat the feed stream during the subsequent period.
This type of operation has been traditionally considered to autothermally run slightly exothermic reactions,e.g. the catalytic purification of industrial exhaust air. Moreover, the excellent heat exchange between the packing and the gas phase offers a huge potential to energetically couple endothermic and exothermic reactions. This is the case of the RFR studied by Glöckler, where the endothermic MSRis coupled with the combustion of a fuel in a regenerative,asymmetric operation mode. This means that the composition of the feed to run the endothermic(production)or the exothermic(regeneration)reactions is switched with every flow reversal.
The reactor is started-up preheating the adiabatic packed bed to temperatures close to 1000℃C,which are required in order to obtain large methane conversions. The MSR reac- tion takes place during the production period as a self-sharpening endothermic reaction and temperature front that travels in the direction of the flow and consumes the energy stored in the packed bed. Large conversions are reached as long as the front remains in the active packing. When its end is reached, the production step is interrupted and the packed bed is reheated during the subsequent regeneration step.
In order to reheat the packed bed,the direction of the flow is reversed so that regenerative heat integration between production and regeneration is possible. Full heat integration can be ideally achieved wheneverthe heat fuxes in both steps are identical. During regeneration, the energy consumed by the reforming reaction is supplied to the packed bed through a combustion reaction. As already shown by Glöcklerin【1】, heat must be released at several positions distributed along the packed bed in order to attain a uniform temperature level. Therefore, the fuel is diluted in a carrier gas that flows through the reactor, while air is introduced through several injection ports, which are confined in inert sections embedded in the active packed bed. The air,the fuel and the carrier gas should mix within the inert sections and react in the catalytic packing downstream. A fraction of the total fuel fed to the reactor is combusted at each of these sections and the energy required to reheat de packed bed is released.
The direct energy release in the packed bed represents an important advantage in comparison to the conventional reforming technology, which generally consists of an externally heated multitubular reactor introduced in radiation furnace. Furthermore, the possibility to take advantage of regenerative energy integration makes it possible to reach large thermal efficiencies, increasing the potential of the reactor concept for the decentralized hydrogen production.
The first experimental validation of the reactor concept was published by Glockler et al.in [2]. These results, taken as staning point for the current study, describe the occurrence of excess temperatures due to the spontaneous combustion of the fuel at the air injection ports. These temperatures damaged the air injectors and the packing itself and are the cause of an irregular temperature profile at the end of the regeneration step. Thus, the present work focuses in the development of an alternative regeneration strategy, which avoids the problematic mentioned above.
The approach followed inspires in the so called flameless oxidation principle (FLOX®),which is a proven technology applied in industrial burners. These burners are generally operated with excess air and are conceived in order to ensure a homogeneous mixing of the air, the fuel and the combustion off-gases. At sufficiently high temperatures in the combustion chamber, the combustion reaction takes place homogeneously and without formation of visible flames. The proper backmixing of the combustion off-gases with the air and fuel introduced to the combustor is a decisive aspect to attain the flameless combustion regime without occurrence of excess temperatures. According to the literature available, this backmixing can be characterized by the so called recirculation ratio Ky, which represents a relation of the recirculated combustion off-gases and the reactants and should exhibit values above 2.
Adaption of this principle to the RFR implies that the inert sections previously described must be replaced by void sections that are run as flameless combustion chambers during the regeneration step. The design of the chambers must ensure a homogeneous mixing of the air, the fuel and the combustion off-gases. In order to design a chamber that fulfills these requirements, several methodologies have been applied: the flow pattern in the chamber, the mixing behavior of the gases and the chamber performance under reacting conditions has been analyzed based on CFD simulations, visualization experiments using tracer methods and combustion experiments in a mock-up of the combustion chamber.
The concept described above, in which the fuel is supplied diluted in an inert carrier gas, is taken as reference for the development of the combustion chamber. The chamber is operated under fuel-rich conditions（λ<1）and the air is injected in such a way that an intense recirculation of the gases in the chamber is achieved. Therefore, the required geometry and alignment of the air injectors as well as the propitious relation between the air and the carrier gas flows are analyzed. This is done based on CFD simulations and visually,using a mock-up of the chamber that is run with water considering the Reynold's similarity of the flows. These studies are followed by combustion experiments in a quartz glass reactor that enable the evaluation of the temperature distribution in the chamber under reacting conditions.In the resulting design of the chamber, the air is introduced through a nozzle head that is radially centered at the outlet of the chamber in the flow direction. Thus, the airis injected in opposite direction to that of the main flow,consisting of the fuel diluted in the carrier gas.
The observations deriving from the investigations mentioned above are consistent and suggest that the design of the air injection nozzles as well as a large ratio between the flow fed through the air nozzles(side stream)and the main stream are decisive parameters to establish flameless combustion regimes. If large side-to-main flow ratios cannot be main- tained during reverse-flow operation, formation of flames may occur. However, the flames are locally confined within the void section of the chamber,so that the resulting temperature peaks do not represent a risk for the stability of the catalyst or the construction materials.
The ignition behavior of the chamber is also an important aspect to be considered under reverse-flow operation. Since the packed bed and the combustionchambers are cooled down during the production step, their temperature at the beginning of the regeneration step is below the ignition temperature of the homogeneous combustion reactions. Thus, the air and the fuel are homogeneously mixed in the chamber and react in the catalytic packing downstream. The combustion reaction takes place in the catalyst until the temperaturein the chamber exceeds the ignition temperature of the gas mixture and homogeneous combustion occurs. According to this description of the ignition behavior,a considerable ignition delay or induction time must be accounted for. In spite thereof,given the homogeneity of the mixture containing the fuel and the combustion air,no excess temperatures occur during his process. Moreover,a controlled,uniform combustion with almost constant adiabatic temperature increase can be run also under dynamic conditions.
The operation of the combustion chamber under fuel-rich conditions exhibits a strong sensitivity towards the fuelif light hydrocarbons are used. Experimental results corroborate that partial oxidation takes place.Part of the fuel reacts at the inlet zone of the active packing causing a temperature increase that is followed by a temperature decrease caused by the cooling effect of subsequent endothermic reactions that the unreacted fuel undergoes in the active packing downstream. This observed behavior helps to describe the iregular temperature profles at the end of the regeneration step when using methane containing fuels,as in the experimental validation of the concept reported by Glöckler.
With the aim to run the RFR in a decentralized hydrogen production facility accounting for energy and material integration, the off-gases of a pressure swing adsorption (PSA) for the purification of the produced hydrogen are to be used as fuel during regeneration. Since this gas contains significant amounts of methane,an alternative regeneration strategy that accounts for the suppression of the methane reforming reactions during the reheating step must be developed.A straightforward approach would consist of introducing air in the main flow and injecting the fuel in each of the combustion chambers, so that these are run under air-rich conditions(>1). However, the oxidation of the catalyst could prove disadvantageous during the subsequent production step.Thus,a reheating strategy basedon stoichiometric combustion （λ= 1） in each chamber is pursued.
Based on thisconcept,the reactor is fed with an inert gas,whereas the air and the fuelare supplied stoichiometrically and separated from each other in every combustionchamber.As already described, the proper mixing of the three streamsis decisive to operate thechamber under flameless conditions. Theexperimental determination of the required parameters and the study of the behaviour under combustion conditions is therefore performed in a metallic combustion chamber. In contrast to the chamber studied so far, the current design accouns for an additional lateral fuel supply positioned at half of its height.
The results obtained confirm the advantages deriving from the stoichiometric operating conditions. These advantages refer not only to the successful suppression of undesired endothermic after-reactions during combustion, but also to the simplification of the control strategy of the chamber during operation under flow reversal.Based on the experimental data and on the analysis of the temperature homogeneity and the fuel conversion, it is possible to define an operation region in which large fuel conversions(>90%)and small temperature differences within the chamber can be achieved. This operation region,how- ever, is strongly limited by the concept followed to introduce the fuel separately into the chamber.
The experimental observations are validated using CFD simulations. The results indicate that the dilution of the fuel in the carrier gas plays a decisive role in order to ensure ahomogeneous distribution of the combustion reaction in the gas bulk. Accordingly, flame sup- pression and a homogeneous temperature distribution in the chamber can be rather achieved with the original chamber design,in which the fuel is diluted in the carrier gas,than in the current concept, in which the fuel is introduced as a side-feed. However,as already described and experimentally verified, the formation of flame fronts occur in the void of the chamber, so that no catalyst and internals damage must be accounted for.
The regeneration concept is verified under reverse-flow conditions taking the findings so far into consideration. Three of the described combustion chambers are embedded in a steel reactor conceived for this purpose. The RFR is operated at reforming loads between 2 and 5kWLHvH,.During regeneration,either H, or PSA off-gas can beused as fuel. The fuelis supplied together with combustion air in stoichiometric proportions to each of the chambers based on a control strategy that enables to heat them up to a constant temperature set point.
The operating parameters for reverse-flow operation are adjusted in order to efficiently recover the energy contained in the effluents, while the duration of the production and rege- neration steps is chosen to be equal.The mass flow during each of the steps is adjusted so that the heat flux in both directions is equivalent. This has an effect on the attainable heat re- covery, which can be derived from the temperature difference between the reactor feed and its effluents. Thus,the reactor is operated in such a way that the feed exhibits temperatures around 150℃C, whereas the temperature of the effluents should not exceed 300°C. The results obtained under reverse-flow operation corroborate that an efficient reheating at a high temperature level,avoiding the occurrence of excess temperatures is possible. In ontrast to the experimental results described by Glöckler et al. in 【2】,a uniform temperature profile at about 900℃C can be established at the end of the regeneration step. Even under operating conditions at which large temperature differences in the chamber cannot be successfully avoided, these differences can be immediately smoothened after begin of the production period, given the low heat capacity of the combustion chamber.e performance of the current reactor concept proves to be reproducible and independent f load variations within the design operation range. Conversions above 90% are achieved moderate reforming loads,corresponding to the equilibrium conversion at the conditions at the outlet of the catalytic packing. The behavior of the temperature profiles at the reactor indaries is a good indicator for the satisfactory heat recovery. In spite thereof, thermal efficiencies of only 40 to 50% are obtained,due to the influence of heat losses in the experimental setup and due to an increase of the heat flux in the direction of the flow during both periods. As discussed later on, the increase of the heat flux is dependent on the physical properties of the flow and on the fuel used for regeneration. The proper adjustment of the operation parameters,i.e. the mass flow during production and regeneration,contributes to partially compensate this effect. This is shown with help of detailed simulations, which are the tool chosen to perform a detailed analysis of the system and for the study of conceptual variations of the RFR concept.
The mathematical model developed for this purpose describes the packed bed as a multiphase, plug-flow reactor with axial dispersion. The combustion chambers are modeled as ideally mixed stirred reactors,assuming total conversion during the regeneration step. The simulation toolused is DIANA,developed at the Max Planck Institute for dynamics of complex technical systems.
As already introduced,an optimal heat recovery assumes constant and equal heat fluxes in both flow directions.However,the heat flux over the reactor length increases about 35% with the heat capacity of the gas during the production step as a result of the reforming reaction. On the contrary, the heat flux during the regeneration step increases due to the supply of fuel and combustion air along the reactor up to 15%if hydrogen is used as fuel, and 30%in case that PSA off-gas is used.Accordingly, the heat flux of the reactoreffluents is always larger than the one of the feed, thus limiting the energy recovery.Moreover, the heat fux of the flow during the production step is almost twice the heat flux of the regeneration gas flow. Hence,a regeneration-to-production mass flow ratio of at least 2 is required during reverse-flow operation. Based on the simulation results,the largest thermal efficiencies by full methane conversion can be achieved at mass flow ratios of around2.5. Although the efficiencies attained are beyond 80%,the results suggest that these cannot be further enhanced and that the difference of the production and regeneration heat fluxes cannot be compensated by means of adjustment of the mass flow ratio or the switching frequency between periods.
The simulation tools, however,enable the study of alternative reactor and packed bed configurations. In the current reactor concept,the inlet section during the production stepis filed with active packing,while the outlet zone is filled with inert packing in order to avoid back-conversion when the effluents are cooled down. Although the reactants are preheated in the inlet section,the temperatures in thiszone of the packing are not suficiently high to substantially contribute to the total conversion of the system. Hence,the proposition made by Glöckleret al.in【3】is adopted and the active section at the reactor inlet is replaced by inert packing. If the energy stored in thissection during the regeneration step is sufficient to preheat the reforming reactants to a high temperature,a stationary reactionzone establishes as soon as the hot gases enter the catalytic bed. This reaction zone is characterized by a steep temperature decrease due to the partial conversion of the reforming reactants. Additionally,a steep moving reaction front where the remaining reactants are fully converted travels in the direction of the flow and consumes the energy stored in the active packing. This behavior,which has been experimentally verified and reported by Glöckleret al.【3】,is also described based on detailed simulations. However,the results indicate that the negative effects caused by the heat flux increase in the direction of the flow are intensified with this reactor configuration. Thus, the thermal efficiencies achieved are lower than in case of the reactor with an active inlet section during the production period.
In order to avoid this problematic,the so called symmetric operation switching pattern proposed by M.van Sint Annaland is taken into consideration [4]. As long as the packedd is symmetrically structured, with inert inlet and outlet sections, the reactor can be operated symmetrically. Accordingly, production and regeneration are performed in the same fow direction in a first period, whereas this is repeated in the opposite direction during the subsequent period.An essential advantage of this concept is the possibility to eliminate two of the three combustion chambers, so that the reactor can be run with a single chamber embedded in the middle of the packed bed.
The duration of the different steps must be adjusted in order to store enough energy at the eactor boundaries,so that the feed can be sufficiently preheated during the consecutive production and regeneration steps. This operation concept enables an optimal energy recovery if the dispersion effects in the evolution of the thermal front can be neglected.However, the detailed simulation results show that this is not the case and that the dispersive effects cause the temperature profiles to strongly fatten during operation.
The optimal realization of the concept requires preheating the reactants during the production step to a high temperature,so that the stationary reactionzone described above can be formed. At the same time, the temperature at the outlet of the catalytic zone needs to be kept at a high level, in order to reach high conversions.According to the simulation results, this is only possible at the expense of reduced thermal efficiencies and large effluent temperatures. Nevertheless, the simplification of the reactor design through elimination of two combustion chambers is an aspect that encourages an experimental validation and the further development of the concept.
Based on this analysis and taking into account the attainable thermal efficiencies of the different configurations studied in this work,the RFR concept originally proposedby Glöckler represents the most appropriate reactor configuration. The regeneration strategy developed in this work to overcome the technical limitations of the original concept has been experimentally validated. Furthermore, large thennal efficiencies and conversions can be attained with this concept over a wide operation range. The efficiencies estimated based on detailed simulations are in the same range than those achieved in the optimized state-of-the-art reforming technologies. Thus, it can be stated that the RFR concept originally analyzed is especially suited for the production of hydrogen in decentralized facilities.

Notation IX

Zusammenfassung XV

Abstract XXII

1 Introduction

1.1Autothermal reactor concepts for endothermic syntheses 1

1.2 Motivation and content overview 4

2 Autothermal Reverse-Flow Reformer 7

2.1 Methane steam reforming 7

      2.1.1 Reaction system 8

2.2 Energy efficient reactor concepts for MSR-State of the art 10

2.3 Reverse-flow reformer 12

      2.3.1 Fundamentals of the operation principle 13

      2.3.2 Short cut method for reactor design 20

      2.3.3 Experimental validation and technical limitations 22

2.4 Summary 24

3 Flameless combustion for heat supply in a packed bed 25

3.1 Flameless Oxidation (FLOX) and its conventional applications 25

3.2 Flameless combustion concept in a packed-bed reactor 27

3.3 Suitability analysis and combustion chamber development 30

      3.3.1 CFD-Simulations 31

      3.3.2 Tracer Methods 34

3.4 Combustion experiments in a single chamber reactor 41

      3.4.1 Stationary operation 41

      3.4.2 Ignition and dynamic behaviour 48

      3.4.3 Understoichiometric combustion 57

3.5 Summary 61

4 Reheating strategy 63

4.1 Stoichiometric combustion concept 63

4.2Design constraints and combustion chamber operation range 65

      4.2.1 Operation constraints 65

      4.2.2 Chamber performance within the design operation range 68

4.3 Systematic combustion chamber design based on CFD 74

      4.3.1 Performance indicators 75

      4.3.2 Simulation results-Correlation with performance indicators 78

      4.3.3 Reduced models for the combustion chamber design 82

4.4 Summary 85

5 Regeneration strategy under reverse-flow operation 87

5.1 Experimental setup 87

5.2 Experimental procedure 92

      5.2.1 Start-up procedure 93

      5.2.2 Periodic operation 93

5.3 Experimental results 96

      5.3.1 Performance indicators 96

      5.3.2 RFR behaviour characteristics 97

      5.3.3 Operability and robustness of operation 103

5.4 Summary 112

6 Modelling the RFR 113

6.1 Mathematical model 113

      6.1.1 Packed bed 113

      6.1.2 Combustion chamber 118

6.2 Numerical solution 120

      6.2.1 Model implementation 121

      6.2.2 Simulation control 122

6.3 Performance indicators 123

7 Simulation study 127

7.1 Asymmetric operation 127

      7.1.1 Convective heat flux disproportion 127

      7.1.2 Adjustment of operation parameters 130

7.2 Alternative reactor configurations for asymmetric operation 135

7.3 Symmetric operation 141

      7.3.1 Simulation results 143

7.4 Summary and outlook 146

A Complementary information and experimental parameter sets 151

A.1 Combustion chamber design 151

      A.1.1 Experimental studies 151

      A.1.2 Temperature distribution in the chamber 153

      A.1.3 Simulation studies 154

      A.1.4 Flow rate variation and effect on the real factor 154

A.2 Reverse flow operation 157

B Technical drawings 159

B.1 Elements of the combustion chamber 159

B.2 Elements of the single chamber reactors 160

B.3 Reverse-flow reformer 161

C Model parameters and physical properties 165

C.1 Model and simulation parameters set 165

      C.1.1 Geometric dependant parameters 165

      C.1.2 Physical and thermodynamic data of the bulk phases 166

      C.1.3 Transport properties 168

      C.1.4 Simulation parameters set 169

      C.1.5 Derivation of the combustion chamber control strategy 170

D Parametric studies based on simulation 175

D.1 Asymmetric operation 175

      D.1.1 Operation parameters 175

      D.1.2 Structure of the packed bed 179

D.2 Symmetric operation 181

E Decentralized hydrogen production facility with RFR 185

E.1 Process description 185

E.2 Relevant process parameters 188

中科院文献情报中心