• No results found

Pyrolysis of wood powder and gasification of wood-derived char

N/A
N/A
Protected

Academic year: 2021

Share "Pyrolysis of wood powder and gasification of wood-derived char"

Copied!
171
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Pyrolysis of wood powder and gasification of wood-derived

char

Citation for published version (APA):

Guo, J. (2004). Pyrolysis of wood powder and gasification of wood-derived char. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR577018

DOI:

10.6100/IR577018

Document status and date: Published: 01/01/2004 Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

providing details and we will investigate your claim.

(2)

WOOD-DERIVED CHAR

By

Jieheng Guo

(3)

Omslagontwerp: Paul Verspaget, Jieheng Guo Druk: Universiteitsdrukkerij, TUE

All rights reserved

No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without the permission of the copyright owner.

CIP-DATA LIBRARY TECHNISCHE UNIVERSITEIT EINDHOVEN Jieheng Guo

Pyrolysis of Wood Powder and Gasification of Wood-derived Char / by J.Guo.

Eindhoven: Technische Universiteit Eindhoven, 2004. -Proefschrift. - ISBN 90-386-1935-9

NUR 961

Trefw.: biomass, houtdeeltjes, houtskooldeeltjes, pyrolyse, vergassing, reactiekinetiek.

Subject headings: biomass, wood powder, char, pyrolysis, gasification, reaction kinetics.

(4)

WOOD-DERIVED CHAR

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de Rector Magnificus, prof.dr. R.A. van Santen, voor een

commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op

donderdag 17 juni 2004 om 16.00 uur

door

JIEHENG GUO

(5)

prof.dr.ir. M.E.H. van Dongen en

prof.dr. W.R. Rutgers

Copromoter: dr. A. Veefkind

The work presented in this thesis has been co-sponsored by The Centre Technology for Sustainable Development (TDO) of the Eindhoven University of Technology (TU/e) and the EU project NNE5-2001-00639. It is carried out within the framework of the J. M. Burgerscentrum (JMBC), Research School for Fluid Mechanics.

(6)

Table of Contents v

1 Introduction 1

1.1 Why biomass gasification . . . 1

1.2 Basic principles . . . 2

1.3 State-of-the-art . . . 4

1.4 Thesis overview . . . 5

2 Materials and experimental methods 7 2.1 Raw biomass . . . 7

2.2 Chars . . . 10

2.2.1 Set-up and preparation procedure . . . 10

2.3 T GA experiments . . . 11

2.3.1 Set-up . . . 11

2.3.2 Analysis procedure . . . 12

2.4 Grid reactor experiments . . . 12

2.4.1 Configuration of the grid reactor . . . 13

2.4.2 IR laser light absorption diagnostics . . . 19

2.4.3 Data acquisition and processing . . . 22

2.4.4 Measurement procedure . . . 23

2.4.5 Temperature measurement . . . 24

2.5 Shock tube technique . . . 29 v

(7)

2.5.2 Conclusions from shock tube theory . . . 31 2.5.3 Diagnostics . . . 33 3 Characterization of chars 43 3.1 Introduction . . . 43 3.2 Char samples . . . 43 3.3 Morphology of char . . . 44

3.4 Physical adsorption of char . . . 45

3.4.1 Principle . . . 45

3.4.2 Apparatus and its analysis procedure . . . 46

3.5 Experimental Results . . . 48

3.6 Theoretical treatment . . . 49

3.7 Conclusions . . . 52

4 Fast pyrolysis of biomass at high temperature 53 4.1 Introduction . . . 53

4.2 Pyrolysis characteristics of several biomasses . . . 54

4.3 High temperature pyrolysis in a shock tube reactor . . . 58

4.3.1 Trajectories of wood particles in the shock tube . . . 59

4.3.2 Heat transfer assessment . . . 66

4.3.3 Kinetics assessment . . . 69

4.3.4 Experimental results . . . 71

4.4 Conclusions . . . 75

5 Gasification model 77 5.1 Introduction . . . 77

5.2 Pore structure of char . . . 78

5.3 C − CO2 gasification mechanism . . . 79

5.4 Diffusive flow in chars: Dusty gas model . . . 80

5.5 Model equations . . . 83 vi

(8)

5.6 Model predictions . . . 87

5.6.1 Binary case . . . 88

5.6.2 Ternary case . . . 92

5.7 Conclusions . . . 96

6 Gasification of the Lignocel-derived chars 99 6.1 Introduction . . . 99

6.1.1 Review of literature . . . 100

6.2 Experimental scheme . . . 102

6.3 Experimental results . . . 103

6.3.1 Formation of nickel carbonyl on the reactor wall . . . 103

6.3.2 The effect of pyrolysis conditions on the char reactivity . . . 104

6.3.3 Gasification kinetics . . . 106

6.4 Conclusions . . . 113

7 Concluding remarks and future work 115 7.1 Concluding remarks . . . 115

7.1.1 Fast pyrolysis in the shock tube reactor . . . 115

7.1.2 Gasification in the grid reactor . . . 116

7.1.3 Gasification modelling . . . 117

7.2 Future work . . . 118

A Property data bank 121 B Temperature measurement by thermocouple 125 B.1 Evaluation of heat flow rates . . . 125

B.2 Heat transfer coefficients and thermal conductivity . . . 129

C Photo-detector response linearity 133

D T GA and DT G curves of biomass 135

(9)

regime 137

F Internal gas flow model 139

F.1 Structure of porous wood particle . . . 139

F.2 Model equations . . . 140

F.3 Results and conclusions . . . 141

F.3.1 Pressure buildup . . . 142

F.3.2 Gas residence time . . . 142

F.4 Appendix: Evaluation of R0 s and C . . . 143 Bibliography 145 Summary 155 Samenvatting 157 Acknowledgement 159 Curriculum Vitae 161 viii

(10)

Introduction

1.1

Why biomass gasification

Concerning the depletion of fossil fuels worldwide and the increasing environmental pollution, numerous endeavors have been attempted to find other renewable and environmental friendly energy sources and to advance the technologies. As to the power generation, biomass gasification technology becomes one of the most promis-ing technologies in the last two decades. Large scale application has two major aspects.

First, biomass is a major source of energy for mankind and is presently estimated to contribute of the order of 10%-14% of the world’s power supply[1]. Basically, biomass is an organic material, which includes plant, wood, crop residues, solid waste, animal waste, sewage, and waste from food processing etc. It offers a number of distinct advantages over other fossil fuels, in particular coal[2]. Biomass typically possesses a higher hydrogen content and a larger volatile component and produces a more reactive char after devolatilization. It contains lower ash and sulfur con-tents. Additionally, biomass, when grown and converted in a closed-loop feedstock production scheme, generates no net carbon dioxide emissions, thereby claiming a neutral position in the build-up of atmospheric greenhouse gases.

Second, gasification technology is an attractive route for the production of fuel gases from biomass. By gasification, solid biomass is converted into a combustible gas mixture normally called “Producer Gas” consisting primarily of hydrogen (H2) and carbon monoxide (CO), with lesser amounts of carbon dioxide (CO2), water (H2O), methane (CH4), higher hydrocarbons (CxHy), nitrogen (N2) and particu-lates. The gasification is carried out at elevated temperatures, 800 K-1700 K, and at atmospheric or elevated pressures. The process involves conversion of biomass,

(11)

which is carried out in absence of air or with less air than the stoichiometric re-quirement of air for complete combustion. Partial combustion produces CO as well as H2 which are both combustible gases. Solid biomass fuels, which are usually in-convenient and have low efficiency of utilization can thus be converted into gaseous fuel. The energy in producer gas is 70%-80% percent of the energy originally stored in the biomass[3]. The producer gas can serve in different ways: it can be burned directly to produce heat or used as a fuel for gas engines and gas turbines to generate electricity; in addition, it can also be used as a feedstock (syngas) in the production of chemicals e.g. methanol. The diversified applications of the producer gas make the gasification technology very attractive.

1.2

Basic principles

A variety of biomass gasifiers has been developed. They can be grouped into four major classifications[1, 4, 5]: fixed-bed updraft or counter current gasifier, fixed-bed downdraft or co-current gasifier, bubbling fluidized-bed and circulating fluidized bed. Differentiation is based on the means of supporting the biomass in the reactor vessel, the direction of flow of both the biomass and oxidant, and the way heat is supplied to the reactor. The processes occurring in any gasifier include drying, pyrolysis, reduction, and oxidation. The unique feature of the updraft gasifier is the sequential occurrence of these processes: they are separated spatially and therefore temporally. For this reason, the operation of an updraft gasifier will be used to illustrate the four processes. In Fig. 1.1 the reaction zones in an updraft gasifier

(12)

are depicted. Biomass and air are fed in an opposite direction. In the highest zone, biomass is heated up and releases its moisture. In the pyrolysis zone, biomass undergoes a further increase in temperature and decomposes into hydrocarbons, gas products and char in the temperature range of 423 K-773 K[7]. The major reactions are given as follows:

Biomass−→ Cheat xHy+ CxHyOz+ H2O + CO2+ CO + H2+ etc.

The hydrocarbon fraction consists of methane to heavy tars (C1-C36 components). The composition of this fraction can be influenced by many parameters, such as particle size of the biomass, temperature, pressure, heating rate, residence time, and catalysts[7]. The obtained char further reacts with the gas stream issuing from the oxidation zone in the reduction zone. Several important reactions occurring in this zone are listed in Table 1.1. The first two reactions in Table 1.1 are usually

Table 1.1: Reactions occur in the oxidation zone. C + CO2 −→ 2CO 14H0 = 172 kJmol−1 C + H2O −→ H2+ CO 4H0= 88 kJmol−1 C + 2H2O −→ H2+ CO2 4H0= 130 kJmol−1 C + 2H2−→ CH4 4H0= −71 kJmol−1 CO + H2O −→ CO2+ H2 4H0= −42 kJmol−1 CO + 3H2−→ CH4+ H2O 4H0= −205 kJmol−1

termed Boudouard and water-gas shift reactions, respectively. Both reactions are highly endothermic and are favorable, kinetically and thermodynamically, at high temperatures. The composition of the producer gases varies widely with the prop-erties of the biomass, the gasifying agent and the process conditions[7]. Depending on the nature of the raw solid feedstock and the process conditions, the char formed from pyrolysis contains 20%-60% of the energy input[8]. Therefore the gasifica-tion of char is an important step for the complete conversion of the solid biomass into gaseous products and for an efficient utilization of the energy in the biomass. The producer gases from the reduction zone rise beyond the reduction zone. When they come into contact with the cooler biomass, the temperature drops down and the aforementioned reactions are frozen. The unreacted char further undergoes the oxidation with air in the lowest zone and

C + O2 −→ CO2 4H0 = −390 kJmol−1 (1.2.1) leaves ash at the bottom of the reactor. The produced CO2 flows upward and is involved in the reactions in the reduction zone. The heat released in the oxidation zone drives both the reduction and pyrolysis processes.

(13)

1.3

State-of-the-art

Gasification technology can achieve a high overall efficiency if it is integrated with the gas cleaning, synthesis gas conversion and turbine power technologies. It is termed ”Integrated Gasification Combined Cycle” (IGCC). Large scale coal-fired IGCC plants have achieved a commercial status, as exemplified by the 250 M W IGCC plant in Polk Power Station, USA, the 235 M W plant at Buggenum, The Netherlands, the 317.7 M W plant at Puertollano, Spain etc. Biomass-fired IGCC technology, however, is still in an early phase of development and demonstration and restricted to small scales. The world’s first biomass-fired IGCC plant is the V¨arnamo plant, which was built between 1991 and 1993. It is until now the only complete and proven IGCC plant based on biomass. It can produce 6 M We2 and 9 M Wth3 from wood chips. Other demonstration plants are still under construction or at the early stage of commercialization such as the ARBRE plant ( 8 M W e) in the UK, the 12 M W e demonstration plant in Italy and the 32 M W e demonstration plant in Brazil etc.

The main constraints to the commercialization of biomass-fired IGCC technology are lack of confidence in the technology, technical and non-technical aspects[9, 10]. The greatest technical challenge for the development of this technology, at all scales, continues to be adequately cleaning the tars and the particulates from the producer gas such that the system operates efficiently and economically. Tar has to be re-moved from the producer gas before entering a gas turbine or an engine. This is not only because of plugging of filters but also for many other reasons: (1) it condenses in exit pipes and plugs them, (2) it is very dangerous because of its carcinogenic character, (3) it contains energy that can be transferred to the flue gas as H2, CO, CH4 etc., and (4) most gas engines and turbines do not accept tar in the incoming gas. Particulates, if not removed beforehand, can cause serious damage to the blades in turbines. Furthermore, there is no standard gasifier, which is able to handle a wide range of fuel types. The quality of the producer gas is affected not only by the types of different gasifiers but also by treatment conditions such as temperature, pressure, hold time, heating rate (which is associated with the nature of biomass, particle size of the feed and temperature, etc.), pyrolysis atmosphere and so forth. Reducing the production of tar and particulates in the gasification demands a thor-ough knowledge on the chemical kinetics of the gasification process. Information about how the reaction rate and product distribution are affected by the temper-ature, pressure, type of reactant gas, flow rate and the biomass properties is quite important. Additionally, in practice, the chemical reactions may be coupled with the transport phenomena e.g. the heat and mass transfer. Information about the

2

M W e means megawatts electric.

3M W

(14)

interaction between the chemical kinetics and the transport phenomena are in turn essential to the optimal operation and a successful design of a gasifier.

The first stage of a logical design procedure requires the availability of reaction rate expressions (e.g. pyrolysis and gasification) that is appropriate for the range of conditions to be investigated in the design analysis. One requires knowledge of the dependence of the reaction rate on composition, temperature, fluid velocity, the characteristic dimensions of any heterogeneous phases present, and any other process variables that may be significant. These information can be obtained by performing bench scale experiment, which are usually designed to operate at constant tempera-ture, under conditions that minimize heat and mass transfer limitations on reaction rates. This facilitates an accurate evaluation of the intrinsic chemical effects. Fur-thermore, in order to predict the gasification performance of the given feedstock and anticipate the technical problems, one has to employ a modelling tool, which accounts for all significant chemical reactions and physical processes. Through an interaction between experimental experience and modelling, the most decisive fac-tors on determining the gasification rate parameters can finally be established. This forms the main scheme of the present work.

1.4

Thesis overview

The objective of this thesis is to provide practical data and a theoretical perspective about the chemical kinetics and the transport phenomena in pyrolysis and gasifica-tion of biomass from both an experimental and a modelling perspective.

First, the basic information about the chemical composition of several biomass ma-terials will be given in Chapter 2. Additionally, the working principles and relevant diagnostics of a number of experimental techniques will be depicted in greater de-tail. They are the shock tube reactor, the grid reactor and the thermogravimetric analyzer. The char preparation procedure will also be specified. In relation to the char gasification in the grid reactor, a char injector enabling depositing the char on a preheated grid and temperature measurement will be addressed. Concerning the importance of the physical properties of the char, the morphology and the pore structure of the char will be independently dealt with in Chapter 3. There, the principle of the physical adsorption method and the characterization results will be presented. Chapter 4 is specially focused on the derivation of chemical kinetics of the high temperature fast pyrolysis process. Chapter 5 is devoted to the modelling of the char gasification with CO2 and CO2/N2. The dusty gas model and the Lang-muir kinetics are incorporated to describe the diffusion of gaseous species and the reaction mechanism, respectively. In Chapter 6 experimental results on the influence

(15)

of the pyrolysis conditions on the char gasification reactivity and the intrinsic ki-netics of the char gasification with CO2are highlighted. The role of diffusion is also validated. Conclusions of the present study, together with some recommendations, are summarized in Chapter 7.

(16)

Materials and experimental

methods

An understanding of the structure and properties of biomass materials is necessary in order to evaluate their utility as feedstocks for pyrolysis and gasification. This chapter first summarizes available information on a variety of such properties in-cluding ultimate analysis, proximate analysis of biomass and chars by using thermal gravimetric analysis (T GA). Subsequently, the char preparation method is elabo-rated in section 2.2. To perform pyrolysis and gasification experiments, both the grid reactor and the shock tube reactor are employed. The grid reactor is used to investigate the char gasification reactivity. The shock tube reactor is mainly used to study the fast pyrolysis of biomass. These two setups were intensively used in the past to study pulverized coal combustion and gasification at high temperatures and high pressures[11, 12]. However, since the biomass properties differ from those of coal, the operating conditions as well as the diagnostics have been optimized to meet the current needs. Therefore, in sections 2.4 and 2.5, the basic principles of the grid reactor and the shock tube are described as well as some improvements of the setup, diagnostic and the new data processing method.

2.1

Raw biomass

In this work a material named Lignocelr HB 120 was intensively used for inves-tigation of the pyrolysis and the gasification. It is a hard wood flour made by a comminution process yielding an average particle size of 120 µm.

The comminution process changes the physical structure of the wood. Typical hard-wood consists of long (l ∼ O(1mm)) hollow fiber tracheids which are connected with

(17)

(a) (b)

Fig. 2.1: The structure of hard wood (a) [2] versus the structure of Lignocel (b).

each other through openings, referred to as pits as shown in Fig 2.1(a). In between these tracheids, large vessels are present, with diameters of 20 to 30µm. During the comminution, the physical structure of the original wood is destroyed. As can be seen in the SEM picture of Lignocel powder (Fig. 2.1(b)), it consists of fragments of the tracheids in which no pore structure in the order of µm is recognizable. During the comminution process, the chemical structure of the wood is retained. Generally, woods can be separated into three fractions: extractables, cell wall components and ash. The extractables, generally present in amounts of 4% to 20%, consist of ma-terial derived from the living cell. The cell wall components, representing the bulk of the cell, are principally the lingin fraction and the total carbohydrate fraction (cellulose and hemicellulose) termed holocellulose. Lignin, the cementing agent for the cellulose fibers, is a complex polymer of phenylpropane. Cellulose is a poly-mer formed from d(+)-glucose while the hemicellulose polypoly-mer is based on other hexose and pentose sugars. In woods the cell wall fraction generally consists of lignin/cellulose in the ratio 43/57[2]. The presence of the organic fraction: lignin, cellulose and hemicellulose, is expected to affect the overall reactivities for pyrolysis and gasification. To have a better understanding, two other materials: microcrys-talline Cellulose and organosolv Lignin are also examined. Organosolv Lignin was obtained from Aldrich Chemical Company, WI, USA (Catalog No.: 37, 101 − 7). It is a polymeric organosolv lignin material isolated from a commercial pulp mill using mixed hardwood (mixture of 50% maple, 35% birch and 15% poplar) as raw material. The microcrystalline cellulose was also obtained from Aldrich Chemical Company, WI, USA (Catalog No.: 43, 523 − 6). It is a purified, partially depolymer-ized cellulose, prepared by treating alpha-cellulose, obtained as a pulp from fibrous

(18)

plant material, with mineral acids. The degree of polymerization is typically less than 400.

The ultimate and proximate analysis of samples as received are presented in Table 2.1. The ultimate analysis generally reports the C, H, N , S and (by difference) O content in the sample. The proximate analysis classifies the sample in terms of moisture (M ), volatile matter (V M ), fixed carbon (F C) and mineral matter(ASH). The volatile matter mainly consists of the organic compounds. The moisture deter-mined by the proximate method represents the water that is bound physically to the matrix structure; water released by chemical reactions during pyrolysis is classified with the volatiles. The ash content is determined by combustion of the volatile and fixed-carbon fractions. The resulting ash fraction is not representative for the ash in the raw material, more appropriately termed mineral matter, due to the oxidation process employed in its determination. The fixed-carbon content is calculated from the material balance. Thus: F C = 1 − M − ASH − V M. The fixed carbon is considered to be a polynuclear aromatic hydrocarbon residue resulting from con-densation reactions which occur in the pyrolysis step. It not only contains carbon but also some other elements such as oxygen, nitrogen and sulphur. To avoid effects of moisture on the pyrolysis reaction, all the samples were kept at 323 K before analysis.

Table 2.1: The ultimate and proximate analysis of all samples as received. Ultimate(wt%, daf)

Cellulose Lignin Lignocel

C 47.23 72.53 53.67 H 5.80 5.43 5.36 N 0.00 0.00 0.00 S 0.00 0.00 0.00 O 46.97 22.04 40.97 Cl 0.00 0.00 0.00 Proximate (wt%)

Cellulose Lignin Lignocel

Moisture 4.30 2.61 9.45

Volatiles (daf1) 84.65 76.66 76.45 Fixed Carbon (daf) 11.05 20.73 13.56

Ash (dry) 0.00 0.00 0.54

(19)

2.2

Chars

2.2.1 Set-up and preparation procedure

The chars are prepared by pyrolyzing Lignocel wood under a continuous N2 flow with a flow rate of 150 ml min−1 in a furnace as shown in Fig. 2.2. About 0.4 g of Lignocel powder was put in a stainless steel sample container. Before the pyrolysis commences, the sample container stays in the cold zone of a quartz tube, where the wood particles can be dried at T0 (typically 423 K) denoted as prior to the pyrolysis process. Note that the drying temperature should not exceed 423 K so that no pyrolysis can take place. The furnace can be heated up to a maximum of 1200 K and can be stabilized at the required temperature under the control of a thermostat. Due to the heat transfer between the quartz tube and the N2 flow, the temperature of the sample can differ from that of the furnace. To check the exact gas temperature inside the quartz tube, a thermocouple is attached to the pipe during N2 flow. When the required temperature Tf T is achieved and stabilized, the sample container is rapidly pushed into the hot zone at time t0, when the pyrolysis of woods starts to take place. The released volatiles can be purged away by the N2 flow so that no secondary reactions occur. Another thermocouple is welded under the sample container. It is used to measure the sample temperature. After a certain residence time, the sample container is moved back to the cold zone at time tf T. There the produced chars are cooled down. Finally, the prepared chars are weighed and stored in glass bottles filled with N2 to avoid oxidation.

(20)

The pyrolysis hold time th is determined by the relation th = tf T − t0. The weight loss of the wood is ml = mw− mc, i.e. the difference between the initial weight of wood and the weight of the char. Thus the conversion ratio can be calculated in the following way,

Xc = ml mw

. (2.2.1)

The heating rate of this process is defined as, βh=

Tf T − T0 tβ

, (2.2.2)

where the parameter tβ is defined as tβ = tf− t0 and tf is the time when the sample temperature reaches Tf T. The conversion ratio is used as an index for evaluating the reproducibility of chars. Chars are taken as identical only if the difference in conversion ratio is less than 3%.

The pore structure of the char is an essential factor in determining the reactivity of the gasification. It can be measured by using a physical adsorption technique. Considering its importance, this part will be independently discussed in Chapter 3.

2.3

T GA

experiments

Thermal gravimetric analysis is a thermal analysis technique used to determine changes in sample weight as a function of temperature. It can also be used to study changes in sample weight as a function of time at a constant temperature (isothermal TGA). The analysis can be performed under nitrogen, in the presence of air, or other reactive atmospheres. Another possible application of T GA is to study the pyrolysis kinetics of the biomass as described elsewhere [13–16]. In this thesis work, it is mainly employed to perform the proximate analysis of both the raw materials and the derived chars. In this section, the configuration and the analysis procedure of T GA are described.

2.3.1 Set-up

Pyrolysis experiments and proximate analysis are performed in a T A Instruments SDT 2960 (Simultaneous Differential Thermal Analyzer) at the group of Thermal Power Engineering, Department of Mechanical Engineering and Marine Technology, Delft University of Technology. Fig. 2.3 shows its configuration. This SDT has horizontal sample carriers and the location of the thermocouple is just below the crucible in the sample carrier. In order to account for buoyancy effects, a correc-tion curve with empty crucibles was first obtained and then subtracted from the

(21)

experimental results. No lids were used on top of the crucibles. Temperature cal-ibration, baseline calibration and weight calibration experiments were done as to each condition, according to the manufacturer-provided manual.

Fig. 2.3: Configuration of the T A Instruments SDT 2960.

2.3.2 Analysis procedure

Similar procedures are used to carry out the pyrolysis and proximate analysis of the samples. As for the proximate analysis, it was carried out under Helium (99.99%) at a constant flow rate of 100 mlmin−1. First, a sample of about 8-10 mg was heated up to 383 K and kept at this temperature for 30 min to remove moisture. Then it was heated to 1173 K at a heating rate of 20 Kmin−1 and kept at that temperature for 30 min before burning the char in air in order to determine the ash content. One example of the proximate analysis for Lignocel is depicted in Fig. 2.4. It shows three steps process: removal of moisture at the temperature below 383 K, devolatilization in the range of 500 K-680 K and combustion of the residue. From the change of the weight in the three steps, M , V M , F C and ASH content can be determined as illustrated in Fig. 2.4. The pyrolysis experiment follows a similar procedure as the proximate analysis, but with variable heating rates and final temperatures.

2.4

Grid reactor experiments

A grid reactor is employed to study the pyrolysis and the gasification of Lignocel wood and its derived char. It was constructed at Eindhoven University of Tech-nology for the purpose of investigating the gasification of coal-derived char at high temperature (up to 1950 K) and high pressure (up to 2.5 M P a). To ensure the oc-currence of an isothermal char gasification experiment, the char sample needs to be

(22)

(a) (b)

Fig. 2.4: An example of the proximate analysis. The weight loss of Lignocel wood as a function of (a) temperature and (b) time.

? 6 6 ? 6 ? 6 M V M F C Ash

fed onto the reactor at constant temperature. To meet this requirement, an injector was installed onto the grid reactor (section 2.4.1). The reaction rate is determined from the CO production. The concentration of CO is measured by means of an infrared absorption method, which will be elaborated in section 2.4.2.

2.4.1 Configuration of the grid reactor

The configuration of the grid reactor is schematically shown in Fig. 2.5. It contains

(a) Top view (b) Side view.

(23)

a closed cylindrical reactor chamber with an inner diameter of 15 mm and a length of 224 mm. In the middle of the reactor a platinum grid is placed and mounted on two supports. These two supports act as electrodes connected to an external power source. The grid has dimensions of 4 mm × 10 mm and consists of interweaved wires as shown in Fig. 2.6. Each weft wire passes alternately over and under each warp wire. Warp and weft wire have generally the same diameter, namely DP t = 0.076 mm. The aperture width Dsp = 0.27 mm. The grid can be heated electrically close to the melting temperature of platinum (2045 K) at atmospheric pressure. The reactant gas is supplied to the reactor through two inlets positioned in the legs of the reactor, symmetrical to the grid. Sintered porous material is

Fig. 2.6: The construction of the grid.

mounted in the tubing at the inlets to provide a homogeneous and gentle gas flow. This construction minimizes the possibility that the feed is blown from the grid when the gas enters.

At the top of the reactor there is a window made of ordinary glass (BK7) for the observation of the grid and the sample on it during the experiments. Moreover, the temperature of the grid can be measured via this window by using a manual color pyrometer. At both sides of the reactor CaF2 windows with good IR transmission properties are mounted. Note that these two windows are not parallel but tilted at a small angle. This particular configuration is chosen to avoid light interference effects between the two windows.

Char injector A char injector2 (Fig. 2.7) has been mounted on the reactor. In this way it is possible to deposit the char directly onto the preheated grid. This con-figuration avoids the problem of having gasification before the sample temperature reaches the required gasification temperature. The char injector consists of a hollow injector tube with a spatula at the end and a piston. Before the experiment, the spatula is set upwards and a small amount of char can be fed into the the injector

(24)

tube and pushed onto the spatula by the piston. When the desired grid temperature is reached, the injector tube is turned so that the chars can fall onto the grid. Then it is pulled back to prevent the spatula from blocking the passage of the infrared

Fig. 2.7: A schematic picture of the char injector.

light. In Fig. 2.8 the pictures of the char particles dropped by the injector are presented in the top and side views. It is seen that the particles are located in the center of the grid and packed up to around 1 mm height. To ensure that this packed particles are gasified isothermally, the following calculation is given with respect to the heat-up process of the packed particles on the hot grid.

(a) Top view (b) Side view

Fig. 2.8: Top view and side view of char particles dropped by the injector. For the gasification experiments, the char particles of diameter Dpc and height Hpc are directly dropped onto the hot grid of height HP t that is held at a constant temperature. On the grid, these particles are packed to a height of HB (see Fig.

(25)

2.9) and react as long as the temperature for gasification is maintained. If the heat-up time of the packed bed is much less than the characteristic gasification time, i.e. τB  τr, we say the gasification is isothermal; otherwise, the temperature

5 4 3 1 2 Hpt (Reference) Grid HB Packed particles

Fig. 2.9: A schematic diagram of a thermal plume.

distribution inside the packed bed should be taken into account in the modelling. To estimate the heat-up time, a better insight will be given to the heat transfer taking place inside the reactor. First, when a hot grid is situated in a cold CO2 environment, a circulating flow exists due to the temperature difference (or buoyancy effect) between the grid and the surrounding gas. Second, since the packed particle bed on the grid is permeable, the gas flow will enhance the heat transfer from the gas to the particles.

The energy balance applied to the bed in the absence of endothermic gasification yields: dTB dt = hlocapc cB (Tg− TB), (2.4.1)

where cB is the specific heat per unit mass of the bed, the parameter apc is the total particle surface per unit mass of bed. The parameter hloc is the heat transfer coefficient between particle and gas in the packed bed and is defined as the local value representative of a cross section through the bed. The characteristic heat-up time of the packed bed associated with Eq. 2.4.1 is then

τB= cB hlocapc

. (2.4.2)

(26)

be estimated according to the following empirical correlation:[17]

jH = 0.91Re−0.51ψ, for Re < 50, (2.4.3) jH = 0.61Re−0.41ψ, for Re > 50, (2.4.4) where the Colburn jH factor and the Reynolds number are defined by

jH = hloc cgρgu0 (cgµg kg )2/3f , (2.4.5) Re = ρgu0 aψµf , (2.4.6)

where cg is the specific heat at constant pressure of the gas, ρg the density of the gas, a the surface area per unit volume of the bed, u0 is the superficial flow velocity. The shape factor ψ is taken as 0.91 for cylinders. In these equations the subscript f denotes properties evaluated at the average temperature of gas and particle surface (0.5(Tps + TB)). The value of jH depends on the flow velocity. The buoyancy-induced flow velocity is estimated by means of an elementary natural convection model. We assume that the temperature below the grid and of some distance aside from the grid are undisturbed with values T2 and so are the gas densities ρ2. Above the grid, the gas temperature is assumed to be T3 and equals the grid temperature. The density is assumed to be ρ3. Regime 3 is assumed to extend to the top wall of the reactor. At steady-state, the mechanical energy balance of the system (see Fig. 2.9) can be described by P1 = P2+1 2ρ2u 2 0+ ρ2gh2, (2.4.7) P3 = P4+ ρ4g(h4− h3), (2.4.8) P4 = P5, (2.4.9) P5 = P1− ρ2gh5. (2.4.10)

The gas flows through the grid and the packed particles due to a pressure drop that can be related to u0 by means of the Ergun equation[17, 18]:

(P2+ ρ2gh2) − (P3+ ρ3gh3) ρ2u20  Ds HB   3B 1 − B  = 150(1 − B) Dsρ2u0/µ2 + 1.75  , (2.4.11) where Ds is the equivalent specific surface diameter defined as Ds = 6Vpc/apc. Combining Eqns.2.4.7-2.4.11 gives an implicit expression for u0:

 0.5ρ2+ 1.75ρ2HB(1 − B) 3 BDs  u20+150µ2HB 3 B  1 − B Ds 2 u0+ (ρ3− ρ2)gh4 = 0, (2.4.12)

(27)

where the subscript B represents the bed and HB the height of the bed.

If τB τr, a steady-state heat transfer analysis can be applied to further estimate the intra-particle temperature difference induced by the endothermic gasification. We consider again a cylindrical particle. At steady state, the heat flow inward by conduction in the radial direction of cylinder must equal the energy consumption by reaction. For one shell of a single particle in the bed, the energy balance is

−4H0DedC dr = −k e p dTp dr , (2.4.13)

where De is the effective diffusivity of the reactant gas, kep the effective thermal conductivity of the particle, 4H0 the enthalpy of gasification. Negative signs are required on the left side of this equation so that for an endothermic reaction, the temperature will be cooler in the core than at the periphery. Integration of the equation between the radius r and the gross particle radius Rp gives

Tp− Tps= 4H 0De ke

p (C − C

ps), (2.4.14)

where Tps and Cps are the temperature and reactant concentration at the external surface of the char particle. The maximum temperature difference between the center of the particle Tpm= Tp(r = 0) and the external surface Tps(r = Rp) occurs when the reactant concentration vanishes at r = 0.

Tpm− Tps = −4H 0DeC ps ke p , = −4H 0DeP RTpskep , (2.4.15)

where P is the total pressure of the CO2 in the reactor.

The property values relevant to the grid and the packed bed are listed in Table 2.2 and those of the char particle can be found in Appendix A. Substitution of

Table 2.2: Property values used in the estimation.

Parameter Value Parameter Value

Tps, K 600 - 1900 h4, m 2 × 10−2

P, P a 1.01 × 105 B 0.6

Dpc, µm 10 CB, kJkg−1K−1 = cp(see AppendixA)

Hpc, µm 50 De, m2s−1 10−7

(28)

the property values into the equations leads to a flow velocity of 0.11 ms−1 and a characteristic heat-up time of the packed particles τB ≈ 21 ms, which is much less than the typical gasification time ranging from a few minutes to hours. The intra-particle temperature difference is less than 1 K. Therefore, it is reasonable to consider the gasification of the char in the grid reactor as an isothermal reaction.

2.4.2 IR laser light absorption diagnostics

During the gasification of Lignocel with CO2, CO molecules are produced, which disperse in the closed volume of the grid reactor. The concentration of CO is time-dependent and can be used to calculate the reaction rate. The time-resolved concentration of CO can be detected by using an IR absorption technique.

To begin with, the principle of the IR absorption technique is described briefly. This technique relies on the fact that molecules absorb light (electromagnetic energy) at spectral regions where the radiated wavelength coincides with internal molecular energy levels. In accordance to well known quantum mechanical theory such energy resonates with interatomic vibrations. At room temperature, the vibrational tran-sition is most often from the ground state to the first excited level. Accordingly, the vibrational spectrum of a diatomic molecule such as CO consists of one line only. For CO it is at 2143 cm−1. In addition to the absorption by the vibration, the rotating oscillator can also absorb corresponding energy quantities exclusively for rotational excitation. Since the energy differences between rotational levels are so much less than between vibrational ones, lines due to rotational transitions appear as fine structure near the frequency of the vibrational transition. In Fig. 2.10 the energy levels and transitions for the vibration-rotation band of CO are depicted together with the resulting spectrum for CO. The absorption spectrum is based on the HIT RAN 96 database. The spectrum consists of approximately equally spaced lines on each side of the center of the band. Transitions to the next higher energy level (∆J = +1), counting from the J value of the lowest vibrational level, belong to the R branch and those with ∆J = −1 belong to the P branch. Transitions corresponding to ∆J = 0 belong to the Q branch. But the Q branch does not occur for gas phase CO at room temperature.

The diagnostic applied in the present work make use of a tunable infrared laser, which scans a very narrow wavelength region. Instead of the entire absorption band the intensity of a single absorption line in the vibration-rotation spectrum is detected and used as a representative of the CO concentration in accordance with

(29)

(a) (b)

Fig. 2.10: Energy-level diagram and the resulting spectrum for CO considered as a harmonic oscillator.

Lambert-Beer’s law3:

I

I0 = exp (−βl[CO]) ,

(2.4.16) where I0is the incident light intensity and I the transmitted intensity. The ratio I/I0 is defined as the transmittance. β in Eq. 2.4.16 is the absorptivity, l the absorption path, i.e. 22.4 cm for the grid reactor. [CO] represents the molar concentration of CO. Rearranging Eq. 2.4.16 yields

[CO] = −βl1 ln I I0



. (2.4.17)

Following the basics about this IR-absorption technique aforementioned, we proceed to the arrangement of the diagnostic system. The diagnostic together with the grid reactor are shown schematically in Fig. 2.11. A laser diode, consisting of a lead salt chip in a gold-plated copper package, is used to generate the infrared radiation. This radiation is reflected by a collimating mirror adjusted in three dimensions with three screw micrometers. Then the reflected beam passes a reference gas cell filled

3In fact, it is the integral of the CO concentration along the pathlength that counts. For a

one-dimensional absorption cell, the CO concentration in Eq. 2.4.16 is the length-averaged one, which is directly proportional to the total amount of CO present in the cell.

(30)

Fig. 2.11: The diagnostics of the grid reactor.

with CO of 933 P a. In front of the gas cell, a thin grating is coated. It splits the incident beam into two parts. One part of the beam passes through the gas cell and is finally projected on a HgCdT e pn detector, which is cooled by liquid nitrogen to suppress thermal noise. The other part of the beam is expanded to a parallel beam with a diameter of 19 mm. Then it passes the grid reactor through two CaF2windows with diameters of 15 mm. So the CO in the entire volume can be irradiated. After passing the reactor, the transmitted light is projected on a second HgCdT e pn detector. The light intensities received by two detectors in terms of the detector voltages are transferred to a data acquisition system4.

The laser diode in our setup is cooled by liquid nitrogen and is tunable by modu-lating the current and adjusting the temperature. The current through the diode determines the wavelength range of the infrared radiation. The modulation current periodically changes the injection current from a value below threshold to another, thus the wavelength of the emitted radiation varies periodically. In this way a time-resolved spectrum can be obtained. The center wavelength of the laser depends on the temperature. Tuning the temperature can shift to another absorption line. Good thermal and electronic stability make sure that the spectrum scan emitted by the diode laser is stable and reproducible. The modulation frequency determines how many scans can be made. The maximum frequency is 20 kHz. It makes this system suitable for a fast process such as the pyrolysis as well as the slow process such as the gasification.

As described before the concentration of CO can be measured from the transmitted light intensity(Eq. 2.4.17). To illustrate the calculation, an example of the trans-mission of CO in one scan is shown in Fig. 2.12. In this figure, the offset intensity is easily determined by blocking the incident light into the detector. The intensity in

4

This data acquisition system and the accompany software (Acquire) were built by Ad Holten, Technische Universiteit Eindhoven.

(31)

the signal maximum is used to determine the transmittance. 4V2b is proportional to the initial light intensity I0 and 4V2a is in the same way proportional to the transmitted light intensity I. Recalling Eq. 2.4.17, the normalized concentration of CO is

[CO] = −βl1 ln(4V2a 4V2b

). (2.4.18)

As for monitoring the gasification rate, we use a normalized concentration C(t). To define this quantity, the concentration of carbon monoxide after all chars is gasified can be taken from:

[CO]∞= − 1 βlln( 4V2a 4V2b )∞. (2.4.19)

The relative concentration can then be expressed as C(t) = [CO] [CO]∞ = ln( 4V2a 4V2b) ln(4V2a 4V2b)∞ . (2.4.20)

For determining 4V2b, the corresponding zero absorption intensity must be deter-mined. To do this, a linear baseline is drawn as recommended by G¨unzler[19]. It can be found by selecting a line drawn through 2 points at each side of the measurement band, i.e. the linear baseline (Fig. 2.12).

1200 1400 1600 1800 2000 2200 0.6 0.65 0.7 0.75 0.8 nsample V 2 (V) ∆V2a ∆V2b Baseline Offset

Fig. 2.12: An example of the CO absorption in one scan.

2.4.3 Data acquisition and processing

The data-acquisition system consists of a computer and 12-bit data acquisition board. A program named Acquire5 has been developed to drive the data acqui-sition system. In this program, the sample frequency, record length, record count and the frequency of a second clock module can be set to record data. During the gasification measurements the sampling frequency was set at 100 kHz. The fre-quency of the laser scan was ∼ 100 Hz, i.e. 1000 samples per scan were taken. To

(32)

Fig. 2.13: A typical time-resolved CO concentration profile.

prevent the collection of too many data, a second clock module is used to make a limited amount of records per period. Its frequency is usually set at 10 Hz at maximum, i.e., 10 records per second. Then the ratio of the total number of records (record count) to this module frequency is then the total measurement time. At the start of each scan, the data-acquisition setup generates a trigger signal. Together with the detector signal, this signal was stored in one file. The data file is used as an input file for a Matlab program, which calculates the time-dependent relative concentration of carbon monoxide. To determine the starting point of the measurement as accurately as possible, a push button was installed near the set-up. When the char is dropped on the grid, the button is pressed. This becomes visible in the trigger signal, which is raised with 0.5 V for 5 seconds. In this way the Matlab program can determine when gasification starts.

A typical time-resolved CO concentration profile is given (Fig. 2.13). The charac-teristic time of gasification, τ90, is defined as the time at which the relative concen-tration of carbon monoxide has reached 90% of the final concenconcen-tration.

2.4.4 Measurement procedure

The gasification experiment starts with preparation of the injector. Char is put in the injector and is moved with the piston towards the spatula. The whole set-up is evacuated till the pressure is below 2 × 10−5bar. Next the reactor is slowly filled with the desired gas up to 1 bar. After that, the external power source is switched on, resulting in the heat-up of the grid. With a thermocouple, the heat-up time of the grid was determined. When the heat-up time has elapsed, an optical pyrometer (Leeds and Northrup R627) is used to determine the grid temperature. The work-ing principles of both the thermocouple and the pyrometer will be described in the

(33)

subsequent section 2.4.5. After these preparations, the data acquisition system is set on the status of ready and waiting for data. By means of the injector, the char is dropped onto the hot grid and the push button is pressed to trigger the data ac-quisition. The injector is pulled backwards immediately to prevent it from blocking the laser light. After a predetermined time (which exceeds the time necessary to complete gasification), the recording of measurement data stops. The measurement is completed.

2.4.5 Temperature measurement

Optical pyrometer

The temperature of the hot grid is measured with an optical pyrometer (Leeds and Northrup R627). This pyrometer operates with nearly monochromatic light. The wavelength is usually a narrow band about 0.01 µm wide at 0.653 µm in the red portion of the visible spectrum. Radiation from the hot platinum grid is focused by the lens onto a screen. The screen is viewed through a red filter glass so that only wavelengths of about 0.653 µm are seen. Inside the pyrometer there is a tungsten wire which can be heated electrically. By making the brightness of the tungsten wire equal to that of the screen image of the grid, the temperature of the grid TP t (K) can be measured. The temperature from the pyrometer Tb (K) is calibrated for black body targets. The lower limit of the temperature range is about 1000 K, determined by the long wave visibility limit of the human eye. When measuring the temperature of the grid, a correction is necessary since it is a gray-body emitter. For a body of spectral emissivity ελ,T, at a temperature, T , and at a wavelength, λ, the monochromatic radiation intensity of the surface Iλ (W/m3· sr) is

Iλ,T = ελ(λ, T )

2hc2λ−5

exp(hc/(κλT )) − 1, (2.4.21) where h = 6.6256 × 10−34Js is Planck’s constant, c is the speed of light propagation taken as 2.998 × 108m/s, κ = 1.3805 × 10−23J/K and ελ(λ, T ) is the emissivity of the body as a function of temperature and wavelength. As a gray-body, the monochromatic emissivity of the grid is independent of the wavelength and is equal to its total emissivity, namely

ελ(λ, T ) ∼= εtot(T ). (2.4.22) Using this relation, the monochromatic radiation intensity of the platinum grid can be correlated to the total emissivity of the platinum grid, εP t,tot. At the moment of reading the measured temperature value, the brightness of the tungsten wire and of

(34)

the grid are equal. If the attenuation of incident radiation of the eyes and the lenses is negligible, the intensity of the wire and of the grid shall satisfy:

Ib(Tb) = IP t(TP t), (2.4.23) In this equation, Tb corresponds to the reading of the pyrometer. Combining Eq. (2.4.23) with Eq. (2.4.21), we obtain

2hc2λ−5 exp(κλThc b) − 1 = εP t,tot( 2hc2λ−5 exp(κλThc P t) − 1 ). (2.4.24)

Rewriting equation (2.4.24) we get Tb = hc κλ ln(1 − ε1P t(1 − exp( hc κλTP t))) . (2.4.25)

By inserting the pairs of TP t and εP t,tot (see Table 2.3) into Eq. (2.4.25), the corresponding Tb can be obtained. Fitting TP t versus Tb by linear regression yields TP t = 1.151 ∗ Tb− 38.136. (2.4.26)

Table 2.3: Total emissivity of unoxidized platinum[20]. Temp.,0C 25 100 500 1000 1500

εP t,tot 0.037 0.047 0.096 0.152 0.191

Thermocouple

The optical pyrometer is limited to the high temperature regime, say T > 1000 K. For the lower temperature regime, we use a type K (Chromel-Alumel) thermocouple with a diameter of 0.2 mm. Its measuring junction is clamped to the grid surface. The cold junction of this thermocouple is kept at the ice point. In order to inves-tigate the accuracy of the thermocouple measurement, a comparison between these two methods has been made in the temperature range of 1100 K and 1400 K. We found that the measured temperature by the thermocouple was ∼ 200 K lower than that measured by the pyrometer. This temperature deviation is attributed to a disturbance of the thermal field of the grid when introducing a thermocouple[21]. A similar problem has been well modelled by Keltner and Beck[22]. In this model, the thermocouple is supposed to be mounted on a thick wall. It is considered as a single semi-infinite cylinder with lateral surface heat loss characterized by a heat transfer

(35)

coefficient, hc, with the ambient at the thermocouple temperature. It is assumed that the heat flow rate and the temperature inside the thermocouple is constant in a cross-section. When the junction is just pressed against the surface of the substrate and has lateral heat loss, its steady-state temperature can be written as

Ttc= TP t 1 + 2K√Bi(B1 +π4), (2.4.27) with K = kT c kP t , B = hptrT c kP t , Bi = hT crT c 2kT c ,

where k is the thermal conductivity, rT c is the thermocouple wire radius, α the thermal diffusivity, B the contact Biot number and Bi the lateral surface Biot modulus. It is found that the accuracy of the measurement is only dependent on the properties of the substrate and the thermocouple. According to Eq. 2.4.27, an estimate of TT cis performed with the parameters listed in Table 2.4. An example of the predicted temperature as a function of the thermocouple temperature is shown

Table 2.4: Parameters used for the estimation of Tgrid in Eq. 2.4.27. Parameters K B Bi

Value 1 1 0.001

in Fig. 2.14. It is shown that the thermocouple technique could give rise to a 100 K difference when the object temperature is over 1000 K.

800 900 1000 1100 1200 1300 1400 1500 800 900 1000 1100 1200 1300 1400 1500 1600

T

tc

, K

T

Pt

, K

Fig. 2.14: Prediction of the grid temperature as a function of the thermocouple temperature according to Eq. 2.4.27.

(36)

rTC Thermocouple Grid z r Tm Hpt

Fig. 2.15: The geometry for a thermocouple attached on the surface of the grid.

It should, however, be kept in mind that Eq. (2.4.27) is generally accurate when the thickness of the substrate is at least ten times the thermocouple diameter. In our case, the thickness of the grid wire has about the same magnitude as the diameter of the thermocouple. Therefore, a modified analysis is quite necessary. In the following, a steady-state thin wall analysis will be presented.

The geometry of the problem is depicted in Fig. 2.15. A thermocouple is in thermal contact with a platinum grid with height HP t over a circular region of radius, rT c. Both thermocouple and grid are considered as homogeneous semi-infinite bodies. To simplify the problem, the following assumptions are made:

• Negligible thermal resistance of the interface: 0 < r < rT c, z = 0.

• The temperature of the thermocouple TT cis approximately constant in a cross-section. Therefore, TT c= TT c(z).

• The temperature of the grid TP tis independent of angular position and height. Therefore, TP t = TP t(r). For 0 < r < rT c, TP t equals the measured tempera-ture Tm.

• The heat transfer coefficients h and the thermal conductivities k are constant. The heat flow rate balance in the isothermal disk (the shaded section in Fig. 2.15) requires that the sum of the heat flow rates from the shaded grid area to the sur-rounding, i.e. thermocouple (Tc), grid (Pt) and environment (∞) equals the electric production of heat:

(37)

Derivation of each heat flow rate has been performed in Appendix B. Here, only the final solution with regard to the measured temperature Tm is given below:

Tm− T∞ Tm− TP t,∞ = − q 8hP tHP tkP t r2 T c K1(rT c0 ) K0(rT c0 ) − 2hP t T∞−TP t,∞ Tm−TP t,∞ hP t+ q 2hT ckT c rT c . (2.4.29)

From this expression, it follows that the measured temperature depends on the physical properties of both the grid and the thermocouple and also on their heat transfer coefficients. In the derivation of Eq. 2.4.29, all these quantities are assumed constant. In reality, this might not be true. Especially, the overall heat transfer coefficient, which must include convection, conduction and radiation, is a function of temperature. In this situation, the above relation is not valid anymore in a wide temperature range. Therefore, the temperature dependence of the heat transfer coefficients has to be investigated in the temperature range of our interest 800 K-1600 K. Also notice that in the above calculation, the grid is taken as a homogeneous solid body. However, it in fact consists of weaved wires as depicted in Fig. 2.6. This construction gives rise to a larger area for heat transfer than a solid body and also to an effective heat conductivity different from that of the single wire.

In Appendix B, the radiative heat transfer coefficient hP t,r, the convective heat transfer coefficient hP t,c and the thermal conductivity of the entire grid were eval-uated by taking into account the construction of the grid. Fig. 2.16 and Fig. 2.17 show the temperature dependence of hP t,r and hP t,c, respectively. In the investi-gated temperature range, hP t,r changes rapidly with temperature. While, hP t,c is nearly constant. According to Fig. 2.18 the ratio of hP t,r to hP t,c is less than unity. Therefore, it is reasonable to state that the convective heat transfer is the domi-nating mechanism and is also constant in the temperature range of interest. The author did not endeavor to estimate the heat transfer coefficient of the thermocou-ple. It is obtained by the following tryout procedure. Firstly, hP t = 459 W m−2s−1 and kP t = 79.1 W m−1K−1 were calculated respectively. Presuming a random value for hT c, a relationship between the estimated grid temperature and the thermocou-ple temperature was obtained. Of course, this relationship has to be verified by the experimental data. By changing the current through the grid, the temperature was changed accordingly and was measured by the pyrometer (TP t) and the thermocou-ple (Tm), respectively. The results are shown as the open circles in Fig. 2.19. The 95% confidence interval of these data are plotted as the dashed line. Giving hT c an initial value, a linear curve between TP t and Tm was obtained according to Eq. 2.4.29. The final relationship was determined when the predicted values of TP t fall within the 95% confidence interval, yielding a hT c value of 80 W m−2s−1. At the grid temperatures above 1000 K, the temperatures estimated by Eq. 2.4.29 agrees well with the those measured by the pyrometer.

(38)

8000 900 1000 1100 1200 1300 1400 1500 1600 10 20 30 40 50 60 70 80 Tpt, K hpt,r , Wm −2K −1

Fig. 2.16: Radiative heat transfer coef-ficient hP t,r in the temperature range of 800 K to 1600 K. 800 900 1000 1100 1200 1300 1400 1500 1600 453 454 455 456 457 458 459 460 461 462 TPt,K hPt,k , Wm −2K −1

Fig. 2.17: Convective heat transfer coef-ficient hP t,c in the temperature range of 800 K to 1600 K. 800 900 1000 1100 1200 1300 1400 1500 1600 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 TPt,K hPt,r /hPt,k

Fig. 2.18: The ratio of hP t,r to hP t,c in the temperature range of 800 K to 1600 K.

950 1000 1050 1100 1150 1200 1250 1300 1100 1200 1300 1400 1500 1600 1700 Ttc, K Tpt , K

Simulated grid temperature measured grid temperature

Fig. 2.19: The estimated grid temperatures according to equation (B.1.37). hT c = 80 W m−2s−1, hP t = 459 W m−2s−1, kT c = 58.6 W m−1K−1, kP t = 79.1 W m−1K−1, HP t = 0.15 mm, rT c= 0.1 mm.

2.5

Shock tube technique

2.5.1 Principle of shock tube

The shock tube reactor, depicted schematically in Fig. 2.20, has an inner diameter of 0.224 m. It consists of a driver section of 4 m and a test section of 8 m, which

(39)

are initially separated by two aluminium diaphragms. Both diaphragms have two perpendicular grooves, causing the rupture of the diaphragms if the pressure drop over them becomes too high (about 6 bar). A vacuum vessel is mounted below the diaphragm section, separated by a plastic diaphragm. This diaphragm can be ruptured by a needle, driven by a device mounted in the vacuum vessel. The

Fig. 2.20: Sketch of the shock tube and of shock wave propagation (x − t diagram).

(1) (2)

(3) (4)

(5)

pyrolysis experiment in the shock tube is done according to the following procedure. Prior to the experiment all sections are evacuated to a pressure below 2.7×10−6bar. Subsequently, the test section is filled with N2, to a total pressure ranging from 100 to 300 mbar. The driver section and the diaphragm section are filled with He or the mixture of He and Ar to 11 bar and to 5.5 bar respectively.

About 3 s before the driver section is completely filled, v 200 mg of the biomass particles are injected at 1.45 m away from the end plate into the shock tube. Due to the higher pressure in the injector, the suspended particles will spread over the test section to form a homogeneous cloud. Experiments in a glass tube have shown that the particles disperse over a distance of approximately 1 m with respect to the injection point. After that, the plastic diaphragm between the vacuum vessel and the diaphragm section is ruptured, which causes a rapid pressure drop in the diaphragm section. When the pressure drop over the diaphragms becomes higher than 6 bar, the diaphragms are ruptured. Due to the large pressure difference between the driver section and the test section, two types of wave are generated: an expansion wave that

(40)

travels towards the front plate and a compression shock wave (incident shock wave) that pushes the test gas towards the end plate. The wave pattern is depicted in the x − t diagram in Fig. 2.20. It starts when the aluminium diaphragms are ruptured. In the x − t diagram, regions (1) and (4) are the initial states of the test section and the driver section, respectively. Region (2) is the state behind the incident shock. The test gas behind the incident shock is compressed to a pressure range of 1 −2 bar and a temperature range of 500 − 1000 K. The contact surface separates the driver gas and the test gas. Its velocity is lower than the incident shock velocity. When the incident shock reaches the end plate, it is reflected. The test gas behind the reflected shock forms a stagnation region. Typical gas conditions in this region are 7 −10 bar and 950−1500 K. The pyrolysis of the particles takes place in this region. At a certain time, the reflected shock meets the contact surface. Depending on the conditions, different events may occur. A particular situation arises when there is no reflected wave at all: the ’tailored’ interface condition. In general, the reflected shock will partly penetrate region (3) and will partly be reflected from the contact surface as a shock wave or as an expansion fan. These waves affect the stagnation conditions which will be discussed later.

The stagnation conditions will be destroyed anyhow when the head of the expansion wave, reflected from the front plate, arrives. The total test time, ttest is defined as the time between the creation of the reflected shock wave and the arrival of the head of the expansion wave. It has a typical value of about 4 ms.

The stagnation conditions depend on the initial pressure ratio and the composition of the test gas and the driver gas. The state change satisfies elementary conservation laws such that a simple measurement of the incident shock velocity at known initial conditions is sufficient to calculate the velocity, temperature, pressure and density in each region shown in Fig, 2.20. The incident shock velocity is determined by the time difference of the arrival of the incident shock at three transducers placed at different positions along the shock tube. The principle of this calculation is based on the shock theory explained elsewhere [11, 12, 23]. In the following subsection, only the results are summarized.

2.5.2 Conclusions from shock tube theory

The following conclusions are derived on the basis of one-dimensional shock tube theory described in reference [23].

(41)

Relations between regions (1) and (2) The gas velocity behind the incident shock u2 reads u2 = 2a1 γ1+1 Ms1−M1s1 , (2.5.1)

where Ms1 = Us1/a1 is the shock wave Mach number and a1 the sound velocity in the test gas in state 1, given by [23]

a1 = s

γ1RT1 MN2

. (2.5.2)

γ is the adiabatic exponent. It equals 5/3 for a monatomic gas and 7/5 for a diatomic gas. The shock jump relations for density, pressure and temperature can be written as follows ρ2 ρ1 = (γ1+ 1)M 2 s1 2 + (γ1− 1)Ms12 , (2.5.3) P2 P1 = 1 + 2γ1 γ1+ 1 (Ms12 − 1), (2.5.4) T2 T1 = 1 +2(γ1− 1) (γ1+ 1)2 γ1Ms12 − 1 M2 s1 (Ms12 − 1). (2.5.5)

Reflection of the shock wave from the end wall The gas behind the reflected shock is stagnant u5 = 0. The shock jump relations for density, pressure and temperature can be written as follows

ρ5 ρ1 = P5 P1 T1 T5 , (2.5.6) P5 P1 =  2γ1M 2 s1− (γ1− 1) γ1+ 1   −2(γ1− 1) + Ms12(3γ1− 1) 2 + Ms12 (γ1− 1)  , (2.5.7) T5 T1 = [2(γ1− 1)M 2 s1+ 3 − γ1][(3γ1− 1)Ms12 − 2(γ1− 1)] (γ1+ 1)2Ms12 . (2.5.8)

An example of measured pressure signal during a shock tube experiment is presented in Fig. 2.21. It represents the pressure history at the location of 7.5 cm away from the end plate. Two steps can be observed in the pressure signal. The first step is from the passage of the incident shock and the second one from the passage of the reflected shock. After the reflected shock, a stagnation pressure of about 7.3 bar is achieved. This stagnation retains till the occurrence of a second compression resulting from the interaction of the reflected wave and the contact surface. This results in a pressure rise as one can see in Fig. 2.21. Moreover, the pressure is also

(42)

affected by the arrival of the wave reflected from the font plate. This will terminate the stagnation conditions and the pressure drops fast. The aforementioned shock tube theory described is one-dimensional. In reality, boundary layers exist behind the incident shock. The interaction between the reflected shock and the boundary layer may result in a bifurcation or forking of the reflected shock front. If this occurs, the pressure and temperature of the gas will deviate from the theoretical values and then have to be measured or corrected. The bifurcation phenomena have been studied in reference [11]. It was found that the bifurcation effect is merely influenced by two factors: the specific heat ratio γ1 of the test gas in state 1 and the Mach number Ms1 of the incident shock. Strong bifurcation is expected for a high Mach number and low specific heat ratio. Commissaris[11] also found that no bifurcation occurred when using Ar as a test gas and strong bifurcation took place when using CO2. No significant bifurcation was found when a mixture of N2 and O2 was used. In this work, the pyrolysis takes place in N2 at a relatively low temperature so that no strong bifurcation effect is to be expected. In the second step in the pressure signal in Fig. 2.21 no disturbance is observed indicative of the absence of bifurcation impact. The same check was done for all the measurements.

Fig. 2.21: Pressure signal near the end plate during the pyrolysis of Lignocel in N2.

2.5.3 Diagnostics

Two-wavelength pyrometry

Two-wavelength pyrometry is used to measure the surface temperature of the react-ing biomass particles. It has been intensively used to measure the temperature of suspensions of coal particles during combustion by Banin[24], Commissaris[11] and

(43)

Moors[12]. In their experiments, very small spherical particles with a typical radius of 2 µm were used. The wood particles used in this study have a cylindrical shape and are much larger than the coal particles.

In Fig. 2.22, the arrangement of the two-wavelength pyrometry is schematically shown. The light emitted by the particles is converged on the IR silica beamsplitter by a parabolic mirror. Then two light beams pass the narrow band filters in front of the photodiodes with 1.36 µm and 2.21 µm as central wavelengths. The wavelengths are chosen in such a way that no gas emission but the particle emission shall be detected. L1: L2,3: G1 : F1,2: D1,2 : parabolic mirrow convex lenses beam splitter f ilters photodiodes

Fig. 2.22: Schematic representation of the two-wavelength pyrometry. The working principle of the two-wavelength pyrometry is based on the ratio of spec-tral radiances at two wavelengths. First, we consider a single particle in the shock tube. From Planck’s law, the intensity of the radiation at the detector (W m−2) is obtained by multiplying the spectral radiance of the particle by the surface of the particle that can be seen, the spectral width of the band detected 4λ and the solid angle of the optical system 4Ω:

I = ελ Ap λ5 1 (exp(hc/λκTp) − 1)4λ4Ω, (2.5.9) where ελ is the emissivity of the particle at the wavelength λ, h is Planck’s constant, c is the speed of light, κ is Boltzmann’s constant, Tpis the temperature of the particle and Ap is the surface area of the particle. If, however, one considers the radiation of the cloud of particles, the intensity at the detector becomes

I = npελ Ap λ5 1 (exp(hc/λκTp) − 1)4λ4Ω, (2.5.10) where np is the number of particles. In this formula the linearity in np is only valid for a optically thin clouds, where the criterion is, according to Banin[24],

(44)

with l the path length and Npthe number density of particles (m−3). The parameter Qextis the extinction efficiency, which in general depends on particle size, wavelength and optical properties of the particle. In the shock tube experiments nearly 200 mg of material was used. After the passage of the reflected shock this amount of material is suspended in the stagnation region over a length of about 25 cm. With the reference of the particle properties in Appendix A, we obtained a number density of the particle Np = 3.184 × 108m−3 if all injected particles are suspended in the cloud. Employing Qext= 2.0, l = 0.224 m, we find ApQextNpl = 0.224, a value for which using the optically thin approximation gives a maximum of 11.6% error in I. However, with the pyrolysis of the particles this error will drop fast. This gives us an indication that the very first part of the emission signal must not be taken into account.

In the temperature range of 950 K-1400 K the term exp(hc/λκTp)  1. Therefore, Eq. (2.5.10) can be approximated to Wien’s expression

I = npελ Ap

λ5 exp(−hc/λκTp)4λ4Ω. (2.5.12) The intensity ratio of two different wavelengths is then

I1 I2 = ε1 ε2 (λ2 λ1 )5exp c1 Tp ( 1 λ2 − 1 λ1 )  , (2.5.13)

with c1 = 0.0144 Km. Here, the subscripts 1 and 2 represent the two wavelengths 1.36 µm and 2.21 µm, respectively.

The light intensity is measured by means of photodiodes. The light intensity has a good linear relationship with the output voltage of the photodiodes with reference to Appendix C. Now, the intensity ratio in Eq. 2.5.13 can be related to the output voltages of the photodiodes as follows:

U1 U2 = BI1 I2 , = Bε1 ε2 (λ2 λ1 )5exp c1 Tp ( 1 λ2 − 1 λ1 )  , (2.5.14)

in which the factor B incorporates the relative sensitivity of the two photodiodes, the apertures and is constant under the same electrical power supply system. Inserting the value of the wavelengths λ1 and λ2 of 1.36 and 2.21 µm, Eq. 2.5.14 reduces to

U1 U2 = 11.3Bε1 ε2exp(− 4069 Tp ). (2.5.15)

Using this equation, the particle temperature Tp can be derived if the term Bεε12 is known. Note that during the pyrolysis, particles may undergo a heat-up process.

Referenties

GERELATEERDE DOCUMENTEN

Als de ondergrond m-eer krimpt dan de pleisterlaag,ontste.at er dus, naast een schuifspe.nni~ in het hechtvlak,die afneemt naarmate de vervormbaarbeid van de

It could not have been easy for either group to live in America during these times, for not only were they immigrants and subject to the racism that surrounded immigrants

The Assisted Driver Model shows for driving situations which are dominated by tactical and operational tasks executed at rule- or skill-based level, that automation

cooperative arrangements for integrated environmental management of the KBR have been identified as the lack of clarity and poor coordination across national, provincial

Such failure distributions are indeed known for non-spare basic events; unfortunately, for spare basic events the failure distribution becomes conditional on the primary’s (i.e.

Turner in part explains the WELL’s impact on public perceptions of networked computing as due to the editorial policy that granted free accounts on the system to journalists and

Zorg voor water en voerbakken die gemakkelijk gereinigd kunnen worden en waar het water gemakkelijk uit kan stromen8. Denk eens

In het Noord-Hollandse deel van Rijnland is langs de kust tussen Zandvoort en Noordwijk de BKL bij een aantal raaien regelmatig overschreden, zonder dat er sprake is van