Thermochemical reactor design & thermal breakdown in middle rank coals R. Kandiyoti Department of Chemical Engineering Imperial College London
Lab-scale operation As a university based laboratory we work at bench-scale Pilot-Scale operation Plant scale operation R.V. Spouted-bed reactor (28 mm id) 1,000 C & 40 bar Feed rate 3 g min -1 ABGC Pilot Plant at CRE Platforming Unit UOP Lab-scale research gives fundamental data & is relatively inexpensive
* We know a great deal about thermal breakdown in coals & we know a lot about reactor design!! * But during the pyrolysis of coal (& other solid fuels), our observations are affected by experimental design * Normal reactor design concepts do not work (directly!) in the case of pyrolytic (i.e. thermal) breakdown. The problem is * Thermal decomposition products (of coals) are reactive difficult to keep track of the primary - secondary - tertiary reaction products
* yields & structures of products depend on how products are removed (or not) from parent coal particles & from the reaction zone therefore * different reactor and/or sample configurations, under similar experimental temps & pressures can give different results!!!! In this paper We try to superpose what we know of thermal breakdown ONTO how reaction products are affected by experimental design parameters We aim to explore the value of bench scale experiments in investigating fuel behaviour in larger scale pilot & industrial plant
we will also explore how similar concepts are required to observe important phenomena such as * the effect of heating rate during pyrolysis * effect of solvent type on coal dissolution * effect of time-at-temperature on the gasification reactivity of chars we will review some of these cases
Meanwhile (as in ANY scientific experiment)..we require data on the behaviour of solid fuels.. to be independent (or as independent as possible ) from the experimental method (design)
in short we need to know How experimental design (including reactor design) can affect product yields and quality Let us begin by looking at initial thermal breakdown in coals
start with electron spin resonance spectroscopy * esr allows observing concentrations of stable free radicals embedded in the coal matrix & we think that during in-situ pyrolysis experiments changes (above T2) represent accumulating stable free radicals from completed covalent bond scission reactions i.e. we see increasing debris from pyrolysis reactions Fowler, Bartle & Kandiyoti; Energy & Fuels 3 (1989) 515
These esr data show the onset of extensive covalent bond scission reactions from about 310-340 C depending on coal rank Fowler, Bartle & Kandiyoti,Carbon27 (1989) 197 Coal Elemental C %, daf T 1 ( C) T 2 ( C) Can a 54.2 250 310 Burning Star 75.5 220 310 Linby 83.0 205 310 Point of Ayr 85.4 220 325 Cortonwood b 87.2 250 340 Cynheidre c 95.2 - -
These experiments were carried out between 1984-7 In another set of experiments (Many years later: 2000-2002) we observed the following, which could be directly linked to our earlier esr experiments!
Atmospheric pressure wire-mesh reactor Heating rate variable between 1 C s -1 & 10,000 C s -1 Multistage heating to pre-set temps. Max. Temp 2,000 C
Wire mesh reactor perspective view Gibbins & Kandiyoti Energy and Fuels 3, (1989), 670
600 Heating patterns in the wire-mesh reactor Fast -hold -Slow hold Or slow-hold-slow-hold 500 1 /s Temperature ( C) 400 300 200 100 1000 /s Temp-1 Temp-2 Temp(Mean) 0 50 100 150 200 Time (s)
Heat coal particles at 1 ºC s -1 OR at 1,000 ºC s -1 to 400 ºC then extract the chars with NMP Extract Yield (mass%,daf) 70 65 60 55 50 45 40 35 30 25 20 Coal A 1000(400) 1(400) 0 20 40 60 80 100 120 Hold Time (s) * NOTE: The internally released extractable material is stable at 400 ºC during at least two-minutes
Thus internally released extractable material a. may be extracted by a good solvent during liquefaction b. or if the temperature is raised without solvent we get dry pyrolysis more on pyrolysis a little further on
Let us see how such information may help us explain observations from coal liquefaction experiments We have an unusual liquefaction reactor design: removes extracts from the reaction zone as soon as they are released from the parent coal particle!
The flowing-solvent reactor assembly Xu & Kandiyoti Energy and Fuels 10 (1996) 1115
Temperature and power control history 5 ºC s -1 to 450 ºC with 400s holding Solvent flow rate : 0.9 ml s -1 at 70 bar Xu & Kandiyoti Energy & Fuels 10 (1996) 1115
Liquefaction Weight loss in tetralin as function of temperature Flowing solvent reactor: 5 C s -1 with 400-s hold Tetralin flow rate: 0.9 ml/s at 70 bar * Extraction yield takes off between 350-375 ºC * Liquefaction to 400 ºC, we get 45 65 % conversion! Xu & Kandiyoti Energy & Fuels 10 (1996) 1115
It appears that * Coal samples must climb high up the ESR curve before significant amounts of extract are released (~375-400ºC) * Many bonds must rupture before large fragments can be released into the coal particle from the solid matrix * This would explain the slow rise between 325 & 375 C Point of Ayr coal shows this better than some others
During liquefaction, we can recover large amounts of extract above 350-375 C in a good solvent But not in dry pyrolysis In dry pyrolysis, bond scission is similar BUT material released from the solid matrix remains within the coal particles and at 400 ºC the extractables are stable for minutes!
Conversions in the Flowing Solvent Reactor solvent --------------------------------------------------- flow rate of 0.9 ml s -1 at 70 bar(g) HRate Holding Medium Weight Loss C s -1 time(s) 350 C 450 C --------------------------------------------------- PoA vitrinite: 1,000 150 Helium 3.3 20.5 (2)* 5 400 Tetralin 28.8*** 77.6 (2) 5 400 Q/P** 38.0 73.8 (2) 5 400 Quinoline --- 72.7 (2) 5 400 Hexadecane 12.5 27.3 (2) PoA whole coal: 5 400 Tetralin 24.6 82.5 (4) 5 400 Quinoline 39.5 74.7 (2) 5 400 Hexadecane --- 24.0 (1) --------------------------------------------------- * Number of repeated runs used for calculating the average value ** Q/P: quinoline/phenanthrene (2.5:1 w/w) mixture. *** Holding time: 500 seconds. The weight loss from 100 s experiments under the same conditions was 29.2 %, within experimental error. All data: (% w/w daf basis)
so for a good solvent that is not a donor solvent, the conversions are close to those of a donor-solvent in the flowing solvent reactor & what happens when flowing-solvent reactor results are compared with extractions in closed bomb (i.e. batch) reactors??
a. Effect of solvent type on conversion. Flowing-solvent reactor. Heating at 5 C s -1 ; solvent flow rate: 0.9 ml/s at 70 bar b. Flowing Solvent & Mini-Bomb reactors. 1-methylnaphthalene solvent; solvent/coal ratio in m-b: 4/1 Flowing solvent reactor Mini-bomb reactor Gibbins, Kimber, Gaines & Kandiyoti, R., Fuel 70, (1991), 380
The type of solvent thus has a determinant effect on the course of the liquefaction processes BUT SO DOES THE CHOICE OF REACTOR!!! Using non-donor (good) solvents in the flowing solvent reactor has a far less dramatic effect: because extract free-radicals dissolve in excess solvent & are greatly diluted This shows us how results can be affected by experimental design
Summarising our findings * Initial thermal breakdown reactions during pyrolysis & liquefaction in coals are similar occur ~ 310-350 *More than one (probably several) bonds must break before the solid matrix releases tar precursors (extractables) into particle * Exit of depolymerized material from particles depends on the chemical characteristics of the surrounding medium: Conversions in hexadecane (a liquid) and helium (a gas) were not much different! * Liquefaction in a good (non-donor) solvent allows much but NOT all depolymerized material to be carried away.
let us now examine the effect of reactor design on the results of pyrolysis/gasification experiments but first we need to think about?-happens during pyrolysis
during dry pyrolysis at 400 ºC, most of the extractables are still intact within the coal particle When the temperature is raised further, - some of the lighter components evaporate - most of the extractables crack, producing lighter tar & light gases -but a significant amount of the extractables recombines to form char That is why we get 50-60 % char during dry pyrolysis!
We need to keep an eye on two key parameters 1. Effect of heating rate 2. Trajectory of the volatiles
Effect of heating rate on tar and total volatile yields Linby coal, atmospheric pressure He, 700ºC, 30 s hold Particle size:106-152 μm Heating rate range: 1 ºC s -1-1000 ºC s -1 We think the effect is due to 1. Rapid volatiles ejection 2. FR quenching by native hydrogen Fuel 68 (1989) 895
& what happens when particles are stacked together? Products of pyrolysis are reactive Tars may re-polymerise to char and/or crack to gas. Product distributions thus (also) depend on extents of volatiles/solid contact * The outcome of experiments depend on how volatiles are removed from the reaction zone * Char gasification reactivity depends on mode of pyrolysis fast/slow heating? tars removed/condensed on char?
Hot-Rod fixed-bed reactor Resistance heating of the high-pressure reactor tube 6 mm id & 8 mm id 800 C & 100 bar or 1,000 C & 40 bar Fuel 66, (1987), 1413 Fuel 77, (1998), 1411
Effect of bed height and sweep gas flow rate on tar yields: "hot-rod" reactor at atmospheric pressure Sample Size (mg) 300 300 50 50 Approximate Bed Depth (mm) 20 20 4 4 Superficial Velocity (m/s) 0.1 9.5 0.1 9.5 Approximate Gas Residence Time in the Coal Bed (s) Tar Yields (%w/w daf coal) % Change in tar yield over the Base Case (%w/w daf coal) 0.2.002 0.03 0.0003 16.0 (2)* Base Case 18.0 18.6 (2)* 21.9 (3)** +2 +2.6 +5.5 * average of two runs ** average of three runs
and at high pressure? Comparing yields between the (i) wire mesh reactor & (ii) a fluidized bed reactor In experiments carried out at 1,000 C & 1 30 bars pressure The designs of the high-pressure rigs are somewhat different
High-Pressure Wire-Mesh Reactor 160 bar at 850 C or 40 bar & 2,000 C at 1 10,000 C s -1 steam injection capability 5 mg sample Messenbock, Dugwell & Kandiyoti Energy and Fuels 13, (1999), 122
High-Pressure Fluidized-Bed Reactor System to product recovery 1,000 C & 30 bars Body: Incoloy 800HT sample injected as a single slug Megaritis, Zhuo, Messenbock, Dugwell, & Kandiyoti, Energy & Fuels 12, (1998), 144
8 9 48 mm o.d. ; 32 mm i.d. 504 mm long 6 7 10 11 12 Creep Rupture Limit: 1000 hr at 1,000 C at 40 bar 5 3 13 4 2 21 20 1 14 2 19 18 13 15 4 16 Megaritis, Zhuo, Messenbock, Dugwell, & Kandiyoti, Energy & Fuels 12, (1998), 144 6 17
Pyrolysis of Daw Mill (UK) coal 1,000 ºC between 1 30 bar fluidized-bed (FBR) & wire-mesh (WMR) reactors conversions tar yields Megaritis, Zhuo, Messenbock, Dugwell, & Kandiyoti, E & F 12, (1998), 144
CO 2 -gasification of Daw Mill (UK) coal 1,000 ºC between 1 30 bar fluidized-bed (FBR) & wire-mesh (WMR) reactors Megaritis, Zhuo, Messenbock, Dugwell, & Kandiyoti, E & F 12, (1998), 144
in CO2 gasification differences show up between between reactors approximating * single particle behaviour (+ fast heating) & * stacked particles ( + slow heating)
CO 2 -Gasification reactivity measured in 3-reactors Daw Mill (UK) coal, 1,000 ºC, between 1 & 30 bar wire-mesh & fluid bed Hot-rod (fixed-bed) Megaritis, Messenbock, Collot, Zhuo, Dugwell, & Kandiyoti, Fuel 77, (1998), 1411
.we have seen that * experimental design also affects the gasification reactivities of chars * condensed pyrolysis tars in H-R reactor can deactivate residual chars * time of exposure (due to slow heating) can also deactivate chars
Combustion Reactivities; Chars from pyrolysis runs Daw Mill(UK) coal: 1000 C in atmospheric pressure He Combustion Reactivity R max (%/min., daf) 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 s 10 s 60 s 1 10 100 1000 10000 Heating Rate ( o C/s ) We observe significant loss of reactivity in the first 10 s, even at 1,000 C!
At 950 1000 C, chars deactivate by as much as a factor of 3 within ~ 10 s! Unless particles are consumed quickly, OR the temperature increased, char consumption will be much slower after 10 s Char deactivation must be quantified and taken into consideration in kinetic and reactor modelling What does this mean for existing kinetic models of coal gasification which contain time-independent reaction rate constants?
summing up * these examples all arise from the reactivity of products formed during thermal breakdown * test reactor design must therefore attempt to take account of changing sample properties ( moving targets ); we always need information on how these properties change * When attempting to generate data to mimic (larger) pilot or plant scale equipment, we are really trying to match data from two moving targets [test reactor and real system]
Thank you for your attention!