Experimental Studies on Municipal Solid Waste and Biomass Pyrolysis

• Main Author: Becidan, Michaël
• Format: Doctoral Thesis
• Language: English
• Published: Norges teknisk-naturvitenskapelige universitet, Institutt for energi- og prosessteknikk 2007
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:no:ntnu:diva-1723; http://nbn-resolving.de/urn:isbn:978-82-471-2758-2

The introduction of this thesis (Chapters 1-9) presents the broader picture of waste management and thermal treatments (situation, trends and novel concepts) with a strong focus on nitrogen (N) in Chapter 6 (a summary of this chapter can be found on page 42). A new insight on N-functionalities is presented, mostly based on plant physiology publications widely ignored by the bioenergy world. N in biomass is found in a variety of chemical compounds and not only in protein compounds. An extensive literature survey concerning N-chemistry during pyrolysis of model compounds and biomass has also been done. A critical light is cast on these studies. Paper I (or P-I) ([Becidan 2004]) presents preliminary results using the experimental set-up and shows its potential in thermal studies. The study of N-release was twofold: NOx release during combustion of biomass and NOx precursors (NH3 and HCN) release during pyrolysis of sewage sludge. The main results confirm known trends: N-release during combustion decreases with increasing fuel-N content; N-release as NH3 and HCN during pyrolysis is clearly dependent on temperature with increasing release with increasing temperature and NH3 as the main component at all conditions. Paper II (or P-II) ([Skreiberg 2004]) presents modelling work realised to assess the potential for reduction of NOx emission formed from fuel-N by implementing staged air combustion. The results obtained from these chemical analysis of ideal reactors (Plug Flow Reactor and Perfectly Stirred Reactor) can be seen as a simplified CFD approach. The reduction potential is depending on a variety of factors and will therefore have to be assessed on a case-to-case basis. However, some conclusions can be drawn: (1) PSR mixing conditions are more favourable than PFR flow; (2) increasing fuel-N content will increase the relative NOx reduction potential; (3) increasing fuel-N fraction of NH3, or HNCO, compared to HCN will increase the NOx reduction potential; (4) increasing amounts of CO, and H2, will increase the NOx reduction potential, but it depends also on the fuel-N compounds; (5) one primary air stage is sufficient, unless also the fuel supply is staged. It is possible to further increase the NOx reduction with more primary air stages at some conditions, but the increase is limited; (6) increasing overall excess air ratio will decrease the NOx reduction potential; (7) increasing residence time will only significantly increase the NOx reduction potential until the main chemistry is completed. However, the time for completion of the main chemistry is significantly longer in a PSR compared to a PFR, and the effect of an increasing residence time is much more pronounced at optimum conditions in a PSR; (8) temperature is an important parameter. However, for a specific set of other parameters there exists an optimum temperature. The temperature in the primary air stage should be high enough to complete the main chemistry. The temperature needed to complete the main chemistry, and the fuel-N chemistry, in a PSR is higher than in a PFR for the same residence time. The temperature in the secondary air stage should be as low as possible, but high enough to ensure complete combustion. Paper III (or P-III) ([Becidan 2007a]) looks at the products distribution and the main pyrolysis products of thermally thick and scarcely studied biomass residues samples. For all fuels, higher temperatures favour gas yield at the expense of char and liquid yields. High heating rate also promotes gas yield. The main gas components were CO2, CO, CH4, H2, C2H2, C2H6 and C2H4. An increase in temperature and heating rate leads to increasing yields for all the gases up to 825-900°C where CO2 and hydrocarbons yields show a clear tendency to stabilise, increase slightly or decrease slightly depending on the fuel. The gas release dynamics reveal important information about the thermal behaviour of the various components (cellulose, hemicellulose and lignin) of the biomass and are consistent with studies using TGA. The gross calorific value of the gas produced increases with increasing temperature reaching a plateau at 750-900ºC. This study provides valuable data of the thermal behaviour of thermally thick biomass samples which is of interest for further work in the area of combustion, gasification and pyrolysis in fixed beds. The study confirms the potential of those unexploited residues for production of energy carriers through pyrolysis. Paper IV (or P-IV) ([Becidan 2007b]) proposes a more extensive study of N-release from 3 biomass residues (coffee waste, brewer spent grains, fibreboard). This study of N-behaviour during biomass pyrolysis of thermally thick samples provided several findings. At high heating rate, NH3 and HCN are the two N-containing compounds, NH3 being the main one at all conditions; NH3 release increases with increasing heating rate and temperature to reach a maximum at 825-900°C while HCN yield increases sharply with temperature without reaching a plateau in the temperature range studied. N-selectivity, N release pattern and N-compounds thermal behaviour are affected by the fuel properties, in all probability including N-functionalities. While the total N-conversion levels to (HCN+NH3) are similar for all fuels at high heating rate, the differences are very significant at low heating rate (more than 2-fold for NH3 and 3-fold for HCN). This can be related to the different fuel properties including N-functionalities. Several attempts have been made previously to correlate N-functionalities and N-release during pyrolysis. However no clear dependence has ever been established for biomass. Furthermore, the intricate and versatile nature of N in biomass samples and its interactions with emicellulose, cellulose and lignin prior to and during pyrolysis are difficult to elucidate. A mechanism of cross-linking between a protein side group and cellulose during pyrolysis was proposed. Further work should focus on the use of the data obtained for improved modelling of biomass pyrolysis. In order to obtain more mechanistic insights the study of model compounds seems more appropriate but may have limited validity because of the intricate structure of “real” biomass. These two types of studies are therefore complementary to obtain a good overview of N-release. Paper V (or P-V) ([Becidan 2007c]) presents the kinetics of decomposition of the three afore-mentioned biomass residues. The results can be summarised as such: (1) The samples were studied at five different T(t) temperature programs. The temperature programs covered a wide range of experimental conditions: the experiments exhibited 10 – 14 times variation in time span, mean reaction rate and peak reaction rate. The experiments on a given sample were described by the same set of model parameters. The optimal parameters were determined by the method of least squares. Three models were proposed that described equally well the behavior of the samples in the range of observations. (2) A model built from three distributed activation energy reactions was suitable to describe the devolatilisation at the highly different T(t) functions of our study with only 12 adjustable parameters. The other two models contained simpler mathematical equations (first order and nth order partial reactions, respectively), accordingly their use may be more convenient when the coupling of kinetic and transport equations are needed. On the other hand, the simpler models needed higher numbers of parameters to describe the complexity of these wastes (3) The reliability of the proposed models was tested in three ways: (i) the models provided good fits for all the five experiments of a sample; (ii) the evaluation of a narrower subset of the experiments (the three slowest experiments) provided approximately the same parameters as the evaluation of the whole series of experiments; (iii) the models proved to be suitable to predict the behavior of the samples outside of those experimental conditions at which the model parameters were determined. Check (iii) corresponded to an extrapolation to ca. four-time higher reaction rates from the domain of the three slowest experiments. (4) The evaluated experiments included “constant reaction rate” (CRR) measurements. This type of temperature control involves a continuously changing heating rate. The simultaneous evaluation of linear, stepwise and CRR experiments proved to be advantageous in the determination of reliable kinetic models. (5) The samples had very different chemical compositions. Nevertheless, the same models described them equally well. Accordingly, the models and the strategies for their evaluation and validation can be recommended for a wider range of biomass studies. Paper VI (or P-VI) ([Becidan 2007d]), this study on thermally thick biomass samples pyrolysis has investigated (1) temperature field, (2) weight loss at two scales (TGA and macro-TGA). The main findings are: (a) Qualitative evaluation of the thermal history: three temperature regimes have been identified: (1) exponentially increasing temperature, (2) linearly increasing temperature (3) 2-slope increasing temperature with a flattening period. The regime at a given point will depend on the sample weight, the reactor temperature and the location in the sample. (b) Quantitative evaluation of the thermal history: significant temperature gradients were measured, with a maximum radial gradient of 167°C/cm for coffee waste at a reactor temperature of 900°C. This will affect the pyrolysis process. (c) The step-by-step pyrolysis chemistry was described and discussed (10°C/min heating rate). By use of a novel concept, i.e. intra-sample heating rate, the exothermic step of pyrolysis was shown. It is related to char and/or char-forming reactions. (d) The comparative study of weight loss in TGA and macro-TGA (10°C/min heating rate, never done before to our knowledge) was performed to investigate the “scaling effect”. Pyrolysis time and pyrolysis rate differences were characterised and quantified. === Paper III reprinted with kind permission of Elsevier, sciencedirect. com