This thesis contains no text, graphics or tables copied and pasted from the Internet, unless specifically acknowledged and the source is detailed in the thesis and References sections. To the technical staff of the chemical technical school for help in setting up the experimental apparatus. Ronelle Gurney and Dhelia Raman, students who helped run the experiments.
Since the early 1920s, when furfural was first produced, several other processing methods have been developed, but none have been able to produce yields comparable to those obtained in the standard TAPPI process for xylan, which almost completely converts xylan to furfural.
INTRODUCTION
Unlike the other processes that use liquid acid catalysts, the catalyst used for s-Suprayyield is gas-phase HCl. Since ozone depletion in stratospheric ice crystals has been observed to be catalyzed by gas-phase HCl, it was hypothesized that due to the structural similarities between the ice crystals and hemicellulose, the catalytic properties of gas-phase HCl might also be applicable to hemicellulose. It has previously been thought that a process like s-Suprayield, which is dependent on gaseous acid catalysis, is not possible at moderate temperatures and pressures, as this type of catalysis is dependent on ionization of the acid – a process which typically takes place in the gas phase at over 1000ºC.
However, studies of ozone depletion reactions in the stratosphere have indicated that acid dissociation may be possible at low pressures and temperatures as a result of it being catalyzed by ice crystals in polar stratospheric clouds.
LITERATURE REVIEW
F URFURAL – ITS MANUFACTURE AND USES
These products are formed by the reaction of furfural with intermediates in the furfural formation process, leading to losses of furfural produced in the liquid phase reaction (Antal et al., 1991). In addition to the reactions and products described in the table, chlorination, nitration, and sulfonation can also occur as a result of interaction with strong acids commonly associated with furfural production (Zeitsch, 2000a). Sufficient excess steam is used to drive furfural out of the reaction zone in the vapor phase to limit the occurrence of loss reactions.
The furfural layer from the decanter is then fed to a dehydration column before going for further distillation (Kirk-Othmer Encyclopaedia of Chemical Technology).
T ECHNOLOGY FOR FURFURAL PRODUCTION
Steam is used for both heating and stripping, as in many other processes, and is continuously supplied to the bottom of the reactor so that the steam currently flows against the feedstock. Both the Rosenlew and Stake methods had the cost-saving advantage of using native acids (i.e., the acetic acid resulting from the hydrolysis of hemicellulose). It is suggested by Zeitsch (2000a) that the high yield in the analytical process is a result of furfural entering the vapor phase during its formation.
Based on the discovery of the ice-catalyzed dissociation of HCl, which will be discussed in Section 2.5.1, Zeitsch suggested that a similar dissociation could take place on hemicellulose—the feedstock from which furfural is produced.
H EMICELLULOSE STRUCTURE
The composition affects the macrostructure of hemicellulose because the number of branch points can be related to the concentration of arabinose (Chaikumpollert et al., 2004). Rodriguez-Chong et al. (2004) reported in their experiments on sugarcane using nitric acid that while glucose was released as a hydrolysis product, no degradation product of hydroxymethylfurfural was detected. To determine the reaction stoichiometry, the molar ratio between acetyl groups and xylose must be determined.
X is the mole fraction of acetyl groups per mole of xylose MMacetyl is the molar mass of the acetyl group (C2H2O-).
L IQUID PHASE KINETICS OF XYLAN HYDROLYSIS
Therefore, the fraction susceptible to hydrolysis depends on the reaction conditions (Garrote et al., 2004). The fraction of hemicellulose that reacts quickly depends on the type of plant material used (Garrote et al., 2004). This causes the acid concentration to decrease and the reaction rate to decrease (Antal et al., 1991).
Particle size has an effect on reaction rate, but the effect is very small (Lavarack et al., 2002).
H YPOTHETICAL MECHANISM OF GAS - PHASE CATALYSIS
- HCl dissociation on ice
- The significance of gas phase catalysis to s-Suprayield
The PSCs have thin layers of mobile water molecules with which the HCl can interact (Baceleo et al., 1999). Recent research (Aguzzi et al., 2003) has shown that the ice surfaces of type II PSCs are not the only ones capable of catalyzing heterogeneous atmospheric reactions, in the troposphere both cirrus clouds and aircraft paths may also be possible sites for such reactions. Research has shown that three water molecules forming part of a hexagonal ice surface are sufficient to catalyze the process (Voegele et al., 2002).
Unfortunately, it is almost impossible to obtain experimental data, since the stratospheric conditions are extremely difficult to create - therefore, many of the studies carried out are by computer simulation (Mantz et al., 2001).
THESIS OBJECTIVES
EXPERIMENTAL WORK
- E QUIPMENT
- Plant description
- Reactor design
- Heating options
- Utilities and control
- P REPARATION FOR TESTWORK
- Pump calibration
- Raw material and bed characterisation
- E XPERIMENTAL P ROCEDURE
- Sample preparation
- Start up
- Experimental run
- Shutdown
- S AMPLE ANALYSIS
- Potentiometric Titration
- Refractive index
Despite being insulated, the top of the reactor and the thermometer opening were areas of heat loss. It was concluded that the voltage to the superheater wire could be successfully used to control the temperature of the superheater glass and thus the gas outlet temperature. The flow of the oil was split, with one part directed to the reactor jacket and one part directed to the evaporator.
Full results and discussion of pentosan analysis can be found in Appendix B.
RESULTS
S TATISTICAL TESTS ON DATA
Analysis of variance was performed on the experimental data to confirm that there was a statistically significant relationship between acid concentration and furfural yield (Figure 27), while the experimental data did not show a clear relationship between temperature and furfural yield (Figure 26). Based on these data, an analysis of variance was performed to determine the significance of the effect of each of the two factors on the reaction yield and the significance of any interaction between each factor. The effect of each factor was then determined from a table consisting of only one data point per condition, viz.
In Table 5, effect A and B refer to the effect of temperature, the effect of acid concentration, respectively, while F0 is the F test statistic. The parameters v1 and v2 are the numerator and denominator degrees of freedom as explained in Appendix F. Therefore, at all significance levels, the interaction effect as well as the effect of temperature were insignificant.
At a 5% level of significance, the effect of HCl concentration was found to be significant, confirming what was deduced from the graph. Despite the statistics indicating that the effect of temperature is not significant, it was felt that these results may have been due to the inconsistencies in the temperature measured and therefore this result should be viewed with caution. More experiments using a better way of temperature control should be performed to verify this result.
In the assumptions made in the modeling section, the effect of temperature is still considered.
MODELLING
- P HYSICAL DESCRIPTION OF THE PROCESS
- S IMPLIFYING ASSUMPTIONS
- The particle model
- The reactor model
- M ODELLING OF THE REACTION MECHANISM
- R EACTION RATE KINETICS
- R EACTION STOICHIOMETRY
- R EACTOR MODELING
- Gas-phase material balances
- Solid-phase material balances
- Calculation of measurables
- O PTIMISATION OF MODEL PARAMETERS
- M ODELLING RESULTS AND DISCUSSION
Diffusion of reactant and catalyst into the particle by intraparticle diffusion. Finally, the spatial domains can be ignored and only the time domain must be considered, because the composition of the gas depends on the reaction in the individual particles. A side reaction also took place involving the hydrolysis of the acetyl groups on the pentosan chain, resulting in the production of acetic acid.
It was also assumed that the porosity of sunflower peels does not change with increasing xylan conversion. Although ultimately all products were in the vapor phase, the change in solid volume over the course of the reaction was believed to be negligible. Because the characteristic reaction time is much greater than the residence time of the gas, several simplifying assumptions regarding the operation of the reactor could be made.
These assumptions allowed the application of the pseudo-steady-state assumption to the gas phase. After the above discussion it was decided to neglect both the accumulation terms and the axial concentration gradients from the gas phase material balances. Cj is the molar concentration of component j in the gas phase (in kmol/m3) Rj is the net rate of formation of component j per unit volume of the solid (in kmol/(m3s)).
The values of the reaction rate constant for each step indicated that reaction 1 proceeded the slowest, while reaction 2 and reaction 3 occurred at more or less the same rate. The concentration of the P-HCl complex followed a similar trend, increasing to a maximum value and then decreasing as it was consumed in the production of furfural.
CONCLUSIONS AND RECOMMENDATIONS
C ONCLUSIONS
R ECOMMENDATIONS
To determine the temperature of the gas outlet, the length of the wire per meter of pipe had to be determined. It was determined by the thickness of the glass "wall" (tw) divided by the thermal conductivity of the glass (kg). A clean and dry 100 ml volumetric flask was weighed and the volume of reagent specified in the table was added to each of the flasks.
After adding the reagent, the flask was weighed again to determine the mass of reagent used. To verify the concentrations of this starting solution, a weighed 10 ml sample of each of the acids was titrated with 1M NaOH. To flask 1, 20 ml of the solution produced above was added and 10 ml was added to each of the subsequent flasks.
The flasks were then weighed again to determine the mass of the original solution used. The cutoff varied depending on the refractive index reading for the distilled water on a given day and therefore it is the gradient of the line that is most important to the calibration curve. Therefore, although the values were likely to give an indication of the trend of change in xylan concentration with size fraction, these values could not be used in yield calculations and a test was carried out using new equipment and a curve of re calibration for it. confirm pentosan concentration.
If H0 is rejected, it would mean that at least one of the factors or the interaction effect would be greater than zero and therefore significant. Since there are only replicates of some of the data points, a separate matrix was formed to determine the sum of squares of errors and the total sum of squares. The occurrence of side reactions cannot be ruled out between the intermediates of the dehydration step as well as reactions of the furfural and the intermediates with the lignin.
The result of the reaction is a net generation of water which therefore begins to diffuse back through the newly formed ash layer into the gas phase.
DIFFUSION CALCULATION
Using all these values to enter the Sherwood number correlation resulted in a Sherwood number of 2.27.
