This study consists of 3 parts. In the first part, experimental parameters which affect the
hydrogen production of olive cores in fixed bed gasifier were investigated. H2 / CO and CH /
H2 ratios in the syngas obtained in the second section were investigated. In the last section,
hydrogen values are modeled by using Linear Regression (LR), K Nearest Neighbors (KNN)
Regression, Support Vector Machine Regression (SVR) and Decision Tree Regression
algorithms.
4
The most critical feature in the treatment of a sample by thermal methods, moisture ash, and
elemental analysis is known. In this study, proximate analyzes of the olive cores to be used in
thermal processes were made and the results obtained are presented in Table 3. When Table 3
is examined, it is clear that the elemental analysis and proximate analysis of olive core
samples are suitable for thermal methods. In particular, the low moisture content and the
44.81% carbon value indicate that the samples are suitable for thermal processing.
The time-dependent values of the hydrogen values obtained from all gasification experiments
are presented in Figures 4, 5 and 6. Figure 4 shows that the highest H2 values are reached in
the GA1 experiment. The highest H2 values obtained in these trials were found to be 33%. It
has been observed that the increase in dry air flow used as agent gas adversely affects H2
formation. The lowest H2 values were obtained in the GA3 experiment with the agent flow
rate of 0.2 l / min. When the change in H2 values due to agent flow rate is examined, it is seen
that the increase in agent flow rate negatively affects the time change of H2. As it can be seen
in Figure 4, in the GA3 experiment, H2 formation was reduced to a minimum in
approximately 100 minutes, while a high rate of H2 formation was observed in the GA1 test
and up to 180 minutes in the GA2 test.
In the gasification experiments with pure oxygen, the process is completed in a short time
(Fig.5). Similar values were obtained with dry air as hydrogen yield, but gasification was
completed in a shorter time. Similar to experiments with dry air, a negative correlation was
found between the flow rate of the agent and hydrogen production in the experiments with
pure oxygen.
Similar results were obtained with the other experiments in the pyrolysis pre-stage
gasification experiments shown in Figure 6. It is observed that the process conditions do not
significantly affect the percentage of H2 obtained and have an effect on the temporal change.
In the PGA1 experiment, the peak reached a peak in approximately 170 minutes and in the
PGA2 experiment the peak was reached in the 85th minute.
In order to model H2 formation, the correlations of the model parameters were determined.
The correlation matrix presented in Figure 9 shows the relationship between all parameters
used in the model both with them and with H2. The ''-'' refers to the figure is negative, and
”+“ refers to a positive correlation.
When Figure 9 is examined, H2, HHV, CH , CO has a positive correlation with temperature,
whereas a negative correlation with O is observed. H2 is highly associated with HHV (0.751). Also, it was calculated that H2 has a high correlation value with CO and CH . When the
H2 - CH correlation value is between 0.50 - 0.75, H2 - CO correlation is between 0.75 - 1. H2
- CO correlation was found in the range 0 - 0.25. In the light of the findings obtained from
the correlation matrix, it can be said that H2 has a high correlation with other synthesis gases
4
2
2
4
4
2
CH and CO. Also; H2 values are expected to have a high positive correlation with HHV as
expected.
4
As can be seen in the figure, all 4 model structures successfully predicted the hydrogen
concentrations. The correlations of model tahini obtained for the four different model systems
used in the study with experimental results are given in Figure 10. When the r2 values
presented in Figure 10 and summarized in Table 8 are analyzed, it is seen that the highest
correlation coefficient was reached in the LR model (r2 = 0.99). In four different models, r2
results were obtained above 0.90.
Smaller MSE values are indicative of an increase in model sensitivity. When the MSE values
obtained in this study are examined, it can be said that the model outputs obtained for four
different model tests are highly sensitive. MAE; shows the difference between measured H2
concentrations and model estimates (H2). The fact that the MAE values are close to zero is an
indication that the model estimates are very close to the actual values. When the MAE values
presented in Table 4 are analyzed, it is seen that there are tiny differences between model
estimates and real values. In this study, MAE <1 values were obtained in four different model
structures.
•
•
Examining the model outputs presented in Figure 10, the higher r2 values obtained
from observed and predicted H2 indicate that the proposed models significantly
model H2 concentrations.
When the statistical evaluation of the model outputs is examined; MAE, MSE and
Variance values showed that the applied models were successful in modelling H2
concentrations.