Analysis Method of Black Liquor Pyrolysis and Gasification Using Deconvolution Technique to Obtain the Real-Time Gas Production Profile

In thermal reaction experiments, e.g., pyrolysis, combustion, and gasification, the gas released from the reaction can be analyzed in gas measuring instruments. There will be some time delay due to the relatively long gas travel from the reactor to the analyzers. Besides, there can be some time lag in the gas measuring instrument. Gas dispersion may furthermore occur and thus alter the gas concentration profile. The observed gas concentration, therefore, can be very different from the original gaseous reaction products profile. A mathematical procedure called deconvolution technique will be used to get the original gaseous reaction products concentrations profile. The deconvolution technique is based on the assumption that original data have been altered by a transfer function to yield observed data. By the deconvolution techniques, the transfer function for each data set will be calculated and then can be used to compute the original data. In this study, the deconvolution technique was applied to the concentration profile of gaseous products from black liquor pyrolysis and gasification reactions measured by gas analyzers instruments to obtain the real-time gas concentration profile during the processes. Tracer gases are injected in the reactor To facilitate the deconvolution calculation, and their concentration profiles observed in the measuring instruments are recorded. Gaseous products that are analyzed are CO2, CO, CH4, SO2, and H2S. This technique can successfully provide the real-time gas production concentration profile from the black liquor pyrolysis and gasification reaction.


Procedures
The pyrolysis and gasification of black liquor involve many complex chemical reactions in addition to the physical processes. A specially built thermogravimetric analyzer (TGA) for a single droplet black liquor reaction has been constructed to study most of the phenomena that occur during pyrolysis and gasification. Only the gas concentration profiles obtained by the gas analyzer were presented in this paper. As can be seen in Figure 3, the gaseous products from the droplet were carried by the carrier gas, leaving the furnace. The temperature of the gas was too high for the gas analyzer instrument; therefore, it needed to be cooled in the gas coolers. Then, the cooled gas entered the gas analyzers, and the concentration could be measured directly.

Mathematical model
The quantity of tracer that was instantaneously injected into a vessel can be calculated by Equation (2).
where a is the quantity of tracer, f(t) is the function describing residence time, and t is time. The residence time distribution function of the tracer will be given by The pulse signal of the tracer will be modified by the RTD to yield the observed signal. Thus if the RTD is known, the observed signal of the gas leaving the vessel can be calculated. This integration method is called convolution integral (Levenspiel, 1999). The derivation of convolution integral can be illustrated in Figure 4 (Levenspiel, 1999).    (Liliedahl et al., 1991). Equation (7) can be written using Fourier transform as in Equation (8).  (9).
The deconvolution procedure is susceptible to the noise in the transfer function (E). Therefore, the calculated Cin needed to be filtered using a digital filter of the digital signal processing procedure. In this study, the filter that is used is the Butterworth filter. The detailed calculation procedure is described by Liliedahl et al. (1991). The calculation is performed using Octave software that is equipped with digital signal processing toolbox.

Experimental Data
The pyrolysis and gasification experiments were performed at a temperature of 900C.
In the pyrolysis experiment, nitrogen gas was used as the input gas. The nitrogen was inert gas to make sure that only pyrolysis or thermal decomposition of the material in the black liquor took place during the process.
For the gasification experiment, a mixture of CO2 gas and nitrogen was used. In general, the detected gas concentration profile had a similar shape with the other references (Jafarikojour et al., 2014;Serres et al., 2018;Wojewódka et al., 2019). However, only four gases were identified during the pyrolysis experiment at 900C. CO is the main gas products from the black liquor pyrolysis. CO2 and CH4 were also detected in significant amounts. However, SO2 is only slightly detected, and the H2S was not detected.
In the gasification process, the CO2 reacted with the carbon in the black liquor according to Equation (10). The observed gas production is presented in Figure 6. Only two gaseous products were detected in the gasification products: CO and CH4. Therefore, the tracer gases that were shown in the graph are also only for CO and CH4. Deconvolution process for CO2 gas from pyrolysis reaction at 900C: (a) RTD and the normalized gas profile, (b) deconvolved signal and its cumulative sum, and (c) deconvolved signal after filtering

Deconvolution Process
The raw experimental data of gas production were deconvolved through sequential processes. At first, RTD was calculated from the tracer gas using Equation (2) and (3). Secondly, deconvolution was done by Fourier transform using Equation   Figure   7(c). Some ripples still can be seen in the deconvolved signal. However, the trend of the real-time gas production profile can be seen from the plotted data.  In all the deconvolved data, ripples can still be found in the profiles. These ripples indicated that not all of the noises could be filtered. A further study is needed to find the more suitable digital filter techniques that can be used to cancel out more noises to obtain a cleaner and more reliable real-time data profile.

Conclusions
The gas production profiles from black liquor pyrolysis and gasification reaction can be measured using the TGA instrument that has been built. The gas concentration profile can then be deconvolved to obtain real-time gas production. The deconvolution processes yield a heavily polluted signal that needs to be digitally filtered to get real-time data. In general, the real-time gas production profile can be obtained by the deconvolution procedure used in this study.
Notation a = quantity of tracer, mol Cin = inflow gas concentration, ppm Cout = outflow gas concentration, ppm E = residence time distribution,f(t) = transfer function,t = time, s (a) (b)