A comprehensive model of NOx and SO2 emissions from advanced coal combustion in a complex geometry CLC equipment

The paper describes experiences in the modeling of complex geometry CLC equipment. The facility consists of two reactors: the air reactor and the fuel reactor. The fuzzy logic (FL) methods are used in the study for the prediction of NOx and SO2 from the solid fuels combustion in CLC equipment. Maximum errors between measured and predicted results are lower than 10 %.


Introduction
Fluidized bed technology is a convenient method for co-firing of coal and biomass [1,2]. Different combustion atmospheres can be applied in such systems, including air-firing mode and oxy-combustion conditions [3][4][5], generating flue gas, mainly composed of CO2 and H2O, which is almost suitable for geological storage [6,7]. Similar applies to CLC and CLOU technologies [6,8]. However, since solid fuels contain nitrogen and sulfur, NOx and SO2 emissions should be considered before this combustion technology is put into practice [7,9,10].
The manuscript demonstrates an application of the Fuzzy Logic approach as one of the leading artificial intelligence methods [11][12][13][14] to predict NOx and SO2 emissions from CLC equipment. The performed model was successfully validated against experimental results.

Experiments
The necessary data were acquired from experiments carried out on a hot CLC facility at Czestochowa University of Technology, Poland [15,16]. The unit consists of two reactors: an air reactor and a fuel reactor ( Figure 1). A detailed description of the system can be found elsewhere [8,15,17]. The experiments were conducted using coal and biomass as a renewable energy source [18][19][20][21], described in Table 1.

Results
The Qtfuzzylite fuzzy logic control application was used to develop the model [22][23][24]. The following input parameters are employed to develop the model:  IDmode tag defining the combustion mode,  the kind of oxygen carrier OC,  oxygen excess OE,  average fuel reactor temperature T,  F.C. ad / V.M. ad ratio, and N ad /C ad molar ratio,  sulfur S ad and ash A ad contents in the fuel,  IDfuel tag, defining the kind of fuel.
Such selected input variables allow describing the outputs in the developed FL-based model [19,25]. The model uses triangular and constant terms for inputs and outputs, respectively [26]. The validation procedure was successfully performed on the hot facility [27] (Table 2). Comparing measured and predicted SO2 and NOx emissions revealed that the maximum relative error is lower than 10 %. This confirms the good accuracy of the model, allowing for the correct prediction of the emission of sulfur and nitrogen oxides.

Conclusions
A comprehensive FL-based model was shown in this paper for NOx and SO2 prediction from coal and biomass combustion under different combustion modes. Airfired, oxyfuel, CLOU, and iG-CLC conditions are considered in the study. The model's accuracy was successfully confirmed by the validation process. with the maximum error below 10 %.
Scientific work was performed within project No. 2018/29/B/ST8/00442, "Research on sorption processes intensification methods in modified construction of adsorbent beds", supported by National Science Center, Poland. The support is gratefully acknowledged.