Simulation of the Steam Gasification of Japanese Waste Wood in an Indirectly Heated Downdraft Reactor Using PRO/II™: Numerical Comparison of Stoichiometric and Kinetic Models

Abstract

The conversion of biomass to olefin by employing gasification has recently gained the attention of the petrochemical sector, and syngas composition is a keystone during the evaluation of process design. Process simulation software is a preferred evaluation tool that employs stoichiometric and kinetic approaches. Despite the available literature, the estimation errors of these simulation methods have scarcely been contrasted. This study compares the errors of stoichiometric and kinetic models by simulating a downdraft steam gasifier in PRO/II. The quantitative examination identifies the model that best predicts the composition of products for the gasification of Japanese wood waste. The simulation adopts reaction mechanisms, flowsheet topology, reactions parameters, and component properties reported in the literature. The results of previous studies are used to validate the models in a comparison of the syngas composition and yield of products. The models are used to reproduce gasification at temperatures of 600∼900 °C and steam-to-biomass mass ratios of 0∼4. Both models reproduce experimental results more accurately for changes in the steam-to-biomass mass ratio than for temperature variations. The kinetic model is more accurate for predicting composition and yields, having global errors of 3.91%-mol/mol and 8.16%-g/g_BM, respectively, whereas the simple stoichiometric model has an error of 7.96%-mol/mol and 16.21%-g/g_BM.

1. Introduction

Biomass resources have been a controversial choice for energy utilization in the last decade because of possible negative effects relating to energy crop deforestation, food security, and biogenic CO₂ emissions [1]. The latest misuse of biomass can be traced to a highly dependent carbon-based economy. Additionally, Dente et al. [2] ascribed the negative effects of biomass use to the lack of a more effective bio-based circular economy and a clear valorization of biomass. It is crucial to consider biomass utilization for added-value purposes, such as the production of food, fertilizers, pharmaceuticals, and chemicals. The chemical sector faces several sustainability threats that can be addressed with rational biomass use [3]. In 2020, the industrial sector contributed 24% of global greenhouse gas emissions, and almost 5.8% of this contribution was from the petrochemical sector (including emissions from direct industrial processes and energy consumption) [4]. Yet the mitigation of CO₂ emissions in the energy sector has been tackled by shifting the carbon cycle economy, such as through the use of fuel cells and photovoltaic solar energy. This strategy cannot be directly applied to most carbon-based chemical subsectors. For instance, containers used for several everyday commodities (from food and pharmaceutical packing to disinfecting products used to limit the spread of COVID-19) are made according to sanitary and health standards, traditionally using organic chemicals such as polymers and resins [1].

In recent decades, biomass has been proposed as a renewable alternative to fossils in the petrochemical sector [3]. Life-cycle analysis has shown that the production of steam-cracking petrochemicals can be mitigated by producing bioplastics from biomass-derived ethylene, propylene, and BTX (i.e., a mixture of benzene, toluene, and xylenes) [5]. Light olefins (i.e., ethylene, propylene, and other C4 streams) are key petrochemical building blocks that are currently under revision for sustainability efforts, particularly in the selection of a suitable feedstock [6]. Within the spectrum of biomass sources, solid waste biomass is of great interest by not competing with food security or promoting deforestation [2]. Regardless of the low availability and expensive collection and management of waste biomass, this feedstock fits the agenda of a sustainable bio-based circular economy and waste management. The Japanese case study of bio-product utilization is of great interest, as Japan is the fourth largest consumer of petrochemicals and the fifth largest biomass market worldwide. Although its efficient waste management program maintained a biomass recycling ratio near 71% in 2020, most available lignified biowaste is forest wood waste (from leftover/thinning waste), rice residue (from straw/husk), and cardboard. Previous studies forecast a theoretical substitution of 21% of fossil-based petrochemicals (olefins and BTX) with biomass-based feedstock in Japan [7].

The conversion of biomass to olefins (BTO) remains at a conceptual level of design and many questions regarding chemical concepts and optimized pathways in a biorefinery remain unsolved [8]. For instance, there is no consensus on the use of direct BTO with catalytic reactions, and several studies have suggested the use of intermediate platforms such as methanol and dimethyl ethers [5]. Currently, most commercial-scale biorefineries are designed to yield methanol as a biofuel, commonly adopting the biochemical conversion of biomass possessing low lignin content [9]. The abundance of forest residue in Japan hinders traditional conversion through biochemical pathways, and lignified biowaste is, thus, converted alongside thermochemical pathways that easily decompose lignin [10]. In the framework of thermochemical conversion, gasification and synthesis gas (syngas) platforms play the main role in decomposing the complex lignocellulosic structures of biomass [11]. Steam gasification is a preferred technology that promotes clean syngas with high contents of H₂ and CO and is favorable for the synthesis of methanol and dimethyl ether [12]. Still, the evaluation of different BTO process scenarios requires calculations that rely on physical or numerical models at a conceptual level [13].

Accordingly, the numerical modeling of gasification is a fundamental approach for assessing different upstream conversion steps (e.g., direct syngas to olefins and additional platforms such as methanol) [14]. Gasification has been widely studied and modeled in recent decades owing to a high interest in the production of syngas for energy use [15]. Different simulation approaches include the use of dimensional and dimensionless models that solve a set of conservation equations (e.g., mass, momentum, energy, and species) with respect to space and time. Regarding space variations, dimensional models solve the conservation equations over a discrete region, while dimensionless models ignore gradients of properties within the reactor. A detailed mechanical design usually adopts dimensional models that include computational fluid dynamic models, while conceptual and plant designs mostly use dimensionless models. Temporal variations during gasification are included in the models by assuming an infinite (e.g., thermodynamic equilibrium approach) or fixed reaction time (e.g., kinetic approach). Equilibrium models neglect the geometry of the gasifier and assume an infinite reaction time to reach thermodynamic equilibrium. This assumption can be made by applying equilibrium constants (in the case of stoichiometric models) or by minimizing the Gibbs free energy (in the case of nonstoichiometric models). Non-stoichiometric models assume a stable equilibrium by solving the conservation equations at a minimum value of the Gibbs free energy of formation and restricting the mass balance of individual elements between reactants and products. Non-stoichiometric reactors are preferred, despite their lower precision, because a particular reaction mechanism is not required to predict the yield of products [16].

Unfortunately, the adoption of a thermodynamic equilibrium idealizes gasification to a maximum yield of products while excluding the residence time of the reactants, resulting in highly unrealistic results that overlook the geometric design. Meanwhile, kinetic modeling acknowledges reactant conversion rates by involving parameters such as residence time, temperature, and hydrodynamic parameters, enabling the simulation of complex reactors (e.g., fixed bed, fluidized bed, and entrained flow gasifiers) [17]. Still, the kinetic approach demands detailed information of the gasification mechanism, normally obtained experimentally or by making several assumptions [18]. In general, both equilibrium and kinetic approaches enable the simulation of gasification with material and energy integration, incurring a narrowing-down error. Mutlu et al. [15] comprehensively reviewed recent simulations conducted in Aspen Plus. They recommended implementing equilibrium models to establish a basic relationship between operating conditions and kinetic models for extensive process modeling.

The implementation of a BTO simulation demands the initial selection of one of the former modeling approaches. Most BTO studies have modeled gasification with a simple non-equilibrium approach (using Gibbs reactors or yield blocks) [8], rarely including an extensive validation of the gasification and reporting a simulation error below 5% [6]. Nonetheless, sensibility analyses of the reaction temperature or reactant content using simplified gasification models can be inaccurate [19]. Currently, there is no clear quantification of the simulation errors when selecting one model or another. For instance, Gonzalez et al. [20] compared the performances of stoichiometric and non-stoichiometric reactors in Aspen, concluding that the former reactor provides a better fit with experimental results. In this comparison, they excluded a detailed error quantification and only reported the global simulation errors of 2.1% v/v and 7.3% v/v for stoichiometric and non-stoichiometric reactors, respectively. Yu et al. [16] made a similar comparison between kinetic and non-stoichiometric reactors, concluding that the kinetic reactor provides a better fit with experimental data. They reported the relative error in the simulation only for the molar fractions of CO and CO₂, which ranged from 11% to 200%, excluding the error in the prediction of product yields.

The present paper compares the fitting of simulation results of stoichiometric and kinetic models with the parametric performance of a downdraft gasifier. This examination aims to identify the model that best predicts the composition of products during the gasification of Japanese wood waste and to quantify the lowest estimation error. The paper focuses on the mass balance of the steam gasification as a function of temperature (T) and the steam-to-biomass ratio ($S/B$). The models are implemented in PRO/II™, following the reaction mechanism for the gasification in a downdraft reactor [21]. The properties of components and the performance of the gasification are obtained from previous studies on Japanese wood waste [22]. Additionally, the previous studies contribute to the validation of the models in a comparison of the syngas composition and the yield of products [23]. The last section discusses the prediction error of each evaluated model and makes comparisons with results from the literature [24]. This study proposes a novel comparison of gasification with stoichiometric and kinetic approaches, using an alternative simulation software such as PRO/II™. This study also conducts an unusually extensive validation that enables a quantitative selection of a steam gasification model for further use in the biomass-to-petrochemical process design. As a result, this research precedes a more comprehensive study of the optimization of the thermochemical conversion of Japanese solid biowaste to biochemicals (light olefins), concerning the assessment of the environmental impact and techno-economic feasibility.

2. Gasification Mechanism

Gasification is a thermochemical conversion of organic compounds into synthesis gas (syngas) conducted from 600 to 1000 °C, using a reaction agent such as oxygen, steam, or CO₂. In the current study, biomass is considered an organic matter composed of carbon, hydrogen, and oxygen and free of inorganic matters and water for simplification. The gasification of biomass ($BM$) encompasses a series of conversion substages that decompose, reduce, and rearrange the lignocellulosic structures of biomass into simpler molecules [15]. The initial stage of the former decomposition is called pyrolysis and occurs in absence of a reaction agent. In contrast to char gasification, where almost pure carbon is directly reduced with the reaction agent, biomass gasification requires pyrolysis to start reduction and rearrangement reactions [17]. Figure 1 summarizes the sequential decomposition of biomass into syngas through the operational units of drying, pyrolysis, gasification (reduction), and combustion. For modeling purposes, these substages are divided into separated zones regardless of the undefined boundaries in a real gasification operation.

The pyrolysis of biomass comprises a series of complex reactions (i.e., carbonization, devolatilization, and recondensation) in the absence of oxygen from 200 to 900 °C [25]. In the current study, pyrolysis is simplified in global reactions $R1$ and $R2$ (as shown in Figure 1), conducted at 500 °C [26]. The carbonization reaction performed during pyrolysis dramatically increases the carbon content in the solid phase, yielding solid carbon ($C_s$) with a yield near the fixed-carbon ($FC$) content present in the biomass. Meanwhile, the devolatilization reaction, $R2$, occurs near the volatile matter ($VM$) and produces several gases (i.e., CO, CO₂, H₂, and CH₄), primary tar ($ta{r}_1$, defined as condensable organic compounds), and pyrolysis water $H_2{O}_{pyro}$ [16]. Primary tar is reformed during the initial gasification stage to produce simple gases and secondary tar ($ta{r}_2$), according to tar cracking reaction $R3$ [9].

During the operation of gasifiers such as downdraft, fluidized bed, and entrained flow gasifiers, pyrolysis products (i.e., carbon, pyrolysis gas, and tar) are collected and forced to stream to the combustion and reduction zones, reacting at a higher temperature [27]. Steam gasification is preferred over other reaction agents to maximize the content of H₂ in the syngas [17]. The presence of O₂ during gasification promotes the production of CO₂ in the combustion zone, leading to a high yield of CO and CO₂ in the reduction zone [28]. Recently, gasification involving supercritical water as a reaction agent (above the water critical point, 22.12 MPa and 374.12 °C) is proposed as an alternative concept for the conversion of wet biomass into syngas, since it reduces drying energy consumption and increases hydrogen yields [29]. Still, technical challenges remain unsolved for large-scale supercritical water gasification, such as the biomass feeding, salt precipitation, corrosion, or efficiency of heat exchangers. Consequently, large-scale steam gasifiers currently operate under subcritical conditions. Directly heated gasifiers supply the energy demand of the process through the internal combustion of the pyrolysis products. In contrast, indirectly heated downdraft steam gasifiers have an external energy supply, and there is no combustion within the reactor (i.e., no oxygen is provided) [9]. The latter configuration hinders the production of CO₂ during gasification and increases the yield of H₂ [17].

The gasification zone includes heterogeneous and homogeneous reduction reactions at temperatures above 600 °C. The solid carbon reacts with the water stream and the pyrolysis gas in an adsorption/desorption heterogeneous mechanism analogous to a Langmuir–Hinshelwood mechanism, increasing the porosity of solid particles [30]. In the present study, heterogeneous reactions include only the Boudouard reaction, $R4$, and the water gas primary reaction, $R5$. The desorption gas, unreacted pyrolysis gas, and water steam undergo a homogenous water gas shift (WGS) reaction, $R6$. This paper excludes the methanation reaction because of the low content of CH₄ in steam gasification. Crude syngas comprises CO, CO₂, H₂, CH₄, and secondary tar. The unreacted char is usually burned in an external combustion reactor to generate the process’s demanded heat or used as activated carbon for filtration systems [8].

3. Model Development and Methodology

The current study evaluated two gasification models, namely $StoicM$ adopting a stoichiometric approach (i.e., using a thermodynamic equilibrium model) and $KinetM$ adopting a kinetic approach. For both models, the process simulation followed the methodology for the material integration of an existing plant proposed by J. Haydary [13]. The operational units and gasification products were selected according to the mechanism discussed in Section 2. Section 3.2 presents the physical properties of the components whereas Section 3.3 presents the development of the flowsheet topology. The kinetic parameters of the two models are described in Section 3.4 and Section 3.5. Finally, Section 3.6 reports the methodology for model validation.

AVEVA™ PRO/II™ Simulation 2020, as part of Schneider Electric SimSci, was used to simulate gasification and to predict the syngas composition and yield of products. PRO/II™ is a steady state and dynamic process simulator used in chemical process design and operational analysis. The functionality of PRO/II™ is similar to that of other commercial tools, such as Aspen Plus and Aspen HYSYS, deploying simulations via graphical user interfaces (GUI) and programming subroutines in Fortran [31]. Notwithstanding the versatility of PRO/II™ in chemical process engineering, gasification literature is scarce compared with the literature for AspenTech tools [32]. Therefore, the current methodology also adopted the simulation flowsheet reported by Mutlu et al. [15] for the modeling of biomass gasification in Aspen Plus.

3.1. Simulation Assumptions

Simplifications were made in the model development to fit the limitations of the simulation software [33], reduce the computational demand, and define unknown data [15]. The assumptions follow those reported in former studies [20]. For simplification, biomass is considered free of inorganic matter and initial water content. This approach limits the model to predict the catalytic effect of ash during gasification or the reactivity of the moisture content of biomass during the pyrolysis and gasification [30]. The following assumptions were applied for both $StoicM$ and $KinetM$:

• The gasifier is a downdraft reactor with a capacity of 10 ton/h of woody biomass (daf) and a volume of 100 m³, reproducing the design conditions of [34].
• The residence time is sufficient to reach steady-state conditions and the simulation blocks are zero-dimensional.
• Temperature gradients are neglected, emulating an isothermal process.
• The biomass is considered as dry and ash-free (daf), containing only carbon, hydrogen, and oxygen, while char contains only carbon. The catalytic effect of ash is ignored.
• Gases are assumed ideal. Pressure and heat losses are ignored.
• Tar is considered a homogeneous organic hydrocarbon containing only carbon, hydrogen, and oxygen.
• The reaction takes place in the absence of oxygen and no internal combustion or oxidation reactions take place.

3.2. Thermodynamic Method and Component Properties

The properties of components were estimated using the thermodynamic method of the Peng–Robinson cubic equation of state, coded in PRO/II™ as PR01. This method enables the calculation of properties for non-polar substances with phase changes. Conventional components included H₂O, CO, CO₂, H₂, CH₄, and $Char$ (defined as pure carbon). The H₂O excludes the initial moisture content of biomass but corresponds to the steam used during the gasification ($H_2{O}_{in}$) and the water produced during pyrolysis ($H_2{O}_{pyro}$). Non-conventional compounds included biomass ($BM$), volatile matter ($VM$), fixed carbon ($FC$), and tar ($tar$). For simplification, the properties of $tar$ were assigned to the components $ta{r}_1$ and $ta{r}_2$ in reaction $R3$. The thermochemical properties of the non-conventional components are presented in

Table 1. Thermochemical properties of non-conventional components: biomass, char, and tar. N.A. means not applicable.

		BM				Char	Tar
		Wood Waste	Japanese	Hinoki	Japanese	Pure	Bio-Oil
Properties	Unit	Average	Cedar [23,35]	Cypress [36]	Red Pine [37]	Carbon [38]	[39]
$C_{daf}$	% g/g	51.4 ± 2	50.4 ± 2.6	51.9 ± 0	51.9 ± 1.5	100	54.7 ± 9.8
$H_{daf}$	% g/g	6.2 ± 0.3	6.2 ± 0.2	6.2 ± 0	6.3 ± 0.4	0	6.6 ± 0.3
$N_{daf}$	% g/g	0.3 ± 0.3	0.7 ± 0.4	0.2 ± 0	0.1 ± 0.1	0	<0.9
$O_{daf}$	% g/g	42.2 ± 2.1	43.4 ± 2.6	41.8 ± 0	41.6 ± 1.6	0	38.7 ± 11.4
$M{C}_{ad}$	% g/g	21.8 ± 7.5	30.3 ± 12.3	27.7 ± 0	7.4 ± 2.7	0	52.1
$V{M}_{daf}$	% g/g	84 ± 4.1	83 ± 5	84.8 ± 0	84.2 ± 3.1	0	N.A.
$F{C}_{daf}$	% g/g	16 ± 4.1	17 ± 5	15.1 ± 0	15.8 ± 3.1	100	N.A.
$HH{V}_{daf}$	MJ/kg	21.1 ± 0.9	19.8 ± 1	23 ± 0	20.7 ± 0.8	39.35	37.3 ± 0.8
${\rho }_{298K,daf}$ [40]	kg/m3	405 ± 140	497	481	238	1611	1130
PRO/II™ properties of unconventional components—CxHyOz
			$BM$	$VM$	$FC$	$tar$
x	atom/atom-C		1	1	1	1
y	atom/atom-C		1.49	1.79	0	1.43
z	atom/atom-C		0.73	0.77	0	0.53
$MW$	g/mol		23.83	26.24	12.01	21.98
$\Delta H_{298K}^0$ [40]	MJ/kg		−10	−8	−30	−5.5

.

The biomass used in the present study is Japanese waste wood. Former studies on Japanese waste management suggested a high availability of forestry residue (i.e., leftover treetops and branches) [7]. Forestry biomass quantification studies have concluded that Japanese cedar, Hinoki cypress, and Japanese red pine are the main tree species harvested [41]. For modeling purposes, the current study considered the properties of Japanese wood waste as an average across the above species. The properties of $tar$, including the ultimate composition, molecular weight ($MW$), enthalpy of formation ($\Delta H_{298K}^0$), and density (${\rho }_{298K,daf}$), were taken as constant values from Shrivastava et al. [39] and Popescu et al. [40]. The boiling point of $tar$ was estimated as being at 180 °C and 1 atm for the manipulation of separation units [17].

3.3. Development of a Modeling Flow Diagram

Figure 2 presents the flowsheet topology of the gasification mechanism reported in Section 2 and shows the designs following similar studies from Chmielniak et al. [42] and Nikoo et al. [43]. Simulation nodes have rhombus shapes and represent the streams of the components. $B{M}_{daf}$ (node 1) is converted into $FC$ (node 3) and $VM$ (node 4) in the decomposer reactor at 500 °C and 101 kPa, according to reaction $R1$. The decomposer block is a conversion reactor (RConv) for both $StoicM$ and $KinetM$, considering a biomass conversion of 100%. $FC$ is converted to the conventional component carbon. $VM$ flows to a reactor called the pyrolyzer, where reaction $R2$ takes place at 500 °C and 101 kPa. Hereafter, the reactor blocks (i.e., the pyrolizer, tar cracking reactor, heterogeneous gasifier, and homogeneous gasifier) differ for each model, $StoicM$ and $KinetM$, according to Section 3.4 and Section 3.5, respectively.

$H_2{O}_{pyro}$ (node 6) flows out of the system, simulating the conditions of evaporation in the pyrolysis zone of a downdraft gasifier. $ta{r}_1$ streams in the pyrolizer flow to the tar cracking reactor at 600 °C and 101 kPa, producing secondary tar ($ta{r}_2$) [9]. The gas produced in the pyrolizer and tar cracking reactor is added to $Ga{s}_{pyro}$ (node 5). $Ga{s}_{pyro}$ and $FC$ are mixed with the $H_2{O}_{in}$ (node 2) or $Strea{m}_{In}$ at 200 °C and 101 kPa. In the simulation, the steam-to-biomass mass ratio ($S/B$) ranges from 0 to 4 and is calculated with Equation (1). This premix to the gasifier ($Premi{x}_{Gas}$ and node 8) is heated to gasification temperature (T) ranging from 600 to 900 °C.

$$S/B=\frac{{\dot{m}}_{H_2{O}_{in}}}{{\dot{m}}_{BMdaf}}$$

(1)

The gasification occurs at 101 kPa, and the reduction zone is divided into two reactor blocks called the heterogeneous gasifier and homogeneous gasifier. $Premi{x}_{Gas}$ initially reacts according to the parallel heterogeneous reactions of Boudouard, $R4$, and water gas primary reaction, $R5$. The produced $Syng1$ (node 9), comprising the unreacted char and primary synthesis gases, flows to the homogeneous gasifier at constant temperature and pressure. This reactor converts $Syng1$ into the secondary synthesis gas $Syng2$ (node 10). The last section of the flowsheet comprises the separation of the unreacted steam $H_2{O}_{cond}$ (node 11) at 0 °C and a cyclone that separates the solid (unreacted $char$ and node 12) and gas phase $Syn{g}_{dry}$. Finally, $ta{r}_2$ and $Syn{g}_{dry}$ are mixed to produce the crude syngas ($Syn{g}_{crude}$ and node 13). $StoicM$ and $KinetM$ have the same simulation flow diagram (according to the following sections).

3.4. Stoichiometric Model $StoicM$

Table 2. Reaction conditions for model $StoicM$. The reaction parameters are taken from the literature; i.e., R1 and R2 are taken from [44], R3 from [40], R4 and R5 from [17], and R6 from [45]. T has a Kelvin unit.

Reactor Block (Rector Type)	Reaction		Reaction Parameters	Simulation Conditions
“DECO+PYRO” (RConv)	R1	${BM}_{daf}\to {\gamma }_{FC}FC+{\gamma }_{VM}VM$	${\gamma }_{FC}=0.34$ ${\gamma }_{VM}=0.75$	Fixed duty = 0 Conversion = 1
	R2	${\gamma }_{VM}VM\to {\gamma }_{CO}CO+{\gamma }_{C{O}_2}C{O}_2+{\gamma }_{H_2}{H}_2+{\gamma }_{C{H}_4}C{H}_4+{\gamma }_{ta{r}_1}ta{r}_1+{\gamma }_{H_2{O}_{pyro}}{H}_2{O}_{pyro}$	${\gamma }_{VM}=0.75$, ${\gamma }_{CO}=0.099$, ${\gamma }_{C{O}_2}=0.077$, ${\gamma }_{H_2}=0.08$, ${\gamma }_{C{H}_4}=0.004$, ${\gamma }_{ta{r}_1}=0.3931$${\gamma }_{H_2{O}_{pyro}}=0.26$
“Tar cracking” (RConv)	R3	$ta{r}_1$$\to {\gamma }_{CO}CO+{\gamma }_{C{O}_2}C{O}_2+{\gamma }_{C{H}_4}C{H}_4+{\gamma }_{H_2}{H}_2+{\gamma }_{ta{r}_2}ta{r}_2$	${\gamma }_{CO}=0.526$, ${\gamma }_{C{O}_2}=0.0987$${\gamma }_{C{H}_4}=0.18$, ${\gamma }_{ta{r}_2}=0.248$	Fixed duty = 0 Conversion = 0.72
“Heterogenous Gasifier”	R4	$C+{CO}_2\to 2CO$	$ln\left(K_P\right)=21.335-\frac{20743}{T}$	Fixed duty = 0
(REquil)	R5	$C+H_{2}O\to CO+H_2$	$ln\left(K_P\right)=17156-\frac{16194}{T}$	Activity basis = partial pressure
“Homogenous Gasifier” (REquil)	R6	$CO+H_{2}O\leftrightarrow {CO}_2+H_2$	$ln\left(K_P\right)=4.33-\frac{4577.8}{T}$	Fixed duty = 0 Activity basis = partial pressure

presents the simulation conditions and defines the reaction parameters for the model $StoicM$. The layout of the simulation in PRO/II™ is detailed in Figure A1 of Appendix A. The pyrolizer block is a conversion reactor (RConv) with a 100% conversion of $VM$. The stoichiometric coefficients ${\gamma }_i$ used for reaction $R2$ were calculated with Equation (2). The mass yield of each product ($y_i$) was taken from the product distribution of the pyrolysis of Japanese cedar at 500 °C [44]. Reactions $R1$ and $R2$ were performed in the same reactor, DECO + PYRO.

The tar cracking block is a conversion reactor (RConv) assuming a 72% conversion of $ta{r}_1$ [26]. Stoichiometric coefficients of $R3$ were taken for the tar lignin cracking [40]. The heterogeneous gasifier and homogeneous gasifier blocks are equilibrium reactors (REquil) fixed with an activity basis of partial pressure. The heterogeneous gasifier block takes equilibrium constants reported in the literature for both $R4$ and $R5$ (where T has Kelvin unit) [17]. Carbon is in solid phase and its activity exponent is zero. Finally, the homogeneous gasifier block takes equilibrium constants for $R6$ from [45].

$${\gamma }_i=y_i{\gamma }_{VM}\frac{M{W}_{BM}}{M{W}_i}$$

(2)

3.5. Kinetic Model $KinetM$

Table 3. Reaction conditions for the model $KinetM$. The reaction parameters are taken from the literature; i.e., R1 is taken from [44], R2 and R3 from [26], and R4 to R6 from [16]. T has a Kelvin unit.

Reactor Block (Rector Type)	Reaction		Reaction Parameters	Simulation Conditions
“Decomposer” (RConv)	R1	${BM}_{daf}\to {\gamma }_{FC}FC+{\gamma }_{VM}VM$	${\gamma }_{FC}=0.34$ ${\gamma }_{VM}=0.75$	Fixed duty = 0 Conversion = 1
“Pyrolyzer” (CSTR)	R2	$VM\to {\gamma }_{CO}CO+{\gamma }_{C{O}_2}C{O}_2+{\gamma }_{H_2}{H}_2+{\gamma }_{C{H}_4}C{H}_4+{\gamma }_{ta{r}_1}ta{r}_1+{\gamma }_{H_2{O}_{pyro}}{H}_2{O}_{pyro}$	${r_j=k_j\exp \left(-\frac{E_{a,j}}{RT}\right)\left(w_{VM}{\rho }_g\right)}_{}$$k_{CO}=9\times {10}^9$ / $E_{a,CO}=110$$k_{C{O}_2}=5.23\times {10}^9$ / $E_{a,C{O}_2}=105$$k_{H_2}=4.73\times {10}^4$ / $E_{{a,H}_2}=92.5$$k_{H_2{O}_{pyro}}=3.6\times {10}^{13}$ / $E_{a,H_2{O}_{pyro}}=150$$k_{C{H}_4}=1.09\times {10}^5$ / $E_{a,C{H}_4}=71.3$$k_{ta{r}_1}=2.09\times {10}^{10}$ / $E_{a,ta{r}_1}=112.7$	$\left\{\begin{array}{c}{k}_j\left(1/s\right)\\ E_j\left(kJ/mol\right)\end{array}\right\}$ Fixed duty = 0 Reactor volume = 5 m ${}^3$
“Tar cracking” (CSTR)	R3	$ta{r}_1$$\to {\gamma }_{CO}CO+{\gamma }_{C{O}_2}C{O}_2+{\gamma }_{C{H}_4}C{H}_4+{\gamma }_{H_2}{H}_2+{\gamma }_{ta{r}_2}ta{r}_2$	$r_{jcrack}={\gamma }_j{\left(10\right)}^{4.98}\exp \left(\frac{-93.37}{RT}\right)\left(w_{ta{r}_1}{\rho }_g\right)$${\gamma }_{CO}=0.563$, ${\gamma }_{C{O}_2}=0.111$${\gamma }_{H_2}=0.017$, ${\gamma }_{C{H}_4}=0.088$${\gamma }_{ta{r}_2}=0.2$	Fixed duty = 0 Reactor volume = 1 m ${}^3$
“Heterogenous Gasifier”	R4	$C+{CO}_2\to 2CO$	$A_r{T}^{\beta }=589T$ / $E_a=222.8$	Fixed duty = 0
(CSTR)	R5	$C+H_{2}O\to CO+H_2$	$A_r{T}^{\beta }=5.714T$ / $E_a=129.7$	Reactor volume = 100 m ${}^3$
“Homogenous Gasifier” (CSTR)	R6	$CO+H_{2}O\leftrightarrow {CO}_2+H_2$	$A_r{T}^{\beta }=2$.78 / $E_a=125.6$	Fixed duty = 0 Reactor volume = 1.5 m ${}^3$

provides the simulation conditions and defines the reaction parameters for $KinetM$. The flowsheet of the simulation in PRO/II™ is detailed in Figure A2 of Appendix A. As in the $StoicK$ model, the decomposer block is (RConv) considering a biomass conversion of 100%. For simulation purposes, the upstream reactors used in $KinetM$ follow the PRO/II definition of the continuous-stirred-tank-reactor (CSTR) block.

The pyrolizer and tar cracking blocks used a kinetic definition of the pyrolysis for beechwood and tar reforming [26]. The heterogeneous gasifier and homogeneous gasifier blocks followed the kinetic formulation reported by Yu et al. [16]. Indeed, the reaction rates of $R4$ to $R6$ were defined with Equations (3) and (4) ($E_a,r$ in units of kJ/mol and $k_r$ in units of 1/s). Accordingly, the blocks were set with an activity basis of the molar concentration, and the activity exponent of carbon is zero.

$$r_{i,j}=k_r\prod _{i=1}^N{\left[c_{i,j}\right]}^{{\alpha }_{i,r}}$$

(3)

$$k_r=A_r{T}^{\beta }exp\left(-E_a/RT\right)$$

(4)

This study acknowledges that a real gasifier is not completely represented by a CSTR reactor. Accordingly, efforts are made to emulate the gasification using CSTR blocks by adjusting simulation parameters. CSTR blocks are thus made to reflect the residence time of the components within the reactor by specifying a reactor volume. This residence time plays an important role in carbon conversion to syngas (X) [30]. For simulation purposes, the volume of the reactor was manually adjusted to fit the yield of products, the residence time, and X reported in the literature [16].

3.6. Methodology of Model Validation

The present paper uses experimental data reported in the literature for comparison and later model validation.

Table 4. Results from the literature used in model comparison.

Source	Gasifier	Feedstock	Reaction Agent	T Range	S/B Range	Ref.
Lit. 1	Downdraft	Japanese cedar	Steam	500–900	0.6–3.2	[46]
Lit. 2	Downdraft	Red Pine	Steam	700–900	0.7–2.8	[18]
Lit. 3	Dual Fluidized	Wood pellets	Steam	770–850	0.75–1.6	[22]
Lit. 4	Fluidized bed	Waste wood	Steam	750–950	0.7	[21]
Lit. 5	Fixed bed	Waste wood	Steam	600–900	3.5	[47]
Lit. 6	Fixed bed	Japanese cedar	Steam	650–850	2.4–12	[23]
Lit. 7	Updraft	Japanese cedar	Steam	750	0.6–3.2	[35]
Lit. 8	Updraft	Japanese cedar	Air/Steam	850	0–1.4	[24]
Lit. 9	Updraft	Wood waste	Steam	950	2–4	[34]

summarizes former studies that evaluated the steam gasification of Japanese woody biomass for different process parameters (i.e., T and $S/B$). Only Lit. 1 and Lit. 4 are used for the validation of the syngas composition and yield of products because they evaluate steam gasification under similar conditions, reporting complete in-service plant records (e.g., syngas composition, the yield of products, and statistic control). Indeed, Lit. 9 is a subsequent study of Lit. 1 and reports the scaling-up results for the steam gasification with Japanese woody biomass. The last approach increases the comparability of simulation results with experimental data, acknowledging the limitation of using different data sources of Lit. 1 and Lit. 4. Thus, the remaining complementary literature is used only for comparison purposes. Lit. 2 [46] is included to complement the data reported in Lit. 1 [18] for syngas composition during the steam gasification in downdraft reactors. Likewise, Lit. 3 enhances the analysis of the effect of temperature on the syngas composition [22]. Unfortunately, these studies did not report product yields or carbon conversion. Hence, Lit. 4 [21] and Lit. 5 [47] were included to extend the revision to different gasifiers. Additionally, the inclusion of Lit. 6 allows analysis for a high steam content (i.e., $S/B$ above 3) [23]. Lit. 1 to Lit. 6 were studies conducted on a laboratory scale, whereas Lit. 7 [35], Lit. 8 [24], and Lit. 9 [34] allow the comparison of gas composition in large-scale gasifiers.

For the model validation, this work adopts the definition of the systematic error ($Bias$) and total allowable error ($TEa$) proposed by Squara et al. [48]. $TEa$ measures the simulation error by considering the dispersion of experimental data, because of the accuracy and calibration of equipment, or uncontrolled testing conditions [49]. The $Bias$ is the absolute difference between a reference simulation value $y_{calc,i}$ and the mean value of the experimental results from the literature $y_{lit,i}$. $TE{a}_i$ includes the coefficient of variation ($C{V}_i$) of the experimental results according to Equation (5). The root mean square ($RMS$) is used to quantify the accumulative difference (i.e., error) between models and experimental data of Lit. 1 [46] and Lit. 4 [21], according to Equation (6).

$$TE{a}_i=|y_{calc,i}-y_{lit,i}|+1.65C{V}_i$$

(5)

$$RMS=\sqrt{\frac{{\sum }_{i=1}^N{\left(TE{a}_i\right)}^2}{N}}$$

(6)

The average $C{V}_i$ for the syngas composition reported in Lit. 1 is 2.6%, 1.26%, 8.27%, and 8% for H₂, CO, CO₂, and CH₄, respectively [46]. In addition to the molar composition of the syngas, this study used X and $y_{products}$ to compare the simulation results with the experimental results reported in the literature on a dry and ash-free basis and according to Equations (7) and (8). The average $C{V}_i$ for the yield of products reported in Lit. 4 is 9.09%, 3.32%, and 6.21% for X, $y_{gas}$, and $y_{tar}$, respectively [22]. Lit. 2, Lit. 3, and Lit. 5 do not report the standard deviation of experimental data.

$$y_{product,i}=\frac{{\dot{m}}_i}{{\dot{m}}_{B{M}_{daf}}}\times 100$$

(7)

$$X=\frac{{\dot{m}}_{B{M}_{daf}}F{C}_{daf}-{\dot{m}}_{Char}}{{\dot{m}}_{B{M}_{daf}}F{C}_{daf}}\times 100$$

(8)

4. Results and Discussion

4.1. Material Balance and Composition

Table 5. Material balance and composition for the model $StoicM$ at a temperature of 850 °C and $S/B$ of 0.7. * The molar composition of $VM$ corresponds to the products of the pyrolysis reaction after $R2$.

					Molar Composition
Node	Stream	$\dot{\mathit{m}}$	T	Pressure	C	Tar	H2O	H2	CO	CO2	CH4
	Unit	ton/h	°C	kPa	mol/mol
Steams In
1	$BM$	10.0	30	101
2	$SteamIn$	7.0	200	101			1.0
Pyrolysis
3	$FC$	1.70	500	101	1.0
4	$VM$ *	8.30	500	101		0.429	0.287	0.087	0.108	0.085	0.004
5	$Ga{s}_{pyro}$	5.30	0	101			0.003	0.164	0.507	0.215	0.112
6	$H_2{O}_{pyro}$	1.98	0	101			1.0
7	$ta{r}_2$	1.026	0	101		1.0
Gasification
8	$Premi{x}_{Gas}$	13.81	850	101	0.195		0.522	0.046	0.144	0.061	0.032
9	$Syng1$	13.81	850	101	0.026		0.250	0.318	0.211	0.163	0.032
10	$Syng2$	13.81	850	101	0.026		0.324	0.245	0.285	0.089	0.032
11	$H_2{O}_{cond}$	4.20	0	101			1.0
12	$Char$	0.224	0	101	1.0
13	$Synga{s}_{crude}$	10.41	0	101		0.089	0.003	0.341	0.398	0.125	0.044

and

Table 6. Material balance and composition for the model $KinetM$ at a temperature of 850 °C and $S/B$ of 0.7. * The molar composition of $VM$ corresponds to the products of the pyrolysis reaction after $R2$.

					Molar Composition
Node	Stream	$\dot{\mathit{m}}$	T	Pressure	C	Tar	H2O	H2	CO	CO2	CH4
	Unit	ton/h	°C	kPa	mol/mol
Steams In
1	$BM$	10.0	30	101
2	$SteamIn$	7.0	200	101			1.0
Pyrolysis
3	$FC$	1.70	500	101	1.0
4	$VM$ *	8.30	500	101		0.544	0.178		0.247	0.026	0.005
5	$Ga{s}_{pyro}$	6.53	0	101			0.001	0.231	0.603	0.033	0.132
6	$H_2{O}_{pyro}$	1.14	0	101			1.0
7	$ta{r}_2$	0.632	0	101		1.0
Gasification
8	$Premi{x}_{Gas}$	15.04	850	101	0.170		0.455	0.087	0.226	0.012	0.050
9	$Syng1$	15.04	850	101	0.031		0.315	0.226	0.366	0.012	0.050
10	$Syng2$	15.04	850	101	0.031		0.211	0.330	0.262	0.116	0.050
11	$H_2{O}_{cond}$	3.13	0	101			1.0
12	$Char$	0.304	0	101	1.0
13	$Synga{s}_{crude}$	12.23	0	101		0.044	0.003	0.415	0.330	0.146	0.063

give the material balance and composition of a baseline simulation for the models $StoicM$ and $KinetM$, respectively. The baseline conditions are $T=$ 850 °C and $S/B=$ 0.7. The nodes used in the material balance correspond to the topology presented in Figure 2. Mass flows (${\dot{m}}_i$) are presented for the case study of a gasifier having a biomass (daf) consumption of 10 ton/h.

Pyrolysis converts 11% to 19% of the initial mass of biomass into $H_2{O}_{pyro}$, whereas 17% of the initial mass is converted into carbon for both models. After pyrolysis and tar cracking, 53% to 65% of the biomass mass is converted into $Ga{s}_{pyro}$ and almost 10% is converted into secondary tar. According to $KinetM$, approximately 85% of the primary tar is converted into $ta{r}_2$ during tar cracking. This value is higher than the assumed conversion of 75% in $StoicM$. Consequently, $StoicM$ predicts a lower ${\dot{m}}_{gas}$ than does $KinetM$, because the stoichiometric coefficients of $R2$ and $R3$ promote a higher production of $H_2{O}_{pyro}$ and $ta{r}_2$ with the model $StoicM$. $Ga{s}_{pyro}$ presents a molar content of H₂, CO, and CO₂ above 86%. Compared with $StoicM$, the kinetic model predicts a lower content of CO₂ in $Ga{s}_{pyro}$, because of the lower value of ${\gamma }_{C{O}_2}$ used in the kinetic tar cracking model (presented in Table 3).

The pyrolysis product distribution is similar to that reported in the literature [44]. The errors in the distribution of pyrolysis products predicted with $KinetM$ and $StoicM$ are, respectively, 7.2% and 6.4% [50]. The highest error is observed for $FC$ owing to the assumption of the complete conversion of the solid phase into pure carbon. As a consequence of the product distribution after pyrolysis and tar cracking, $StoicM$ produces almost 12% less $Ga{s}_{pyro}$ than does $KinetM$. This flow difference in the reactants results in lower reactivity during the gasification with $StoicM$, as evidenced by a higher unreacted ${\dot{m}}_{H_2{O}_{cond}}$ for this model. Indeed, $StoicM$ seems to be more affected by the concentration of the reactants, in agreement with Le Chatelier’s principle and the formulation of an equilibrium model [20].

While pyrolysis and tar cracking produce almost 53% g/g_BM of $Synga{s}_{crude}$, the remaining 47% g/g_BM reacts in the gasification zone, as a result of the reaction of $Char$ and $Ga{s}_{pyro}$ in $R4$ to $R6$. Almost 3% of the initial biomass remains as unreacted carbon after the heterogeneous gasifier. The differences in ${\dot{m}}_{char}$ between $StoicM$ and $KinetM$ reveal a higher char conversation X in the former model. Indeed, the remaining char in $Stoic$ is 13% g/g whereas that in $KinetM$ is 18% g/g. The former variation is due to the equilibrium conditions of $R4$ and $R5$ in the stoichiometric model, where the char residence time in the reactor is ignored, and char conversion is overestimated [15]. In contrast, the reactor volume regulates the reaction rate in $KinetM$, providing a more realistic approach, with similar data reported in the literature [12].

The heterogeneous gasifier, as the final reactor, exposes different behaviors between models under the baseline conditions. In $KinetM$, the WGS reaction reduces the molar content of CO while increasing the contents of H₂ and CO₂. In contrast, $StoicM$ adopts the reverse definition of the WGS reaction under baseline conditions, suggesting an equilibrium that reduces the contents of H₂ and CO₂. According to Sharma et al. [27], a rise in temperature above 770 °C reduces the hydrogen yield, in agreement with Le Chatelier’s principle for the exothermic reaction $R6$. The effects of T and $S/B$ on the syngas composition and the equilibrium conditions are discussed in the following sections to clarify the role of the reduction zone.

4.2. Model Validations

The models are validated by presenting the effects of T and $S/B$ on the distribution of the products. The effect of $S/B$ is assessed in the range from 0 to 4 with steps of 0.35 for a constant T of 850 °C. The effect of T is assessed in the range from 600 to 900 °C with steps of 50 °C for a constant $S/B$ of 0.7. The prediction error of each model is presented as $RMS$ for each gasification parameter (T and $S/B$) relative to results from the literature.

4.2.1. Effect of the Steam-to-Biomass Ratio ($S/B$)

Figure 3 presents the effect of $S/B$ on the syngas composition at $T=$ 850 °C, comparing $KinetM$ and $StoicM$. The gas composition is reported on a dry basis (free of tar and water). Lit. 1 and Lit. 2 results are taken from [18,46], respectively.

The composition of the syngas at $S/B=0$ illustrates the effect of pyrolysis and tar cracking blocks, specifically low char conversion below 10%. In this state of gasification, CO has the highest molar content among the constituents in the syngas for both models. A rise in the steam flow promotes the production of H₂ and CO₂ while reducing the contents of CO and CH₄. The profile trends are similar for $StoicM$ and $KinetM$ but with a sharp asymptotic behavior occurring for $KinetM$. This rapid change in the molar composition at $S/B$ below 2 for $KinetM$ is attributed to the concentration dependency of $r_{i,j}$ (Equation (3)) and not the partial pressure of $StoicM$ (Table 2) [17]. In Figure 3, both models reproduce the trends of experimental results reported by Umeki et al. [46], and Huang et al. [18]. $KinetM$ makes a better prediction of CO and H₂ contents whereas $StoicM$ better reproduces variations in CO₂ and CH₄ contents. Both models overestimate the molar fraction of CO₂ relative to results obtained by [20] (for an equilibrium reactor) and [16] (for Gibbs and CSRT reactors).

4.2.2. Effect of the Gasification Temperature (T)

Figure 4 presents the effect of T on the syngas composition (free of tar and water) at $S/B=$ 0.7, comparing $KinetM$ and $StoicM$. Raising T increases the yield of H₂ whereas it reduces the contents of CO and CH₄. The latter behavior is attributed to the greater conversion of char, boosting the roles of the heterogeneous reactions $R4$ and $R5$ [15]. Indeed, the water gas primary reaction ($R5$) increases the production of H₂ while consuming CO₂ and competing with the WGS reaction $R6$ [12]. Consequently, variations in the molar content of CO₂ do not follow clear trends in either the modeling or experimental results. The higher content of H₂ and the reduced content of CO after 700 °C suggest a leading role of $R4$ over the WGS and Boudouard reactions. Excluding CO₂, the profile trends are similar for $StoicM$ and $KinetM$.

In contrast to the effect of $S/B$, there is no asymptotic behavior within the temperature range simulated. Additionally, the numerical difference between the results of $StoicM$ and $KinetM$ is notably higher when predicting $w_{CO}$ and $w_{C{O}_2}$ in the lower range of T. The latter disparity suggests differences in the reactivity of the char assumed in the models. Both models overpredict the molar content of H₂ and CO reported in the literature. Furthermore, Gonzalez et al. [20] reported a similar overestimation with equilibrium models related to differences in the experimental formation of secondary tar. A more extensive revision of tar reforming and cracking is advised in future simulation studies.

The char conversion affects the discrepancies between the literature and models. Figure 5 presents the effect of T on X and the yields of products ($y_{gas}$ and $y_{tar}$) at $S/B=$ 0.7, comparing $KinetM$ and $StoicM$. There is a notable difference between the experimental and simulation char conversions at temperatures near 800 °C. Nevertheless, a better fitting of X was observed at 600 and 900 °C. Compared with the simulation results, a lower X in the experimental data suggests the lower production of CO and H₂ during gasification at 800 °C. Variations in X explain the better prediction of the syngas composition at higher temperatures. Furthermore, the former difference in X explains the inaccuracy of $StoicM$ in predicting $w_{CO}$ and $w_{C{O}_2}$ in the lower T range.

The estimation of $y_{gas}$ made using $KinetM$ is higher than that made using $StoicM$. This discrepancy is explained by a higher production of gas during pyrolysis and tar cracking (as reported in Table 5 and Table 6). The stoichiometric model seems to better predict $y_{gas}$ at lower temperatures, but the fitting with experimental results is enhanced at temperatures near 900 °C. The results of $y_{tar}$ obtained using the two models are similar and follow the literature data. The following section discusses the global performance of the models in detail, quantifying the discrepancies with the literature data and suggesting general trends in the fitting errors.

4.2.3. $RMS$ Prediction Error of the Models

Table 7. The errors $RMS$ of models $StoicM$ and $KinetM$ in the prediction of the syngas composition ($w_{gas,i}$) and mass yield of the products ($y_{product,i}$) when evaluating T and $S/B$.

	Error vs. T						Error vs. $\mathit{S}/\mathit{B}$
	$\mathit{RMS}-{\mathit{w}}_{\mathit{gas}}$			$\mathit{RMS}-{\mathit{y}}_{\mathit{product}}$			$\mathit{RMS}-{\mathit{w}}_{\mathit{gas}}$
	$\mathit{KinetM}$	$\mathit{StoicM}$		$\mathit{KinetM}$	$\mathit{StoicM}$		$\mathit{KinetM}$	$\mathit{StoicM}$
Unit	% mol/mol			% g/g			% mol/mol
$H_2$	1.28	2.91	X	9.34	14.85	$H_2$	1.53	3.14
$CO$	5.24	13.88	$Y_{gas}$	9.66	22.14	$CO$	2.47	9.77
$C{O}_2$	4.01	5.49	$Y_{tar}$	4.38	8.79	$C{O}_2$	4.72	1.52
$C{H}_4$	3.05	1.22				$C{H}_4$	2.48	2.88
${\mathit{RMS}}_{\mathit{global}}$	3.91	7.96		8.16	16.21		2.91	5.62

presents the errors of models $StoicM$ and $KinetM$ in the prediction of the syngas composition ($w_{gas,i}$) and mass yield of the products ($y_{product,i}$) when evaluating T and $S/B$. $KinetM$ has lower global error ($RM{S}_{global}$) than $StoicM$ under all evaluation conditions. Both models are more accurate in predicting the effect of $S/B$ than in predicting the effect of T. Indeed, the evaluation of $S/B$ using $KinetM$ has the best global fitting with the experimental data of $w_{gas,i}$. In contrast, the prediction of the effect of T on $w_{gas,i}$ has a higher error using $StoicM$, mainly because of the overestimation of CO content. The error in the prediction of $y_{product,i}$ is higher than that in the prediction of $w_{gas,i}$, as the stoichiometric approach has a global $RMS$ error exceeding 12%.

$RM{S}_{global}$ reported in the literature for predicting $w_{gas,i}$ ranges from 2.1% to 14% mol/mol when using stoichiometric models [16] whereas $RM{S}_{global}$ for kinetic models ranges from 1.3% to 4.4% mol/mol [16]. As an additional comparison, $RM{S}_{global}$ reported in the literature for non-stoichiometric models (using Gibbs reactors) ranges from 5.15% to 11% mol/mol, similarly to the error range for stoichiometric models [51]. The simulation error is the largest for CO₂ among the constituents, but this error has not been discussed extensively. As presented in Section 4.1 and reported by Gonzalez et al. [20], the models are sensitive to overestimations of char conversion and tar cracking and reforming. These limitations suggest that kinetic models that include Langmuir–Hinshelwood mechanisms should be implemented.

5. Conclusions

The simulation of the downdraft gasifier with Japanese wood waste in PRO/II™ allowed the prediction of the syngas molar composition ($w_{gas,i}$) and the yield of products ($y_{product,i}$) with $RMS$ errors below 7.96% mol/mol_gas and 16.21% g/g_BM, respectively. A comparison between the fitting of kinetic and stoichiometric models to experimental data confirmed a higher accuracy when using the former model. Indeed, the adoption of a simpler stoichiometric model reduced the accuracy of the prediction of $w_{gas,i}$ and $y_{product,i}$, increasing the $RMS$ error by 4.05% mol/mol and 8.05% g/g_BM, respectively. Both models better reproduced experimental results for changes in the steam-to-biomass ratio than for temperature variations. The error of the evaluated models is ascribed to an overestimation of char conversion in heterogeneous reactions, as well as the excessive production of pyrolysis gas previous to the gasification zone.

The results of the present study support recommendations made in the literature to use equilibrium models for basic relationships between operating conditions and kinetic models for more precise modeling evaluations. The current study makes an additional contribution by measuring and suggesting the simulation error when using Kinetic or Stoichiometric models Still, this paper acknowledges the limitations of the error definition and the narrow comparison of a single gasification case of study. Consequently, a further revision of different gasification scenarios is advised. Such a revision should include extended model validation with systematic reviews, contrasting non-equilibrium and dimensional models, and the evaluation of different biomass sources and reaction mechanisms.