Introduction

The energy requirement is increasing on a global scale as a result of industrialization and population growth. With the decrease in natural resources in line with the energy requirement, the need for alternative energy sources has emerged. Landfill, which is the lowest step of the waste management hierarchy depending on the principle of sustainable development, has started to be replaced by energy recovery processes today (Rigamonti et al. 2016; Asefi and Lim 2017). Energy recovery from waste will inevitably become the dominant option in the future with the decrease in fossil resources and population growth. Combustion, pyrolysis, and gasification are widely used as thermal energy recovery methods (Wissing et al. 2017; Gurgul et al. 2018). Energy production by burning organic materials is almost as old as human history. The combustion of organic substances can be defined as the chemical reaction of oxygen with flammable elements. Since high amounts of oxygen are used during combustion, fly ash, metal oxides, particulate matter, and emissions that are difficult to control such as SOx, NOx, CO occur in high quantity (Chen et al. 2017; de Foy 2018; Tumsa et al. 2018).

Gasification is the degradation of carbonaceous organic materials at high temperatures after undergoing thermochemical transformation using partial oxygen. Gasification is a clean and efficient process compared to combustion (Bartela et al. 2017; Kabalina et al. 2017; Watson et al. 2018). The main purpose of gasification is to provide a reaction of the carbon in the organic substances with the gasification agents to obtain a gas product with a high heating value (HV). The content and amount of gasification products vary according to the type of organic matter, its activity, the type of gasification agent, and the application method (Ongen et al. 2019). Syngas produced when air is used for partial oxygen during gasification containing high hydrocarbon gases of CO, CO2, H2, CH4, N2, as well as particles and tar (Lv et al. 2007; Sforza et al. 2012; Yoon and Lee 2012; Ongen et al. 2016; Cerone et al. 2020). The difference of the gasification process from the combustion is that the gasification is carried out under partial oxygen conditions and the heat and synthesis gas produced in the system is used as the feed materials of the fuel system.

The main factors affecting the gasification processes are raw material property, temperature, pressure, gasification agent, and equivalence ratio (Said et al. 2020). Biomass selection is of great importance to achieve successful results in gasification. Almost all biomass containing carbon can be used as fuel in experimental conditions. However, it is not always possible to achieve the desired results. This is because when the humidity in the raw material exceeds 30%, ignition becomes difficult and the HV of the product decreases. If the fuel with high humidity reaches the oxidation zone, it will lose heat to remove the moisture. The high rate of mineral matter also impedes gasification. Since the oxidation temperature is higher than the melting point of ash, it causes slag formation, which leads to clogging of the feed. In cases where the ash content exceeds 5%, slag formation reaches dangerous levels and in this case, new mixtures with low alkali value and high dust content may emerge (Pangaliyev 2014).

Since the gasification process is energetically an auto-thermal process, no extra thermal input (heat source) is required. Gasification reactions are various endothermic reactions supported by the energy released as a result of combustion. Combustible gases such as CO, H2, and CH4 that are released from gasification emerge as a result of these reactions. Gasification reactions are water–gas reaction, Boudouard reaction, Shift transformation, and Methanization reaction. The basic stages of gasification are drying, pyrolysis, reduction, and oxidation. The duration and size of these stages vary according to the gasifier reactor feature. The most important pillar of gasifier design is determining the reactor type (Kirsanovs et al. 2016). Since the carbon conversion, temperature distributions throughout the reactor, and the number of particles in the product gas will change depending on the reactor type chosen, the composition and HVs of the final products to be obtained are directly affected by the reactor type. Reactor types are divided into three categories as fixed-bed gasifiers, fluidized-bed gasifiers, and entrained-flow gasifiers (Kumar et al. 2009; Das et al. 2020).

This study aimed to model the synthesis of gas and HV produced by gasification of the carbon black obtained from waste tires and sewage sludge originating from the yarn industry using the response surface method (RSM). Process optimization was carried out by modeling the effects of process parameters (gas flow rate, reaction time, and biomass amount) on the synthesis gas composition and HV. All experiments were conducted at Istanbul University-Cerrahpasa, Environmental Engineering Department’s Laboratories in the year 2019.

Materials and methods

Sampling and analytical procedures

In the study, commercial carbon black obtained from the pyrolysis of waste tires and polymeric-structured process waste with hazardous waste character originating from the yarn production industry were used as samples. Both samples are organic materials. Moisture content, ash content, burning loss, and solid substance content of carbon black samples used in experimental studies were conducted according to Standard Methods (SM 2540 B; SM 2540 E) (APHA 2005). Elemental analysis experiments were carried out with Thermo-Flash 2000 CHN-S elemental analyzer using the ASTM-D5373-16 method (ASTM-D53732018). IKA C200 Bomb Calorimeter was used to determine the calorific values of carbon black and the gasification products. The calorific value analyses were carried out according to the ASTM-D5865-13 method (ASTM D5865/D5865M-19 2013). ABB brand AO2020 continuous gas analyzer was employed to perform synthesis gas measurements. In gasification experiments, the volumetric percentages of CO, CO2, H2, CH4, and O2 in the produced gas were measured, and their calorific values were calculated depending on their volumetric percentages (Ongen et al. 2013; Ozcan et al. 2016; Ozbas et al. 2019).

Experimental setup

Gasification experiments were conducted in an up-flow fixed bed steel reactor with an inner diameter of 7 cm that was circulated employing a cyclone unit. The heating rate of the system is constant, and the reactor temperature can be controlled according to the operating conditions. The reactor has two gas inlet lines that are properly connected to dry air or pure O2 gas. Besides, there is an outlet line in the coverage area of the reactor to ensure the evacuation of the synthesis gas formed. The thermocouple was used to measure the internal temperature of the reactor during the experiments. Synthesis gas was directed to the gas analyzer to determine its composition, and the amounts of CO, CO2, H2, CH4, and O2 in it were measured as the percentage by volume. Gasification experiments were conducted at 700 °C using dry air as an agent. By keeping the temperature constant; the effects of agent flow rate, biomass amount, and test duration on experimental processes were investigated.

Experimental design

The experimental design applied to evaluate the system efficiency was carried out using Design Expert V.11.0.10. software. A factorial design was applied with three independent variables (gas flow rate, reaction time, and biomass amount) used in gasification experiments. The levels of the variables (low, medium, high) were arranged as − 1, 0, and + 1. The experimental levels of each variable were determined according to the results of the preliminary studies and values in the literature. Dependent variables (responses) were chosen as percentages of H2, CO, CH4, and HV. To estimate a pure error sum of squares, 15 experiment sets were planned in the design center with 3 repetitions. The target function, which is a function of independent variables, can be expressed with a quadratic polynomial model (Eq. 1).

$$Y = b_{0} + \mathop \sum \limits_{i = 1}^{n} b_{i} X_{i} + \mathop \sum \limits_{i = 1}^{n} b_{ii} X_{i}^{2} + \mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{n} b_{ij} X_{i} X_{j} + \varepsilon$$
(1)

Here, Y is the estimated response, b0 is the intercept parameter, bi, bii, and bij are linear, quadratic, and interaction factor effects, Xi and Xj are independent variables, and ε is the error. The compatibility for the regression model was evaluated with the coefficient of determination (R2, R2adj), the statistical significance level was checked with the variance analysis where the F value and the P-value were calculated. Model terms were selected or rejected based on probability value at 95% confidence level. Three-dimensional response surface graphs were drawn using these values to visualize the individual or interactive effect levels of the independent variables. The optimization step of the proposed method was completed using the Box–Behnken design (BBD). In this study, the process variables were optimized using the response surface method to reach the maximum HV in the gasification of the selected biomass. For the optimization of the variables, a three-factor 3-level (BBD) approach was used and the value ranges of the variables are given in Table 1. Within the scope of the study, a total of 15 experimental sets were employed to analyze the effects and interactions of independent variables.

Table 1 The coded and actual values of variables of the experimental design matrix for syngas composition
Full size table

Results and discussion

The physical and chemical properties of the raw material to be used as fuel in gasification processes have a direct impact on the process efficiency. Therefore, in the first stage of the study, the elemental analysis, HV, moisture content, and ash content of biomass to be used in gasification experiments were analyzed. Elemental analysis (C, H, N, S–O) and proximate analysis of waste tire samples are given in Table 2 together with the literature comparison. When the values given in Table 2 are examined, it is seen that the biomass mixture particularly fits the gasification process with its C and H content. Besides, low moisture and ash values are other indications that the samples conform to the process.

Table 2 Ultimate and proximate analysis of biomass samples
Full size table

The effects and simultaneous interactions of operating parameters (gas flow rate, reaction time, and biomass amount) on syngas composition and HV were investigated by applying BBD, and the values of the variables in the 15 experimental sets are presented in Table 3. Design Expert V.11.0.1.0 software program was used to analyze experimental data and to determine simultaneous interactions between independent variables and responses with ANOVA. The optimization of the syngas composition with the highest HV was aimed at with the model applied.

Table 3 The actual and predicted values of responses for H2, CH4, CO, and HV
Full size table

The fit and statistical significance of the developed model were evaluated using factors such as correlation coefficient, F-test, P-value, coefficient of variation (C.V.). (Dil et al. 2018; Tripathy et al. 2019). The correlation between independent variables and system responses is presented in the second-degree quadratic equations given in Eqs. (2–5). Pareto analysis, in which the contribution of each factor is calculated, provides information about the effect of operating conditions on the process. The graphical representation of Pareto analysis is given in Fig. 1. All of the linear parameters and gas flow rate in quadratic parameters are effective on H2 generation. Besides, the gas flow rate in linear parameters along with reaction time and biomass amount is more effective on CH4 generation, whereas all of the linear parameters and biomass amount in interactive parameters are effective on CO generation. Furthermore, gas flow rate and biomass amount among quadratic parameters are the most effective parameters on the amount of the HV. The results obtained by the Pareto analysis confirm the ANOVA results.

$${\rm{H}}_{2} {\rm{vol}}\% = + 12.56 + 1.64A + 2.31B + 1.66C - 3.88A^{2}$$
(2)
$${\rm{CH}}_{4} \;{\rm{vol}}\% = + 8.24 - 0.9725A - 1.62B - 2.38A^{2} - 2.40C^{2}$$
(3)
$${\rm{CO}}\;{\rm{vol}}\% = + 4.11 + 1.06A - 1.14B + 1.30C + 1.14C^{2}$$
(4)
$$\begin{aligned} {\rm{HV}},\frac{{{\rm{kcal}}}}{{{\rm{m}}^{3} }} &= + 1295.60 - 118.34B + 87.84C\\&\quad - 122.77BC - 352.61A^{2} - 217.14C^{2} \end{aligned}$$
(5)
Fig. 1
figure 1

Pareto chart showing the effect of independent variables and their interactions

Full size image

The results of the second-degree response surface method and quadratic statistical summary of the models in the ANOVA are given in Table 4. As seen in the ANOVA tables, the Probe > F (P-value) value being below 0.05 for all developed models confirms the significance of the models. The F values, which are expressed as mean square regression/mean square residual, highly indicate that the quadratic regression model is quite significant (Yetilmezsoy et al. 2009). The F values of all models were large, their probability values (P values) are smaller than 0.05, and the lack of fit values are greater than 0.05 (Table 4). This indicates that the models are statistically significant. The fit of the experimental data obtained for each response was verified with high R2 values. Having R2 values close to 1 indicates that each quadratic model is statistically acceptable (Asaithambi et al. 2016; Bashir et al. 2016). For a good degree of model fit, the R2 value is desired to be > 0.8 (Myers, 1999). It can be seen from Table 4 that the R2 values obtained for the generation of H2, CO, CH4 gases, and the amount of HV obtained are 92.86%, 95.40%, 96.15%, and 96.80%, respectively.

Table 4 ANOVA results obtained from the quadratic model for H2, CH4, CO, and HV
Full size table

R2 is defined as the measurement of the amount of reduction that occurs in the variability of the response obtained using the independent variables in the model. Testing the regression model's fit with the data cannot be justified by high R2 values only. Adjusted R2 is preferred to check the fit of the regression model as it does not always increase when the variables are added (Silveira et al. 2015). In addition to high R2 values, high adjusted R2 values are another indication that the proposed model is approaching the actual values. It can be seen from Fig. 2 that there is a satisfactory correspondence between the experimental data and the estimated values for all responses. The distribution of data points close to the relevant predicted values is emphasized with low standard deviation values. Normal probability plots evaluating whether residuals fit the normal distribution are given in Fig. 2. In such models, points are expected to accumulate around a line. As can be seen from Fig. 2, experimental data are scattered around the estimated values. The data show the possible normal distribution in certain model responses.

Fig. 2
figure 2

Normal probability plots of residuals and actual and predicted values of H2, CH4, CO, and HV

Full size image

It is the desired result that the Adequate Precision value is higher than 4 (Ölmez 2009), and the AP values were above 4 in all models. The fact that AP values of all models were greater than 4 in this study indicates that adequate signals could be used to navigate the design space and predict the responses (Mohajeri et al. 2010; Azmi et al. 2015; Bashir et al. 2016). The coefficient of variation can be defined as the ratio of the standard error of estimate to the mean value of the observed response and is a measure of reproducibility of the model. Although the desired value for CV varies depending on the research area, basically CV < 10 is very good, 10 < CV < 20 is good, 20 < CV < 30 is acceptable and CV > 30 is not acceptable. Low CV values indicate that deviations between experimental and predicted results are low. Besides, low CV values show not only a high level of precision but also a high degree of reliability of the experimental study conducted (Kuehl 2000; Prakash Maran et al. 2013; Silveira et al. 2015). It can be seen from Table 4 that the CV values of the model for H2, CH4, CO were in the range of 10–15, while it is below 10 for the HV.

The sum of squares of the relevant term is used to calculate the contribution percentages of each model component. To calculate the contribution percentages of first-degree, quadratic and interactive terms, the equations given in the study by Yetilmezsoy et al. (2009) were used. The total contribution percentages were calculated based on the sum of squares values of linear, interactive, and quadratic terms (Fig. 3). It can be seen from Fig. 3 that the first degree (linear) terms have the lowest contribution percentages for all responses. Quadratic terms have the highest contribution percentages for CH4 and HV, and interactive terms for H2 and CO. The high contribution percentage of interactive terms indicates that the combined behavior of these variables and complex reactions are dominant in the process. Based on the results, it can be interpreted that interactions between independent variables have a high effect on H2 and CO percentages. For all responses, it can be said that there is no direct and linear relationship between each parameter.

Fig. 3
figure 3

A detailed schematic diagram showing the percentage contributions of the components

Full size image

Three-dimensional (3D) response surface graphs as a function of two factors were drawn based on model equations to determine the individual and interactive effects of the factors that are desired to be monitored, keeping the other factors constant at certain values. In this study, since the regression model has three independent variables, one variable is kept constant in the center, and the change in the response is determined by monitoring the other two variables within the specified range. Nonlinear 3D response surfaces and relevant contour plots indicate that there is a significant interaction between each independent variable and the response (Yetilmezsoy et al. 2009). The visualized version of the regression model is expressed with response surface plots and contour plots.

The RSM plots given in Figs. 4, 5, 6 and 7 show less and more curvature depending on the significance level of the interactive and quadratic terms. Quadratic response surface graphs may exhibit different profiles. These profiles can be grouped in four as; (1) profile having a surface in which the maximum and minimum points are within the experimental region and having contour plot in the form of ellipse or circle form (2) profile having contour plot in the form of a hyperbolic system (profile in which the minimum or maximum point is not at the region that remains within the experimental range, but at the borders of the region, and in which the torsion point remains in between the relative minimum and relative maximum), (3) profile in which the maximum point is remaining outside the experimental region, and (4) profile that values of variable draw a plateau (in this profile, change at the level of the variable does not affect the system) (Nair et al. 2014).

Fig. 4
figure 4

3D plots surface of H2 vol (%) as a function of a Time and flow b Time and biomass, c Biomass and flow

Full size image
Fig. 5
figure 5

3D plots surface of CH4 vol (%) as a function of a Time and flow b Time and biomass, c Biomass and flow

Full size image
Fig. 6
figure 6

3D plots surface of CO vol (%) as a function of a Time and flow b Time and biomass, c Biomass and flow

Full size image
Fig. 7
figure 7

3D plots surface of HV (kcal/m3) as a function of a Time and flow b Time and biomass, c Biomass and flow

Full size image

The response surface and the corresponding contour plots drawn to evaluate the individual and interactive effects of the independent variables and their contribution in the prediction of the responses are analyzed to obtain maximum or minimum responses and relevant optimum conditions. Optimum conditions in multi-response systems should meet the required criteria for all parameters simultaneously. Since the maximum HV was aimed in this study, the optimized conditions were determined based on the maximum HV and indirectly the maximum values of the syngas components, and the optimum conditions are given in Table 5.

Table 5 Performance evaluation of the optimized experimental conditions
Full size table

Optimum conditions obtained with the help of the model are as follows: reaction time is 31.14 min, the gas flow rate is 0.05 L/minute and biomass amount is 19.66 g. Under optimum conditions, the percentages of H2, CH4, CO were estimated as 13.13%, 8.50%, and 8.09%, respectively, and the HV was 1445.03 kcal/m3. As a result of validation experiments carried out under optimum conditions, H2, CH4, CO percentages were determined as 12.75%, 8.07%, and 7.87%, respectively, and the HV was 1420.3 kcal/m3. The low error in the experimental and predicted values indicates good agreement of the results predicted by the models and obtained from validation experiments. A good correlation between the experimental values and predicted responses confirms the reliability of modeling by RSM.

Conclusion

This study aimed to obtain energy by gasification of a mixture of the carbon black obtained from waste tires and the polymeric biomass, which is a yarn industry waste, and to optimize the experimental process. Based on elemental analysis results, it was observed that both materials contain high levels of carbon and hydrogen. High carbon content and low moisture content values of carbon black and sewage sludge show that both materials are suitable for the gasification process. To maximize the H2, CO, CH4 content, and heating value of the synthesis gas, independent variables of the process were optimized by the response surface method and Box–Behnken design. The effects of the operational parameters on the system’s responses were assessed by response surface plots. Correlation coefficients of the second-degree polynomial equation were high with the values of 92.86%, 95.40%, 96.15%, and 96.80%, for H2, CO, CH4 content, and HV, respectively. High correlation coefficient values of the variance analysis indicate that the second-degree model conforms to a sufficient level with the experimental data. Under optimum conditions, H2, CH4, CO content of the syngas were 12.75%, 8.07%, and 7.87%, respectively, whereas the HV was determined as 1420.3 kcal/m3. The results of the study indicated that the gasification process is an appropriate treatment for obtaining high-quality syngas from waste materials with high carbon and low moisture content and the Box–Behnken design was successfully applied in the optimization of the gasification process.