Introduction

Finding environmentally friendly methods for handling biomass-rich municipal solid waste (MSW) is a particularly important subject worldwide (Siwal et al. 2020). Annually 1.3–1.6 billion tonnes of MSW is generated with a composition of 25–30% of paper and cardboard, 26–35% of organic waste, 5–12% of metal and glass, 7–10% of plastics and 2–4% of wood, textile and rubber (Glushkov et al. 2018). The amount of MSW could exceed 2.2–2.4 billion tonnes per year by 2025 (Veses et al. 2020). About a quarter of biomass-rich MSW is utilized; meanwhile, the remaining part is sent to landfilling or incineration (Gabbar et al. 2018). However, landfilling and incineration are not viable methods and cause serious environmental issues (e.g. emission of greenhouse gases and hazardous volatile compounds) (Sakaki et al. 2014). Therefore, biological, hydrothermal and thermochemical techniques were proposed in the literature for waste utilization (Sipra et al. 2018).

Among the thermochemical processes, two-step pyrolysis seems to be a promising method because of the improved contact between the derived pyrolysis products and catalyst and the higher hydrogen content of the produced syngas (Veses et al. 2020).

As a result of the above, several research groups have already studied the two-step pyrolysis technologies. Yao et al. (2018) carried out their experiments on nickel supported ZSM-5, Beta and Y catalysts. The examined Ni/ZSM-5 catalyst showed excellent thermal stability and coke resistance owing to its high acidity and adequate pore structure. Regarding the catalyst, similar findings were made by Wu and Williams (2009). However, their experimental results also confirmed that the temperature of the second stage and steam rate also have a significant influence on hydrogen production. Increasing the reaction temperature and steam rate usually involves higher hydrogen content, but finding the optimum is required.

Barbarias et al. (2018) found that type and characteristics of the raw material also greatly influence the reforming behaviour. During their experiments, the highest hydrogen production was observed for polyethylene (PE) and polypropylene (PP) (34.8 and 37.3%) which was followed by polystyrene (PS) (29.1%) and polyethylene terephthalate (PET) (18.2%). Czernik and French (2006) also reported a similarly high hydrogen production value for PP (34%). Namioka et al. (2011) were able to further enhance the hydrogen production using a ruthenium (Ru) catalyst. To understand the correlation between plastics and hydrogen contents, Cai et al. (2021) were also made experiments. Their results showed that two-step pyrolysis of PP, HDPE and LDPE on Fe/Al2O3 catalyst yielded 55–60% hydrogen content. In contrast, continuous reforming of the volatiles of biomass pyrolysis results in a hydrogen yield of about 1.6–10.2% (Santamaria et al. 2018).

Real MSW contains PP, PE, PET, PVC, PS, paper, organic waste, wood, glass and textile. The different components degrade in different ways and affect the yield and composition of the gas product (Du et al. 2021). Plastic waste has a high hydrogen/carbon ratio and low oxygen content. Therefore its pyrolysis mainly results in the formation of H2 and hydrocarbons. In contrast, biomass has a high oxygen content; therefore, its addition increases the CO and CO2 content of the pyrolysis gas (Kai et al. 2019).

To find the optimum for hydrogen production, many experiments are needed. However, response surface methodology (RSM) can be a promising way to reduce the number, time and cost of experiments required (Hemmati et al. 2019). In addition, the system itself and the reactions that take place are quite complex, which poses a serious challenge in standard kinetic modelling and parameter identification. Moreover, there is not enough information about the individual reactions to make a robust model. This was another important reason to consider RSM.

RSM was first proposed by Box and Wilson (1951). The methodology is based on the use of factors and outcomes and calculates correlations using multidegree functions. RSM allows the identification of key influencing factors and facilitates the determination of optimal operation intervals. The resulting expressions are not easy to extrapolate beyond the limits of the system. However, they can provide insight and estimates for further experiments and simulation steps.

RSM has already been used in some other research areas. Alaba et al. (2017) applied this technique to find the optimum formulation of hierarchical nanoporous HY zeolite. The effects of ageing, crystallization, and NaOH/kaolin ratio were investigated on specific surface areas and hierarchical factor. Each variable was statistically significant for high crystallinity and specific surface area, but the most statistically significant variable was the NaOH/kaolin ratio. Mohammed et al. (2017a, b) used RSM to interpret the effects of temperature, heating rate, and nitrogen flow rate on the pyrolysis of Napier grass biomass and to optimize the mentioned variables. It was found that temperature, nitrogen flow rate, and heating rate significantly affect the yield of gas and bio-oil, while the temperature mainly affects the biochar yield. The optimal parameters were 600 °C temperature, 50 °C/min heating rate, and 5 L/min nitrogen flow rate. Mohammed et al. (2017a, b) pyrolyzed Bambara groundnut shell in another publication. The parameters and the methodology were the same. It was found that the optimal bio-oil yield (36.5%) can be achieved at a temperature of 600 °C, a heating rate of 50 °C/min and a nitrogen flow rate of 11 L/min. Das and Gound (2021) used the RSM to optimize the process parameters of slow pyrolysis of rice husk. In order to determine the trend and relationship between the experimental responses and the process variables (temperature: 300–600 °C, nitrogen flow rate: 0.87–1.5 L/min, holding time: 20–60 min), a quadratic model was developed. Temperature showed the maximal effect among the studied variables, followed by holding time and nitrogen flow rate. The optimal conditions were 427 °C, 0.8 L/min and 45 min, resulting in a bio-oil yield of 35.5% (biomass conversion: 50.8%). Hanandeh et al. (2021) extended the RSM modelling with a significant evaluation of the particle size of oak acorn shells, deseeded carob pods and olive mill solid waste. They found that the effect of particle size on lower heating value (LHV) varies as a function of biomass. The highest LHV was observed in oak acorn shells when the particle size was below 1000 µm. In the case of olive mill solid waste, the larger particle size was the most advantageous.

Gupta et al. (2022) combined the RSM with an artificial neural network (ANN). The output of the pyrolysis process was predicted by ANN, while the significance and optimum of the process parameters (temperature, heating rate and inert gas flow rate) were determined by RSM. Their study revealed that the combinational approach is a good ability to model the pyrolysis process of pine needles. The most predominant variable was the temperature in terms of product yields, and the optimized process parameters were 552.06 °C temperature, 50 °C/min heating rate and 164.40 mL/min inert gas flow rate.

Singh and Tirkey (2021) developed a robust model for gasification in the ASPEN plus simulator. They optimized the gasification temperature (600–900 °C) and equivalence ratio (0.2–0.6) for hydrogen production and in terms of higher heating value (HHV). The optimal hydrogen production and HHV values are 0.1 and 3.96 MJ/kg, which corresponds to the optimized temperature at 887–879 °C and an equivalence ratio of 0.32. The composite desirability observed was 0.59.

Based on the literature, it can be concluded that RSM modelling is practically not used in the two-step pyrolysis of MSW to optimize the process parameters. Moreover, the composition of the waste varies from country to country, so it would be necessary to extrapolate RSM to different raw materials and systems and to determine the statistical significance of the parameters. These can significantly contribute to the research and development of the field.

Experimental

Based on the aforementioned, two-step pyrolysis of biomass-rich MSW was studied in a two-stage fixed-bed reactor system. RSM was used to understand the effects of temperature and steam rate on yields and composition of the gas products and to determine the significance of the parameters.

Raw materials

Shredded and milled real biomass-rich MSW was used as raw material. The raw material had carbon, hydrogen, oxygen, and nitrogen contents of 34.48, 4.70, 59.78 and 1.04%, respectively. The moisture and ash contents were 5.39 and 37.22%. The volatile matter was 62.35%, while the fixed carbon was 0.14%. To increase the gas yield and hydrogen content Ni/ZSM-5 synthetic zeolite catalyst (Si/Al molar ratio: 25.9, surface area: 425 m2/g) was used. The Ni/ZSM-5 catalyst was prepared by incipient wetness method using an aqueous nickel(II) nitrate hexahydrate (Ni(NO3)2·6H2O) solution. The sample was stirred continuously at 85 °C for 2 h, dried at 110 °C for 10 h and conditioned at 650 °C for 5 h in air. The SEM micrograph of the fresh Ni/ZSM-5 catalyst is shown in Fig. 1. It is well shown that the catalyst has a spherical shape with a similar size without any larger agglomerates. The nickel content of the catalyst was 2.08%.

Fig. 1
figure 1

SEM micrograph of the fresh Ni/ZSM-5 catalyst

Full size image

Pyrolysis process

Pyrolysis of biomass-rich MSW (1.0 g) was carried out in a two-stage fixed-bed reactor system (Zou et al. 2018). The reactor system consisted of a two-stage independently controllable furnace. In the first stage of the reactor, pyrolysis was carried out at constant temperature (T = 550 °C), and in the second stage at 500, 700 and 850 °C. When the reactor was heated and the set temperatures stabilized, Ni/ZSM-5 catalyst was fed to the second stage. The catalyst was placed in a quartz basket with silica wool, and its load was varied between 0.5 and 2.0 g. Nitrogen was used as carrier gas (4.5 L/h).

In some cases, the effects of the steam rate (1 and 5 g/h) were also investigated. In these cases, the water was introduced by a syringe pump through a stainless steel tube that passed through the thermal pyrolysis stage and reached the second stage entrance. After the reactions, the products were cooled down in two condensers by the water–ice mixture. The non-condensable gases were then dried and collected in a gas bag. The gas samples were collected for 60 min.

Analysis

To characterize the biomass-rich MSW proximate (TGA2000, Las Navas) and ultimate analyses were made. Ultimate analysis of the MSW sample was obtained with a CHNS/O analyser (Vario Micro cube, Elementar). In the analysis, the weight percentages of carbon, hydrogen and nitrogen were determined simultaneously, while the weight percent of oxygen was calculated by the difference.

The surface morphology of the catalyst was investigated by high-resolution scanning electron microscopy (SEM, Quanta 200, FEI).

Gas samples were analysed using a Panna A91-type gas chromatograph (GC) containing thermal conductivity and flame ionization detectors. Column A was Porapak Q and was used for CO2 analysis at 80 °C. Column B was a 5A zeolite molecular sieve (MS-5A, He as carrier gas) for analysis of H2, N2, O2, CO, and CH4 at 100 °C.

Gas yields (H2, CO2, CH4, CO, C2+) were calculated from the results of the gas chromatographic analysis. The char yield was determined by weight measurement, and the liquid yield was calculated by the difference. LHV was calculated using (Eq. 1) according to He et al. (2009).

$$\begin{aligned} & {\text{LHV}} \left( {\frac{{{\text{MJ}}}}{{{\text{Nm}}^{3} }}} \right) \\ & = \left( {{\text{CO}} \times 126.36 + {\text{H}}_{2} \times 107.98 + {\text{CH}}_{4} \times 358.18 + {\text{C}}_{2} {\text{H}}_{2} \times 56.002 + {\text{C}}_{2} {\text{H}}_{4} \times 59.036 + {\text{C}}_{2} {\text{H}}_{6} \times 63.772} \right)/1000 \\ \end{aligned}$$
(1)

Results and discussion

Effect of the temperature on gas yields and compositions

Gas yields changed between 17.3 and 29.8%, and yields of liquid and solid products were 15.9–24.6% and 54.2–58.0%, respectively. As expected, gas yields increased with the reaction temperature, and yields of liquid and solid products followed the opposite trend.

Temperature also significantly affected the composition (Fig. 2). In the function of the temperature, hydrogen (H2) and carbon monoxide (CO) contents increased, and concentrations of carbon dioxide (CO2), methane (CH4) and the heavier hydrocarbons (C2+) decreased. The higher gas yields and the increased H2 and CO contents were attributed to the lower thermal stability of the C–C bonds. The main reactions of two-step pyrolysis (Eq. 3–5) are endothermic, and their equilibrium can be shifted towards the formation of CO and H2 at higher temperatures. As mentioned earlier, the C2+ content also decreased with increasing reaction temperature due to the increased cracking activity of the catalyst.

Fig. 2
figure 2

Composition of gas products in the presence of 0.5 g catalyst (red curve shows the H2 yield in mmol/g waste)

Full size image

Reactions of two-step pyrolysis:

$${\text{C}}_{x} {\text{H}}_{y} {\text{O}}_{z} \to {\text{aCO}}_{2} + {\text{bH}}_{2} {\text{O}} + {\text{cCH}}_{4} + {\text{dCO}} + {\text{eH}}_{2} + {\text{f}}E_{2 + }$$
(2)
$${\text{Boudoard}}\,{\text{reaction}}: {\text{C}} + {\text{CO}}_{2} \to 2{\text{CO}} \left( { + \,164.2\, {\text{kJ}}/{\text{mol}}} \right)$$
(3)
$${\text{Water}}\, {\text{gas}}\, {\text{reaction}}: {\text{C}} + {\text{H}}_{2} {\text{O}} \to {\text{CO}} + {\text{H}}_{2} \left( { + \,131.3\, {\text{kJ}}/{\text{mol}}} \right)$$
(4)
$${\text{Steam}} \,{\text{reforming}}\,{\text{of}}\, {\text{methane}}: {\text{CH}}_{4} + {\text{H}}_{2} {\text{O}} \to {\text{CO}} + 3{\text{H}}_{2} \left( { + \,203.6\, {\text{kJ}}/{\text{mol}}} \right)$$
(5)
$${\text{Water}} \,{\text{gas}}\,{\text{shift}}: {\text{CO}} + {\text{H}}_{2} {\text{O}} \leftrightarrow {\text{CO}}_{2} + {\text{H}}_{2} \left( { - 41 {\text{kJ}}/{\text{mol}}} \right)$$
(6)
$${\text{Secondary}}\, {\text{cracking}} \,{\text{of}}\, {\text{tar}}: {\text{Tar}} + {\text{n}}_{1} {\text{H}}_{2} {\text{O}} \to {\text{n}}_{2} {\text{CO}}_{2} + {\text{n}}_{3} {\text{H}}_{2}$$
(7)

Effect of steam on gas yields and compositions

The effect of steam was investigated with 1 and 5 g/h rates at 500, 700 and 850 °C temperatures (Fig. 3). Due to the shorter residence times, the highest gas yield (27.1%, T = 850 °C, steam: 1 g/h) and hydrogen production (9.96 mmol g−1 waste) were slightly lower than in the base case (29.8% and 10.40 mmol g−1 waste). At a steam rate of 5 g/h, the residence time was about 42% of the base case; thus, significant decreases could be observed not only in gas yields but also in hydrogen contents. It is important to note that hydrogen yields of two-step pyrolysis of biomass/waste are around 10 mmol/g, which means that our results are consistent with those of other research groups (Alvarez et al. 2014).

Fig. 3
figure 3

Effect of steam rate on gas composition in the presence of 0.5 g catalyst (red curve shows the H2 yield in mmol/g waste)

Full size image

To evaluate the effect of steam rate, H2/CO molar ratios and LHVs were also calculated (Table 1). In general, H2/CO molar ratios changed between 0.7 and 1.8, and LHVs were in the range of 8.5 and 14.8 MJ/Nm3. H2/CO molar ratio determines the applicability of syngas in chemical synthesis. For example, the theoretical H2/CO molar ratio of the Fischer–Tropsch synthesis is about 2. This ratio was best approximated by the product of the experiment carried out at 850 °C and 1 g/h steam.

Table 1 H2/CO molar ratios and LHVs of selected products, MJ/Nm3
Full size table

Effect of the catalyst load on gas yields and compositions

Figure 4 shows the effect of catalyst load (0.5, 1.0 and 2.0 g) on gas composition and H2 yields at 850 °C. At higher catalyst loads, more catalytically active sites were available for the conversion of the molecules and thus gases (yields: 29.3, 51.2 and 51.4%), and H2 could also be formed in higher concentrations.

Fig. 4
figure 4

Effect of catalyst load on gas composition (T = 850 °C, Steam rate = 5 g/h, the red curve shows the H2 yield in mmol/g waste)

Full size image

RSM modelling of two-step pyrolysis of biomass-rich waste

As mentioned earlier, RSM was used to study the effects of operating parameters. For RSM, two factors were considered: temperature of the second stage and steam rate. Table 2 shows the chosen factors and outcomes. These two factors are the main effectors of the pyrolysis process and were defined as continuous variables. The lower and upper limits of the factors T and F are the boundaries of the system (500 and 850 °C, and 0 and 5 g/h steam). The coded values were calculated using 0 as the lower level and one as the higher level. As outcomes, the amount of char, the liquid and gas phases, the four important gases (H2, CO2, CH4, CO) and the C2+ fraction were investigated. A full factorial measurement was performed, extended with a middle point, resulting in 9 different experiments.

Table 2 The chosen factors and outcomes of RSM
Full size table

For each set of outcomes, the following quadratic polynomial expression was considered:

$$\begin{aligned} y = & a_{1} + a_{2} \times T + a_{3} \times F + a_{4} \\ & \times T^{2} + a_{5} \times T \times F + a_{6} \times F^{2} \\ \end{aligned}$$
(8)

Polynomial planes were fitted to the measurement data using the MATLAB curve fitting toolbox, and the data were evaluated by ANOVA methods. Figures 5 and 6 show the resulting response surfaces for each different outcome. CO2 response had a maximum of around 0.5–0.6 coded values. This finding means CO2 production is lower, and the reaction equilibrium shifts towards CO formation. At higher temperatures, the yield of CH4 also increased, but the reaction to hydrogen was not fast enough to counteract the decrease in the residence time. Figure 7 shows the differences between the measured and simulated values. The difference between the measured and simulated values is minimal in most cases. However, due to the inhomogeneity of the waste, larger differences can also be observed.

Fig. 5
figure 5

The response surface for gas product (coded values for T: 500 °C = 0; 700 °C = 0.57; 850 °C = 1.0; for F: 0 g/h steam rate = 0; 1.0 g/h steam rate = 0.2 and 5 g/h steam rate = 1.0)

Full size image
Fig. 6
figure 6

a The response surface for H2, b for CO2, c for CH4, d for CO, e C2+ (coded values for T: 500 °C = 0; 700 °C = 0.57; 850 °C = 1.0; for F: 0 g/h steam rate = 0; 1.0 g/h steam rate = 0.2 and 5 g/h steam rate = 1.0)

Full size image
Fig. 7
figure 7

Comparison of the measured and the simulated values

Full size image

Table 3 shows the identified model parameters and p values. The lower the p-value, the higher the contribution of the given factor to the outcome. As we can see, the temperature had a more important effect on the steam rate. However, in some cases (especially in the case of the hydrogen), the synergic effect of the two factors was also significant. The temperature had a negative effect on the char and liquid and a positive effect on the gas yield. The steam rate had the opposite effect on yields. The increasing steam rate promotes the water–gas shift reaction, leading to a negative effect on the CO and a positive effect on the CO2 yields. The synergic effects are shown mainly in the case of the C2+.

Table 3 The model coefficients and the p values
Full size table

Our primary goal was to maximize the H2 production, which can be achieved by applying the highest second stage temperature, and 0 g/h steam rate, but to determine the optimum, H2/CO molar ratios and the LHVs of the gas products were also taken into account (Table 1). Table 4 shows the minimum and maximum values of the variables. We also performed a desirability analysis, which is included in the table. Most of the responses have an antagonistic effect on each other, meaning the contrary changes in the yield of the products. Similarity can be observed in the case of gas components; all of the molecules affect each other due to the equilibrium reactions. The composite desirability value (0.81) is acceptable for this system.

Table 4 Minimum and maximum values of the variables (g)
Full size table

Conclusions

In this study, two-step pyrolysis of biomass-rich municipal solid waste was studied on nickel supported ZSM-5 catalyst. In the first stage of the reactor, constant temperature (T = 550 °C) was applied, and in the second stage, the reactions took place at 500, 700 and 850 °C. Effects of operating parameters (temperature = 500, 700, 850 °C, steam rate = 1, 5 g/h) and catalyst load (0.5, 1.0, 2.0) were studied on yields and compositions, particularly on hydrogen contents of gas products. In order to determine the significance of the operating parameters, response surface methodology was used. Considering the experimental data, the values of hydrogen/carbon monoxide molar ratio, lower heating value and the simulation, it was found that the favourable operating parameters of two-step pyrolysis of biomass-rich municipal solid waste were 850 °C temperature and 1 g/h steam rate (gas yield: 27.1%, hydrogen yield: 9.96 mmol g−1 waste, hydrogen/carbon monoxide molar ratio: 1.8). From the data of response surface methodology, it was also concluded that temperature has a more critical effect on gas yields and composition than steam rate. However, in some cases, the synergistic effect of the two factors can also be significant. Therefore, the determination of the significance and extrapolation of the RSM for other wastes was particularly important because the results can contribute to the development of the research area and allow the determination of the optimal parameters of hydrogen production with a reduced number of experiments.