Integrated Waste-to-Energy Process Optimization for Municipal Solid Waste

Abstract

Within the past few decades, thousands of experiments have been performed to characterize urban waste and biomass to estimate their bioenergy potential and product identification. There is a need to develop an integrated process model based on the experimental literature, as well as simulations to obtain suitable products. In this study, municipal solid waste (MSW), including paper and plastic characterization and an integrated process model, were developed to optimize the final products in a reactor system. The process model has two modes, R&D and reactor control (RC), to obtain suitable products including bio-oil, char, and gases. A database was integrated based on thermokinetics, machine learning, and simulation models to optimize product efficiency. The experimental data include those obtained by thermogravimetric analysis and Fourier-transform infrared spectroscopy, which were linked to a pyrolysis experimental setup. Feedstock product mapping models were incorporated into the database along with the temperature, heating rates, elemental analysis, and final product concentration, which were utilized for the pyrolysis reactor setup. Product feasibility was conducted based on life cycle cost, affordability, and product efficiency. The present work will bridge the gap between experimental studies and decision-making based on obtained products under several experimental conditions around the world.

1. Introduction

In recent decades, the focus of energy generation has diverted from fossil fuels to renewable energy resources. Sustainable waste conversion into renewable energy gained significant momentum due to the initiative from the European Green Deal (EGD) [1,2]. The European Green Deal is a key policy initiative toward the sustainable goal aims of transforming the EU into a resource-efficient and competitive economy, achieving net-zero greenhouse gas emissions by 2050, which is also a priority of the Canadian government. The ambitious strategy shared by both continents plays a pivotal role in our current research, which concentrates on key factors such as providing affordable energy, creating a fully integrated system for minimizing waste by converting it into valuable products, and enhancing energy efficiency by exploring new possibilities in renewable energy sources, as indicated in industry references [1,2]. Various initiatives under the European Green Deal (EGD), like the transition to clean energy through renewable energy resources and reducing emissions, are underway. These efforts align with the Canadian government’s goals to cut down emissions and optimize integrated solutions. The overarching objective is to establish a comprehensive system for effectively managing waste streams to produce beneficial products and chemicals.

Municipal solid waste (MSW) management is a crucial component in the advancement of the bioeconomy, as it is considered the primary source for bioenergy generation in the United States, Canada, and Brazil [3]. However, the utilization of biomass presents a challenge due to the diverse range of feedstocks and complex chemical structures [4]. Global plastic consumption was estimated to be 299 million tons in 2013, with an expected increase by several folds by 2050, which is alarming to the earth due to its difficulty in degrading in the environment and rapidly trailing landfills [5,6]. Process optimization plays a critical role in converting raw materials like paper waste and plastic waste (PPW) into valuable products and chemicals through experiments and computational work [7]. Currently, there are only a limited number of innovative process integration models that have emerged in the United States, Brazil, and certain regions of Europe, and further investigation into their development is necessary [8,9]. Nonetheless, a significant amount of research is being conducted on low-carbon and carbon-neutral technologies [10]. A wide range of waste-to-energy conversion technologies are available, including gasification, anaerobic digestion, coal-fired boilers, power plants, and biomass-to-ethanol processes (as well as other methods for producing transportation fuels) [11]. Pressurized anaerobic digestion through autogenerative high pressure [12], pressurized anaerobic digestion [13], and two-stage digestion [14] has improved the conversion of methane content into biogas. In another study, a reliability model using diesel engines while operating on biofuel was proposed [15]. Therefore, it is reasonable to inquire about the optimal combination of technologies, biomass types, and their locations that would result in the most efficient (i.e., cost-effective) approach to the expansion of the MSW industry both domestically and internationally. This work aims to formulate an optimization strategy to address these questions.

The process of thermochemical conversion of plastic and waste involves the utilization of a distributed network of fast pyrolyzers to thermally transform biomass into synthesis gas (syngas), bio-oil, and biochar [16]. This approach presents a sustainable means of generating energy from plastics and paper waste, as the syngas produced can provide the energy needed for the pyrolysis process. The resulting bio-oil is a valuable energy source, with a calorific value of approximately 17 MJ/kg, which can be used for heat generation or sent to a refinery for further processing into transportation fuels [17]. Biochar produced from the co-pyrolysis of plastic and wastes can improve soil quality in many ways. However, it is challenging to define the exact thermochemical conditions required, due to the complex nature and structure of biomass, to obtain different products (biofuel, gases, biochar) [18]. The co-pyrolysis process of PPW including corn straw, corn stover, and barley straw and plastic waste needs to be optimized, which will help in improving product recovery and energy efficiency alongside machine learning processes.

In the past few decades, numerous studies have focused on optimizing thermochemical processes (pyrolysis, gasification) to produce valuable products (biofuel, biogas, and biochar). A few have focused on utilizing biosolids from large industrial and municipal solid waste plants to detoxify the residuals and produce biochar and/or bio-oil simultaneously. In addition, minimal research has been conducted on the co-pyrolysis of biomass and plastic. A recent study showed that the biochar yield improved by 44–49% by using the co-pyrolysis of biomass and plastic instead of single pyrolysis [19]. However, the synergetic effects of char yield lower the pyrolysis temperature when mixing biomass with different ratios (25–50%) of plastic derivatives [20]. Another study focused on achieving different favorable optimizations of development temperature ranges for the co-pyrolysis of plastic and tires. But more critical aspects of co-pyrolysis were not explored including the underlying mechanism, thermokinetics, and synergetic effects of developing improved mechanistic models to obtain co-pyrolysis products [21]. Likewise, a co-pyrolysis study on truck tires and fruit bunches increased oil yield by 20% as compared to single tire pyrolysis [22]. Furthermore, a recent study quantified pyrolytic and gaseous products through machine learning approaches [23]. Therefore, using machine learning in the co-pyrolysis process of paper and plastic MSW is a novel idea in integrated biorefinery. Therefore, an optimization framework for the strategic identification, development and commercialization of bio-based feed stocks for biomaterials, biochemicals, and bioenergy will first be developed. The model will seek to provide an understanding of emerging opportunities and their underlying challenges for solid waste management companies, and the applicability of the model will be demonstrated with a case study focusing on new markets for MSW’s operations. An integrated analytical framework will also be applied to comparing the results with the recent literature [24]. However, several barriers [25] to the integration of the process model must first be identified including by using recommendations provided to develop integrated process models in other parts of Canada by policy makers wishing to facilitate the development of novel biomass or other paper waste in other parts of Canada.

The current study presents an exploratory case study analysis of paper waste and plastic MSW through process integration to develop biobased production systems. MSW was selected due to its abundance and easy availability, whereas plastic derivatives are hard to degrade and combustion is an environmental harm. Therefore, it would be important to know the co-pyrolysis behavior of MSW, wherever paper waste is a profuse biological resource while plastic is a detrimental environmental waste. Considering Canada’s abundant natural resources, the utilization of lignocellulosic residues obtained from agricultural sources and plastic derivatives as feedstocks presents a favorable opportunity for the development of advanced biofuel and biochar utilization industries [26]. Therefore, it would be worthwhile to produce valuable chemicals and products from an MSW-handling industry. In the upcoming years, with the growing population of Ontario, it will be crucial to address the proper conversion of these waste materials into useful chemicals and products [27]. The economic feasibility of biobased materials relies on the optimization of their efficiency and the reduction in toxic gas emissions during pyrolysis, gasification, and combustion processes.

A complete range of experimental and simulation analyses were performed including process system engineering, reaction engineering, pyrolysis, and co-pyrolysis experiments to obtain useful energy products (bio-oil, gases, biochar) in terms of energy generation and product optimization. Therefore, each pyrolysis product (bio-oil, gases, biochar) recovery will also be optimized through some complex mechanistic models through machine learning and detailed experimental characterization [28]. As a result, it is possible to provide the cheapest solution for the industrialist to adopt our expanded new strategy to contrivance in large-scale reactor modeling. It will also help to determine the overall cost of the process, including the rate at which the MSW is bought, and the financial and environmental benefits. We will also address the loopholes to finalizing this co-pyrolysis technology.

The integrated process includes a series of experiments linked with simulations and mathematical models to obtain optimized products and chemicals. The process has four major stages linked with each other throughout the process integration. It started with the characterization of a raw MSW sample and performance of thermogravimetric analyses to obtain different patterns and temperature profiles with heating rates and specific temperature ranges. The products were identified based on their temperature regions and kinetic parameters were identified to be used in the second stage, the simulation process. In the second stage, simulation through Aspen Plus was performed by inputting temperature profiles and kinetic parameters to obtain products. In the third stage, elemental analysis data were used that were generated through experiments and the literature to fit with the final products. Moreover, a life cycle cost analysis was performed to obtain the cost and benefit analyses of the major products. The novelty of this work resides in the amalgamation of experiments performed compared to recent decades of characterization of waste. No such effort was made for each process that was linked with one another through control systems to obtain useful products by changing input parameters such as temperature ranges, heating rates, and process conditions. It was only possible when an integrated system was introduced by incorporating experimental data and using this data to build experimental strategies to obtain optimized products. This strategy may reduce the cost, chemical usage, and time required to target the desired products.

2. Materials and Methods

A series of experiments were conducted in triplicate including thermogravimetric analysis, machine learning, pyrolysis, and simulation modeling as shown by the detailed framework in Figure 1. The various ranges of parameters were investigated during each stage including thermal degradation patterns, kinetic and thermodynamic parameters, elemental contents, reaction conditions, gaseous products, and life cycle costs.

2.1. Integration of the Experimental Setup (Reactor) with a TG-FTIR-GCMS System

The experimental setup was connected, including a TGA-FTIR and GCMS, to characterize the small amount of the MSW to obtain products and to study their reaction chemistry. For this purpose, a 10 mg sample was characterized in a controlled nitrogen environment at different heating rates (10–40 °C/min) to study its reaction chemistry. A kinetic analysis was also performed using different isoconversion models to obtain thermogravimetric parameters including the activation energy, pre-exponential factor, enthalpy, entropy, and Gibbs’ free energy. These parameters were stored in a database for future reference and to obtain suitable products based on experimental conditions. The obtained analysis results were stored in a database integrated as a part of process model to analyze the feasibility of the products through a laboratory-scale pyrolysis process. The integrated algorithm to obtain the thermodynamic and reaction chemistry is shown in Figure 2. Interventionary studies involving animals or humans, and other studies that require ethical approval, must list the authority that provided approval and the corresponding ethical approval code.

The evaluation of the reaction mechanism underlying biomass degradation is difficult due to the complexity of the chemical makeup of biomass. Kinetic analysis is a useful tool for assessing the thermal stability of materials during combustion and pyrolysis. A single-step biomass degradation process is shown in Equation (1) as the following:

$$MSW\stackrel{heat}{\to }volatiles+liquids+residual solids$$

(1)

The instantaneous conversion function $\alpha$ and the pyrolytic temperature have an impact on the rate of the chemical reaction. Consequently, it can be expressed as follows in Equation (2):

$$\alpha =\frac{m_o-m_t}{m_o-m_{\infty }}$$

(2)

where $m_0$ denotes the initial mass, $m_t$ denotes its mass at a certain point in the reaction time (t), and $m_{\infty }$ denotes its mass at the end of the reaction.

The response rate, $d\alpha /dt$, which is a function of α in these processes, can be expressed as follows in Equation (3):

$$\frac{d\alpha }{dt}=kf\left(\alpha \right)$$

(3)

where Equation (3) is the fundamental kinetic equation for solid-state mass loss processes. In the case of thermally stimulated processes, the Arrhenius equation is frequently used to replace the value of the rate constant k, like in Equation (3), which then has the following form:

$$\frac{d\alpha }{dt}=A\exp \left(\raisebox{1ex}{$-E_{\alpha }$}\!\left/ \!\raisebox{-1ex}{$RT$}\right.\right) f\left(\alpha \right)$$

(4)

where $A$ is the pre-exponential factor, $E_{\alpha }$ is the energy of activation, $f\left(\alpha \right)$ is the function of the degree of conversion, called the reaction model, and R is the gas constant. In terms of physics, $A$ describes the frequency of collisions between the particles involved in the production of activated complex(es), $E_{\alpha }$ is the activation energy barrier(s), and $f\left(\alpha \right)$ is an expression for the reaction mechanism.

Re-arranging Equation (3) gives

$$\frac{d\alpha }{f\left(\alpha \right)}=A\exp \left(\raisebox{1ex}{$-E_{\alpha }$}\!\left/ \!\raisebox{-1ex}{$RT$}\right.\right) dt$$

(5)

Substituting the constant heating rate (with increasing temperature),

$\beta =\frac{dT}{dt}=\frac{dT}{d\alpha }\times \frac{d\alpha }{dt}$, in Equation (5) gives

$$g\left(\alpha \right)={\int }_0^{\alpha }\frac{d\alpha }{f\left(\alpha \right)}=\frac{A}{\beta }{\int }_0^T\mathrm{e}\mathrm{x}\mathrm{p}\left(\raisebox{1ex}{$-E_{\alpha }$}\!\left/ \!\raisebox{-1ex}{$RT$}\right.\right)dT$$

(6)

The term $g\left(\alpha \right)$ is the integral expression of the conversion-dependent function, while the term used in Equation (6) is the integral of the temperature, which lacks an explicit integral solution. After resolving numerous empirical interpolation equations, several researchers have put forth an approximation of the solution.

Calculations of the KAS, OFW, and Starink methods can be established as the following:

$$\ln {\beta }_i=\ln \left(\frac{A_{\alpha }{E}_{\alpha }}{Rf\left(\alpha \right)}\right)-5.331-1.052\frac{E_{\alpha }}{{RT}_{\alpha ,i}}=\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}-1.052\frac{E_{\alpha }}{{RT}_{\alpha ,i}} (\mathrm{OFW})$$

(7)

$$\ln \left(\frac{\beta }{T^2}\right)=\ln \left(\frac{AR}{E_{a}g\left(\alpha \right)}\right)-\frac{E_{\alpha }}{RT} (\mathrm{KAS})$$

(8)

$$\ln \left(\frac{\beta }{T^{1.92}}\right)=\ln \left(\frac{AR}{E_{a}g\left(\alpha \right)}\right)-\frac{E_{\alpha }}{RT} (\mathrm{Starink})$$

(9)

The thermodynamic parameters in this study, including $∆G$, $∆H$, and $∆S$, were determined using the following formulas:

$$∆H=E-RT$$

(10)

$$∆G=E+R{T}_p\ln \left(\frac{k_{\mathrm{B}}{\mathrm{T}}_{\mathrm{p}}}{\mathrm{h}\mathrm{A}}\right)$$

(11)

where k_b is the Boltzmann constant, which has a value of 1.38 × 10⁻²³ J K⁻¹.

$$∆S=\frac{∆H-∆G}{T_p}$$

(12)

2.2. Simulation Integration in the Process Model with Machine Learning

The simulation software results were integrated to obtain suitable products and analysis based on the simulation results related to the temperature and heating rate conditions. The simulation was used to obtain the gaseous products based on the elemental analysis results which also integrated with the process model. Using the ASPEN Plus process simulator, a stoichiometric steady-state model was created in order to mimic the process of municipal solid waste. The elemental analysis results of five samples, MSW1, MSW2, MSW3, MSW4, and MSW5, were used as inputs to be simulated based on our previous studied data [29]. The model was used to study the influence of proximate and ultimate analysis on gasification behavior. Additionally, it was utilized to break down volatile yields that were produced during the process.

On the other hand, machine learning algorithms were used to map the elemental results to obtain suitable pyrolysis products. The major equations used are described below and the algorithm is shown in Figure 3.

Equations for Machine Learning

The following equations were used to calculate the residual sum of squares, meaning the error between the predicted and modeled values; the hyperbolic tangent function, representing the activation function in our model; the logistic sigmoid function, which is a non-linear function that allows neural networks to learn and represents complex patterns; and the Gaussian function, representing the initial weights of random values and used to perform regression and classification tasks.

$$SSE=1/2N{\sum }_{i=1}^n\left(y_i-t_i\right)$$

(13)

$$Hyperbolic Tangent=\frac{e^a+e^{-a}}{e^a+e^{-a}}$$

(14)

$$Logistic Sigmoid=\frac{1}{1+e^{-a}}$$

(15)

$$Sin=Sin (\propto )$$

(16)

$$Gaussian={\frac{1}{2\pi \sigma } e}^{\left\{{(x-u)}^2|{2\sigma }^2\right\}}$$

(17)

2.3. Process Model Integrated with a WTE Database

To further power the process model designs, a database was integrated in the laboratory-scale reactor. It was integrated with datasets obtained from the experimental, simulation, and literature data related to the temperature profiles, mass degradation patterns, heating rates, and product analyses of similar waste samples. It gave information regarding mass degradation patterns at different heating rates and temperature ranges which could further be validated through our experimental setup values, as shown in Figure 4.

2.4. Life Cycle Analysis (LCA)

The primary purpose of this objective was to estimate the sustainability of the products using a cost analysis with an environmental impact assessment. The process was taken from the European ELCD database to conduct the LCA, based on the CML 2001 ecoinvent method, to study 1000 kg of the reactor; if we increased the capacity of the reaction, the emission would increase at the same ratio as the capacity of the reactor.

LCC = IC + ESPWF (Cost _energy + Cost _main)

LCC = Life Cycle Cost

IC: Initial Cost

ESPWF: Equal Series Present Worth Factors

3. Results and Discussion

3.1. Integration of Thermal Analysis

Thermal analysis was performed to obtain the degradation patterns, temperature profiles, and kinetic analyses to integrate in the process design to optimize conditions for various products. The thermal behavior of the MSW was evaluated at four different heating rates (10, 20, 30, and 40 °C/min). Thermogravimetry (TG-DTG) revealed that mass was continuously lost as the temperature increased. Different stages of mass loss were observed from the TG-DTG curves of the model compounds. During the degradation reaction, dehydration occurred at temperatures below 140 °C, followed by the decomposition of the organic components (Figure 5A). As the material was composed of plastic and paper, therefore, hemicellulose, cellulose, and lignin decomposed at temperatures between 180 ℃ and 330 °C, 330 °C and 490 °C, and 180 °C and 800 °C, respectively [30].

The rate of heating ($\beta$) has a substantial impact on the rate of breakdown, total weight loss, and temperature at which the highest weight loss occurs (DTG_max). When the heating rate was increased, a higher rate of biomass decomposition was observed, causing the DTG_max peaks to move toward the higher temperature region (Figure 5B), which is attributed to the poor conductivity of the biomass. Therefore, a higher heating rate results in a shorter residence time for heat transmission and a higher temperature is needed to break down organic matter [31]. Interestingly, here, the cellulose breakdown predominated (300–361 °C) with DTG_max peaks centered at 273 °C, 276 °C, 279 °C, and 281 °C, at 10 °C/min, 20 ℃/min, 30 ℃/min, and 40 ℃/min, respectively. The weight loss was found to be 18%, 17%, 18%, and 18% at heating rates of 10 °C/min, 20 °C/min, 30 °C/min, and 40 °C/min, respectively. The phenomena of the degradation pattern and temperature associated with each stage are shown in Figure 5A,B. It is obvious that a reaction rate depends primarily on the temperature of the reacting particles, with higher temperatures causing faster reactions with a lower reaction time, thus subsequently increasing the surface temperature. However, MSW almost approaches thermal equilibrium due to lower heating rates and a prolonged thermal energy supply [32]. Thus, the maximum decomposition is shifted to an area of high temperature as a result of a high heating rate. Interestingly, increased gas production is made possible by rapid degradation of volatile elements. Therefore, a lower residue and higher devolatilization rate were obtained [33]. The thermal degradation pattens were also compared with already studied waste samples including plastic [34], plastic waste [35], municipal solid waste [36], solid waste [37], and high-density waste plastic [38], indicating good synchronization which means it can also be used to generate energy products. Furthermore, the pyrolytic decomposition of biomass components occurs individually at a lower heating rate, while the immediate reaction causes the disintegration of various components concurrently at a higher heating rate.

Furthermore, thermal data obtained from TGA experiment were directly stored in the integrated process model for the laboratory-scale reactor for product optimization and control based on the temperature ranges, mass degradation patterns, and heating rates.

3.1.1. Kinetic Analysis

The activation energy ($E_a$) of the MSW at 10, 20, 30, and 40 °C/min was investigated using the Friedman, FWO, and KAS methods. The process was conducted using slow pyrolysis with gradual, non-isothermal heating. This approach facilitates the determination of kinetic parameters through thermogravimetric analysis, enabling the easy acquisition of temperature-specific data over a prolonged duration. Calculations of the kinetic parameters were made for conversion values ranging from 0.2 to 0.8 and activation energies were calculated using the slopes. For the conversion of 0.1, the regression coefficient is very low; thus, it is considered negligible and therefore ignored [39]. For all models, smooth fits were achieved and for various conversion fractions, the estimated $E_a$ values and R² are shown in

Table 1. The thermodynamic parameters with a continuous increase in the conversion (α) with all kinetic models.

Alpha	Activation Energy (E) kJ/mol	Pre-Exponential Factor (A)	R2	Enthalpy (∆H) kJ/mol	Gibbs’ Free Energy (∆G) kJ/mol	Entropy (∆S) J/mol
FWO Model
0.10	132.37	3.52 × 1097	0.68	126.02	−1099.63	1606.35
0.15	167.92	2.54 × 10124	0.91	161.57	−1456.38	2120.52
0.20	174.11	1.20 × 10129	0.96	167.77	−1518.50	2210.04
0.25	177.56	4.84 × 10131	0.98	171.21	−1553.09	2259.90
0.30	178.55	2.73 × 10132	0.99	172.21	−1563.06	2274.27
0.35	178.17	1.40 × 10132	0.99	171.82	−1559.20	2268.71
0.40	177.97	1.00 × 10132	1.00	171.63	−1557.27	2265.93
0.45	177.19	2.55 × 10131	1.00	170.84	−1549.39	2254.56
0.50	176.44	6.93 × 10130	0.99	170.10	−1541.88	2243.74
0.55	176.26	5.05 × 10130	0.99	169.91	−1540.05	2241.10
0.60	180.04	3.63 × 10133	0.98	173.70	−1578.00	2295.80
0.65	182.40	2.20 × 10135	0.81	176.05	−1601.66	2329.90
0.70	137.77	4.24 × 10101	0.61	131.42	−1153.84	1684.49
0.75	130.94	2.95 × 1096	0.55	124.60	−1085.32	1585.73
0.80	243.30	2.25 × 10181	0.98	236.96	−2212.82	3210.72
0.85	243.87	6.02 × 10181	0.99	237.52	−2218.49	3218.89
0.90	240.91	3.48 × 10179	1.00	234.56	−2188.76	3176.04
	185.03			178.69	−1628.09	2367.99
KAS Model
0.10	123.22	4.33 × 1090	0.65	116.88	−1007.85	1474.086
0.15	158.35	1.49 × 10117	0.90	152.00	−1360.35	1982.111
0.20	164.26	4.39 × 10121	0.95	157.92	−1419.70	2067.653
0.25	167.49	1.21 × 10124	0.98	161.15	−1452.10	2114.354
0.30	168.30	4.95 × 10124	0.99	161.96	−1460.24	2126.075
0.35	167.76	1.93 × 10124	0.99	161.42	−1454.82	2118.263
0.40	167.44	1.11 × 10124	0.99	161.10	−1451.60	2113.624
0.45	166.55	2.32 × 10123	1.00	160.20	−1442.59	2100.643
0.50	165.69	5.26 × 10122	0.99	159.35	−1434.02	2088.298
0.55	165.40	3.15 × 10122	0.99	159.05	−1431.07	2084.034
0.60	169.01	1.70 × 10125	0.98	162.67	−1467.36	2136.338
0.65	170.84	4.09 × 10126	0.79	164.50	−1485.71	2162.784
0.70	124.87	7.69 × 1091	0.56	118.53	−1024.44	1498
0.75	115.57	7.19 × 1084	0.48	109.22	−931.07	1363.429
0.80	227.05	1.18 × 10169	0.98	220.70	−2049.68	2975.598
0.85	227.27	1.73 × 10169	0.99	220.92	−2051.91	2978.817
0.90	224.06	6.58 × 10166	1.00	217.72	−2019.76	2932.471
	172.77			166.43	−1505.07	2190.69
Starink Model
0.10	123.59	8.19 × 1090	0.65	117.24	−1011.52	1479.38
0.15	158.73	2.91 × 10117	0.91	152.39	−1364.19	1987.65
0.20	164.66	8.72 × 10121	0.96	158.31	−1423.65	2073.35
0.25	167.90	2.43 × 10124	0.98	161.55	−1456.14	2120.18
0.30	168.71	1.01 × 10125	0.99	162.37	−1464.35	2132.00
0.35	168.18	3.99 × 10124	0.99	161.84	−1458.99	2124.28
0.40	167.86	2.30 × 10124	0.99	161.52	−1455.82	2119.72
0.45	166.97	4.87 × 10123	1.00	160.63	−1446.86	2106.80
0.50	166.12	1.11 × 10123	0.99	159.78	−1438.34	2094.52
0.55	165.83	6.71 × 10122	0.99	159.49	−1435.42	2090.32
0.60	169.45	3.66 × 10125	0.98	163.11	−1471.78	2142.72
0.65	171.30	9.15 × 10126	0.79	164.96	−1490.34	2169.47
0.70	125.39	1.89 × 1092	0.56	119.05	−1029.62	1505.46
0.75	116.18	2.09 × 1085	0.49	109.84	−937.24	1372.32
0.80	227.70	3.65 × 10169	0.98	221.35	−2056.21	2985.00
0.85	227.93	5.50 × 10169	0.99	221.59	−2058.58	2988.42
0.90	224.74	2.12 × 10167	1.00	218.39	−2026.52	2942.21
	173.26			166.92	−1509.99	2197.78

During pyrolysis, it was observed that the $E_a$ values ranged from 132–243 kJ mol⁻¹ when using the OFW technique, 123–227 kJ mol⁻¹ when using the KAS model, and 123–224 kJ mol⁻¹ when using the Starink method. Variation in $E_a$ with conversion values (Figure 6A) indicates that the pyrolysis of the MSW involved a parallel, multi-step devolatilization reaction pathway. The mean $E_a$ values were found to be 185 kJ mol⁻¹, 172 kJ mol⁻¹, and 173 kJ mol⁻¹ by the OFW, KAS, and Starink models, respectively. A comparison of each activation energy at a continuous increase with the conversion point (α) is shown in Figure 6A. The linear plots of the Friedman, OFW, and KAS analyses with small correlation factors suggest the accuracy of the $E_a$ recorded from the experimental data. The kinetic energy values were compared with already studied waste samples including plastic [34], plastic waste [35], municipal solid waste [36], solid waste [37], and high-density waste plastic [38] and indicated correlation, suggesting MSW can also be used effectively to generate energy products. The results obtained from the kinetic model were integrated in a database to optimize the efficiency of the reaction chemistry that was directly linked with the production of biochar, biofuel, and gaseous products.

3.1.2. Thermodynamic Analysis

In an inert environment, thermodynamic parameters were determined as shown in Table 1. For designing simulations, reactor thermodynamics parameters play a pivotal role [40]. The average changes in enthalpy ($∆$H) were recorded as 178 kJ mol⁻¹, 166 kJ mol⁻¹, and 166 kJ mol⁻¹. A comparison of the $∆$H with the conversion point (α) is shown in Figure 6B. A positive $∆$H and negative values indicated that MSW undergoes endothermic reactions and exothermic reactions during the pyrolysis process [41]. The reaction mechanism of the MSW comprises a function with multiple reactions because of the complex nature of the material, as shown in Figure 7. However, keeping in view the first and second laws of thermodynamics, the change in the Gibbs’ free energy, $∆$G, indicates the overall increase in energy of a system and the tendency of a reaction to occur in a particular direction. The results showed increasing trends in the $∆$G for the conversion values, 0.2 to 0.8, which confirmed the mechanism’s suitability for co-pyrolysis and co-gasification. Significantly, it has been noted that the reaction consistently produced a relatively stable energy output throughout its duration. The results showed decreasing trends in the $∆$S for the conversion values, 0.2–0.75, which tend to move a system toward thermodynamic equilibrium by employing physical and chemical processes [42]. This may be attributed to the complex nature of biomass, so properties are difficult to predict.

3.2. Simulation and Comparison of Experimental vs. Model Results

Several product species were discovered through the thermochemical process; the previous results showed that the feasibility of the process is comparable with already studied municipal waste samples. CO concentration is higher as compared to other gaseous products, but if we compare CO₂ with already studied waste samples, it has also been feasible in the gasification experimental data in recent years. CO was higher when compared to PE, PP, and PC at 750 ℃ but the same when we compared it with waste polythene [43]. When it was compared with H₂, it was lower in PE and PP, as studied by Cao et al. [44], but it was almost comparable to a recent study by Dogu et al. [45]. However, CO₂ percentage was high as compared to polythene bags [46], but lower as compared to Cao et al.’s experimental work related to gasification [44]. Our gasification method used was almost the same when compared to the experimental gasification of general plastic waste [47]. On the basis of these results, it can be predicted the reactor model showed prominent results when compared with experimental data in recent years as shown in

Table 2. Comparison of experimental and simulated results.

Feedstock	Gasification	H2	CO2	CH4	CO	C2H4	C2H6	Reference
Polyethylene (PE)	750 °C	41.67	20.77	30.22	1.14	1.35	4.86	[44]
Polypropylene (PP)	750 °C	36.99	27.82	26.57	1.68	0.86	6.08	[44]
Polycarbonate (PC)	750 °C	36.35	36.80	23.19	2.34	0.27	1.06	[44]
Plastic waste	650–1100 °C	64	4–6	3.30	25.7			[47]
Plastic waste (polyethylene box used for fruit packing)	850–1100 °C	40.86	0.37	9.10	0.1	C3H8 (12.8)	28	[46]
Thermoset-insulated wire and cable waste	750 °C (10–14 L in 20 g waste)	4–7		10–13	7–13			[48]
Waste polyethylene	700 °C to 900 °C	16–36	35–20	21–9	20–27	3–4	2	[43]
PE	740 °C (Pyro)	0.8		4.2		11.4	7.3	[45]
PP	760 °C (Pyro)	0.7		4.8		6.6	6.4	[45]
MSW1	750 °C	0.46	3.3	0.03	21.67			Simulation Work
MSW2	750 °C	0.72	3.71	0.06	19.67			Simulation Work
MSW3	750 °C	0.57	4.86	0.03	32.78			Simulation Work
MSW4	750 °C	0.39	3.22	0.05	20.54			Simulation Work
MSW5	750 °C	0.42	4.51	0.02	24.89			Simulation Work

.

3.3. Integration of Characteristics of Temperature Profiles with Degradation Stages

The mass loss curves provide insights into the physical and chemical transformations that occurred when the MSW was thermally converted into various products [49,50]. Thermogravimetric analysis is commonly employed to quantify the percentage of mass loss in a sample as a function of pyrolysis temperature. In this study, the mass loss percentage of MSW was examined at different heating rates ranging from 5 to 50 °C min⁻¹. Interestingly, the results indicated that the heating rate did not significantly impact the mass loss percentage. The curves displayed a typical pattern of degradation observed in lignocellulosic MSW, which is comparable to the DTG curves obtained from the pyrolysis of various waste materials, including Delonix regia, food waste using leachate, potato stalk, and peanut shell [33,51,52,53].

Based on a literature search, the samples with plastic and paper showed the same trends with the TGA analysis. The mass loss can be classified into three stages, followed by a long tail: the initial stage began at room temperature and extended to a certain temperature, to 110–120 °C for moisture removal, and the second stage occurred with a major degradation pattern to degrade cellulose, hemicellulose, and some lignin components which ranged from 120 °C to 450–520 °C for all observed plastic and paper samples [29,54,55,56,57,58,59,60,61,62]. The third stage observed when temperature ranged from 520 ℃ to 800 ℃ that corresponded to degradation of complex products.

Table 3. Relationships between the conversion (α), temperature, and activation energy.

Conversion Range (α)	Temperature Range (K)	Reactions	Activation Energy (Ea)
α ≤ 0.17	25–382	Liberation of retained moisture and volatiles	Increased from starting point to 111–145 kJ mol−1
0.17 ≤ α ≤ 0.5	382–474	Degradation of secondary and tertiary structured compounds	Increased from 145 to 176 kJ mol−1 and then decreased from 176 to 131 kJ mol−1
0.5 ≤ α ≤ 0.8	473–516	Degradation of lignin	Activation Energy from 180 to 220 kJ mol−1
0.8 ≤ α ≤ 1.0	516–800	Lignin degradation and char formation	Decreased from 240 to 120 kJ mol−1

and

Table 4. Relationships between the conversion (α), temperature, and activation energy.

Conversion Range (α)	Temperature Range (K)	Reactions	Activation Energy (Ea)
α ≤ 0.17	25–365	Liberation of retained moisture	Increased from starting point to 105–143 kJ mol−1
0.17 ≤ α ≤ 0.5	365–486	Degradation of secondary and tertiary structured compounds	Increased from 135 to 167 kJ mol−1 and then decreased from slightly with 5 kJ mol−1
0.5 ≤ α ≤ 0.8	486–511	Degradation of lignin	Increased from 184 to 224 kJ mol−1
0.8 ≤ α ≤ 1.0	511–800	Lignin degradation and char formation	Decreased from 247 to 112 kJ mol−1

demonstrate the reaction chemistry and activation energy of plastic with temperature profiles that can be used for process optimization.

3.4. Machine Learning Integration

The dataset for machine learning (ML) models including artificial neural networks (ANNs) and advanced regression trees were taken from already built datasets related to plastic pyrolysis. Using ML models, this model accurately predicted biochar outputs from the co-pyrolysis of plastic and paper based on ultimate and proximate analysis results. It was discovered that SANN and C&RT are considered as best models to predict the co-pyrolysis yields. Researchers in the scientific literature who employed machine learning techniques did not compare their outcomes to those of conventional models. Model outcomes provide a way of identifying the influence of characteristics on the prediction of biochar and yields from the co-pyrolysis of biomass and plastics. The data could also assist in forecasting gas outputs from the co-pyrolysis of biomass and plastics.

Figure 8A–D depict the mean values of the features used to train the biochar data, in order to forecast yields using proximate and elemental analysis. For the machine learning analysis, the top six training features proved to have a considerable impact on yield prediction, while other features had a small effect. With mean absolute values of 3.88 and 6.47, respectively, the weight percentage of polymers in the mixture was shown to have a substantial effect on the accuracy of biochar and bio-oil model predictions. This is also evident in the effect of w% P on the prediction: a high value led to a low biochar yield while more plastic in the feedstock led to a higher liquid yield and a lower char yield. Fixed carbon in biomass influences char production; at all pyrolysis temperatures, the presence of more fixed carbon increases char formation. With a mean value of 2.47, the ML analysis indicated that fixed carbon was the second most influential factor in predicting biochar output. A similar conclusion was suggested by the effect of fixed carbon. Co-pyrolysis produced greater solid residue from biomass containing more fixed carbon.

Using machine learning, this work accurately predicted biochar yields from the co-pyrolysis of paper and plastics based on feedstock properties, reaction conditions, and elemental analysis results. It was discovered that dense neural network models excel at predicting biomass and plastics co-pyrolysis yields. The optimization of hyperparameters and the incorporation of proximate analysis into the features enhanced the forecasts for bio-oil substantially. In light of the effects of parameters driving the synthesis of biochar, SHAP analysis helped us comprehend the pyrolysis of the biomass and plastics.

There are some recent studies that have shown that our model is better as compared to already studied ML models. Praporn predicted a model using XBoost to find the synergistic effect of the co-pyrolysis of plastic with a <90% accuracy [63]. Similarly, predicted the biochar percentage using machine learning models with an accuracy of 85% with experimental data [64]. Additionally, a bio-oil predicted model was made with an accuracy of an RMSE of 92% by Tang et al. [65].

3.5. Life Cycle Analysis

This approach was used to calculate the environmental effects of pyrolyzing mixed plastic waste. However, 1 ton of mixed plastic waste was the functional unit often used in all the impacts discussed below. The net climate change impact is estimated to be 713 kg CO₂ eq./t. Before substitutions, the pyrolysis process accounts for two-thirds of the overall impact, followed by waste collection and sorting at 26% and pyrolysis oil purification at 8%. The 419 kg CO₂ eq./t reduction in total impact can be attributed to the avoidance of naphtha production, which was replaced by pyrolysis oil. This substitution led to a significant decrease in greenhouse gas emissions, thereby contributing to the reduction in the carbon footprint. It is essential to recognize the positive environmental impact of this substitution and give credit to the efforts that led to this achievement. Such efforts could include the development and implementation of sustainable practices that promote the use of alternative fuels or the adoption of cleaner production processes. By acknowledging these contributions, we can encourage further progress toward more sustainable and eco-friendly practices in various industries. Ultimately, this can help to mitigate the adverse effects of climate change and promote a more sustainable future for generations to come.

Energy recovery, though, is a preferable choice for the other aspects considered here. All of the options are depicted by the net savings they bring and the contributions of the material or energy substitutions they bring outweigh their systemic consequences. Chemical recycling and energy recovery have comparable process emissions, but the latter has contributions that are roughly twice as high as the contributions of a naphtha replacement in a pyrolysis system. Apart from a few toxicity impacts (carcinogenic and non-carcinogenic human toxicity as well as terrestrial and freshwater ecotoxicity), a similar pattern is observed for all other impacts measured by the Environmental Footprint techniques. For a 1000 kg reactor, the cost would be CAD 176,093 with a 10-year lifetime warranty, and the profit would be CAD 33,318 based on rough estimation as shown in

Table 5. The product’s value with LCC and revenue generation.

	H2	CO2	CH4	CO
Percentage	4	40.5	0.2	240
Value (CAD)	12	3.5	1.2	0.2
Cost	48	141.75	0.24	48
Revenue	CAD 237.99/Run and CAD 33,318/Year

. If the process is steady and efficient, it can increase up to CAD 55,000.

3.6. Process Controller

By utilizing the three distinctive modes of operation, namely experimental testing, simulation, and literature database analysis, we could effectively establish the critical connections required to pinpoint the ideal product outcomes under precise heating rate and temperature conditions for a given feedstock. This holistic approach to integrated process design not only streamlines the production process but also minimizes operational costs. Instead of allowing reactions to occur at varying temperatures and heating rates, leading to unnecessary time and financial resource wastage, we can strategically control these variables to yield the desired products, as exemplified below.

For every distinct feedstock, the selection of optimal product parameters can be achieved by defining specific operational conditions within a laboratory-scale pyrolysis reactor. This approach empowers us to finetune the pyrolysis process for each unique feedstock, enhancing the efficiency and effectiveness of product development and resource allocation.

Further investigations through a process controller were carried out based on different scenarios of temperature, heating rate, and flow control to evaluate the developed products of the pyrolysis reactions. It was determined that the best heating rate to achieve the highest efficiency occurred at 20 °C/min with a flow rate of 60 mL/min. It was intriguing to observe a significant presence of products associated with them during the experiment. In the reaction, a range of products was formed, including water (H₂O), carbon dioxide (CO₂), carbon monoxide (CO), nitric oxide (NO), and various hydrocarbons such as butane, ethane, ethylene, propylene, acetaldehyde, ammonia (NH3), and formaldehyde. The order of hydrocarbon evolution was distinct, beginning with acetaldehyde, followed by propylene, ethane, ethylene, butane, and ending with formaldehyde. Oxygenated gases like CO and CO₂ were also observed, with their formation linked to the breakdown of carbon–carbon (C–C) and carbon–oxygen (C–O) bonds. This occurs during the decomposition of hemicellulose and cellulose at temperatures ranging from 240 to 500 °C. The decomposition process extends to lignin and carboxyl (COOH) groups at higher temperatures, leading to the further generation of CO and CO₂. During the initial pyrolysis stage at lower temperatures, there is a predominant release of hydrogen (H2) and water vapor (H₂O). In contrast, at elevated temperatures during secondary pyrolysis or charring reactions, there is a release of additional H2, indicating the breakdown of heavier hydrocarbons.

4. Conclusions

The current study provides a well-coordinated process integration for optimizing waste-to-energy processes for municipal solid waste (MSW) that is an essential prerequisite for achieving sustainable technology. The data obtained from thermogravimetric analysis and evolved product analysis were determined by considering different heating rates ranging from 10 to 40 °C min⁻¹ to obtain kinetic and thermodynamic parameters. The results obtained from the characterization through thermogravimetric analysis predicted average values of activation energy of (173–185) kJ/mol and enthalpy of (166–178) kJ/mol with greater R² values that showed the suitability of MSW to generate energy. The product analysis by incorporating coupling between TGA and evolved gas analysis established the potential to obtain useful product species including a high ratio of acetaldehyde ˃ propylene ˃ ethane ˃ ethylene ˃ butane ˃ formaldehyde, which were identified in triplicate at different heating rates. Furthermore, the simulation modeling results showed that the gasification process is also comparable to pyro-gasification, gasification, and catalytic pyrolysis at producing methanol and hydrogen. Several product species have been discovered through the thermochemical process; previous results showed that the feasibility of the process is comparable with already studied municipal waste samples. The experimental and simulation results were compared and gasification occurred at 750 °C. Life cycle analysis and process optimization can bridge the gap for rapid product optimization and profits were estimated at CAD 33,318/Year for a small-scale plant. These values were based on the present value of the products in CAD and can vary with inflation and management practices. All results concluded the suitability of MSW for co-pyrolysis and the co-gasification process with various feedstocks to increase the process efficiency. Hence, we can unlock the potential of waste as a valuable resource for sustainable development. In the case of every unique feedstock, we have the capability to handpick the most suitable product parameters by establishing precise operational conditions within a laboratory-scale pyrolysis reactor. Different stakeholders may benefit from this integrated system including waste management companies, waste treatment plants, pharmaceutical companies using waste problems, and agricultural and farm waste management. This method provides a means to finely calibrate the pyrolysis process to cater to the specific characteristics of every feedstock, ultimately elevating the efficiency and efficacy of our product development efforts and resource utilization. The main contribution is linking products through the co-pyrolysis process by connecting an experimental and simulation setup to obtain optimized products by reducing cost, time, and resources. This might help in the design of a reactor system connected to an experimental and control system where the process can be controlled by changing the temperature, heating rate, and reaction conditions. From the experimental system including, thermogravimetric analysis, evolved product analysis data were generated which were used in kinetic models and reaction chemistry from which an AI system decided the best conditions to perform experiments to obtain different products.