Design of Pyrolysis Plant for Waste Methyl Ethyl Ketone from the Polarizer Manufacturing Process

Abstract

The rapid growth of the semiconductor industry has made significant strides in addressing clean energy concerns. However, there are still unresolved issues related to waste solvents. One promising approach to tackle these challenges is through pyrolysis. This study selected waste methyl ethyl ketone (MEK) from the industrial sector as the feedstock for pyrolysis, resulting in various residual products such as fixed carbon (char), carbon soot, and fuel gases. Experimental results demonstrated that operating temperatures between 750 and 900 °C under anaerobic conditions yielded 5% to 10 wt% of fixed carbon, along with a small amount of tar and 80% to 90% of fuel gases. The research included lab-scale pilot experiments and field-scale system studies to develop a comprehensive concept for a thermal cracking plant. SolidWorks and Aspen Plus software were applied for calculations involving heat-transfer coefficients, residence time, and the utilization of fuel gases with a boiler or burner. A field system was constructed to scale up the pyrolysis process and effectively eliminate waste solvents, incorporating an automated procedural process.

1. Introduction

Over the past few decades, the demand for high-quality cleaning solvents for the photoelectric material industry has increased due to the rapid development of industrial technology [1]. Unfortunately, with the growth of the domestic electronics and semiconductor industries, the output of waste solvents has increased annually, and there is no efficient way to deal with all kinds of waste solvents. These organic waste solvents are proven to be toxic and carcinogenic to humans and harmful to organisms and the environment. They are critical industrial waste projects managed by the Environmental Protection Agency (EPA) [2].

Spent solvents are a significant product of imaging and chemical Processes due to their polarity and reactivity. Storing the solvents in safety vessels is not a viable option due to their high calorific heat and flammability. Solvent treatment procedures with a higher boiling point and complex molecules are typically divided into physical, thermal, and recycling parts (11.42%, 63.77%, 24.16%); Those with lower boiling points and smaller molecules have different proportions (44.49%, 28.82%, 24.16%). While physical methods such as filtration or distillation were once more common, the formation of side products at the bottom of the tower and the higher cost of reused solvents have made these processes less practicable [3,4]. The thermal method has gained popularity in various industries. While there is research on directly burning solvents in cement kilns or burners [5], the blending of the burned stream may pose challenges in terms of operation and de-NOx protection mechanisms compared to the pyrolysis process. In addition to incineration, thermal cracking methods have been increasingly adopted in recent years to produce synthesis gas, which is similar to natural gas. The heat generated from this process can be directly used in boilers, burners, or power generators. Methyl ethyl ketone (MEK) is widely used in detergents. In the research discussed, MEK is obtained from the polarizer roll-to-roll process. Additionally, some formulations of ink involve the use of diluted MEK as a solvent in the subsequent cleaning process [6]. Several studies have focused on the chemical reactions of MEK [7,8,9,10], and it has been concluded that during the pyrolysis process, MEK generates a methyl radical initially, followed by the formation of low carbon chain molecules and C-O components in subsequent steps of the mechanism. In many cases, waste solvents are contaminated with polymers due to the complex nature of high-tech manufacturing processes involving materials such as plastics or glues. Considering that several studies have investigated the pyrolysis process of plastics such as PP, PE, and PS, blending plastic waste with solvents during pyrolysis has been proposed as a promising approach [11]. However, it should be noted that the cracking reactions of plastics can be complex. Despite this complexity, blending plastics with solvents during pyrolysis has been explored as a potential strategy in some studies.

Computational fluid dynamics (CFD) and Aspen Plus were popular software to simulate and adjust the design reactor; Aspen Plus is a numerical program that is based on thermodynamics and chemical kinetics, making it suitable for analyzing those aspects. However, it lacks capabilities for dealing with flow-related problems. Further, CFD is well-suited for handling complex geometries. SOLIDWORKS flow simulation is widely used for solving various heat transfer-related problems, including predicting heat transfer coefficients [11], designing heat exchangers for optimal heat transfer capacity [12], analyzing heat transfer in thermal generators [13], applying nuclear energy principles [13], and designing pyrolysis tanks [13]. In the mentioned research, the heat transfer coefficient and overall heat flux are the primary parameters considered in the reactor design.

Governments are actively promoting the transition towards a circular economy, where waste treatment policies focus on resource utilization. This has become a global concern in recent years, with many articles discussing the potential of waste tires, plastics, biomass, and carbon fibers in achieving circular economy goals. Researchers such as Chew [14], Martínez [15], Naqvi [16], Sakthipriya [17], and Zabaniotou [18] have published articles on blending pyrolysis as a way to improve reactivity and achieve high utilization through the use of residual char in traditional chemical plants as recycled raw materials or adsorption materials [14]. The circular economy approach involves redefining waste, redesigning products and processes, and utilizing new technologies to reintroduce resources, raw materials, and refuse back into the industry chain for circulation, leading to significant reductions in energy and resource consumption, as well as waste generation.

The objective of this study was to develop an on-site waste solvent system that eliminates the need for off-site incineration and instead recovers and utilizes the energy contained in the waste. By implementing such a system in manufacturing factories, it is possible to prevent the annual increase in treatment fees and generate fuel value. The study encompassed the following tasks: (1) Utilizing CFD and Aspen Plus (EDR), the gas phase convection/conduction heat transfer capacity was examined, along with determining the heat transfer capacity of the reactor and heat exchanger; (2) Designing an experimental pyrolysis system for preliminary testing and assessing the amplification effect. The specific conditions for the optimal continuous operation were determined; (3) Defining the production process of waste MEK pyrolysis and establishing the control system for the chemical process; (4) Constructing a field system to verify the feasibility of quantification. Through the completion of these tasks, the study aimed to develop a robust waste solvent system that can be implemented on-site, making the process of organic waste elimination more economical.

2. Materials and Methods

This research mainly focuses on obtaining optimized design parameters for the experimental machine to avoid the system amplification effect. While many studies on thermal cracking process efficacy rely on residence times as a convenient and intuitive parameter, results can be distorted in most reactors due to heat transfer issues caused by incomplete product formation, such as tar or solvents, resulting from differences in substrate conductivity. The main objective of this research is to establish a solvent treatment plant with a capacity of 15 tons per month. This capacity is approximately half of the monthly amount of waste MEK produced in the photoelectric material industry, and a lab-scale system was built to conduct long-term tests instead of micro-tests. The optimal conditions determined from the lab-scale system, including heat transfer coefficients and theoretical residence times at different temperatures, will be adopted to scale up the process using commercial software such as EDR or CFD to size the machines accordingly. The thermal cracking reactor and cyclone-heat exchanger are the central units in this research, with pumping, filtration, and combustion chosen based on the capabilities of commercial units. The fuel gas generated during the process can be reused in a boiler or burner. The concept of the research is illustrated in Figure 1.

2.1. Feedstock

The feedstock used in this research comprises methyl ethyl ketone sourced from the industry, with a PMMA-glue content ranging from 3% to 5% and an MEK content ranging from 97% to 95%. During the cleaning process, the mass fraction of coating glue may increase with repeated cleaning cycles. However, due to the impurities in the glue, traditional distillation methods are inefficient in removing these components, leading to high recovery fees. Furthermore, the increasing incineration fees are attributed to the combustion instability of the glue content, which can also result in carbon tax charges. Pyrolysis is considered a viable method for removing impurities from the substrate, as it can handle the complex nature of the glue and its contaminants more effectively.

2.2. Lab-Scale System

The reactor body (Figure 2a) for this research was constructed using SUS 310, which is suitable for high-temperature operations. The reactor had a height of approximately 1300 mm and a diameter of 4 inches, with an effective volume of 10.7 L. To ensure uniform heat transfer and minimize heat transfer limitations between evaporation and cracking, larger pipe diameters were used. The internal flow channel was designed in a spiral shape to increase thermal cracking efficiency by changing the gas flow vector towards the outer barrel wall. The flow channel had an area of about 100 × 37 mm with a screw pitch of approximately 100 mm. A double pipe arrangement was used for nitrogen injection to stabilize gasification during the pyrolysis of MEK, allowing the MEK to drop down to the bottom of the reactor (Figure 2b). The double pipe arrangement also helped reduce clogging due to heat and minimized pre-oxidation reactions. Nitrogen was used as the inert gas to lower the oxygen content in the system. The furnace had a heat capacity of 10 kW and could reach temperatures of up to 1200 °C.

A Cyclone-heat exchanger constructed with SUS 304 was installed after the thermal cracking reactor to lower the flue gas temperature. The cyclone was included to minimize fouling effects on the heat exchanger tubes caused by particles generated during the pyrolysis process. The particles were separated from the gas stream by the cyclone and collected in a box at the bottom of the equipment. The gas stream, which still contained micro-sized particles, then passed through the heat transfer zone to further decrease the gas temperature. The design of the Cyclone-heat exchanger, as shown in Figure 3a, allowed for efficient use of space, increased safety, and higher cooling rates. Additionally, the separable cyclone design simplified the process of collecting clean carbon and oil, as depicted in Figure 3b. The experimental furnace machines used in this work had a heat capacity of 10 kW and could operate at temperatures of up to 1200 °C.

2.3. Field System

The field system was designed based on experimental machine tests, taking into account the optimized residence time for pyrolysis to ensure operational convenience. The sizing of the heat exchanger was estimated using Aspen Plus EDR results, as mentioned earlier. An overview of the plant can be seen in Figure 4. To avoid potential creeping deformation in the fully welded reactor due to high-frequency temperature fluctuations, AutoPIPE Vessel software was used to determine the appropriate thickness of the reactor. In particular, a thermowell tube was installed to monitor the temperature at the top of the cylinder instead of the side to prevent weld chipping.

2.4. Detection and Computational Detail

The product gas was analyzed using Nondispersive infrared gas detectors (NDIR, Molecular Analysis 6000i), Fourier-transform infrared spectrometer (FTIR, Jasco FTIR-6700), and Gas Chromatography (GC, Agilent 7890B). Fuel gas yield (mmole/g) was defined as the mass ratio of the sum of H₂, CO, CH₄, and C₂H₄ produced in the pyrolysis reactor.

This study does not consider the potential deviation of H and O resulting from the glue due to its negligible amount. The carbon ratio was determined through weight comparison to check against the theoretical reaction equation. The amount of carbon formed was measured by extracting samples from the reactors in each experiment to validate the equation. The fraction of gas mixtures was defined using the ideal gas principle and the following equations.

$${\mathrm{H}}_2 \mathrm{yield} ({\mathrm{Y}}_{{\mathrm{H}}_2})=\frac{{\mathrm{H}}_2 \mathrm{yield} \mathrm{in} \mathrm{the} \mathrm{reactor} \left(\mathrm{mmole}\right)}{\mathrm{MEK} \mathrm{fed} \mathrm{to} \mathrm{the} \mathrm{system} \left(\mathrm{g}\right)}$$

(1)

$$\mathrm{CO} \mathrm{yield} ({\mathrm{Y}}_{\mathrm{CO}})=\frac{\mathrm{CO} \mathrm{yield} \mathrm{in} \mathrm{the} \mathrm{reactor} \left(\mathrm{mmole}\right)}{\mathrm{MEK} \mathrm{fed} \mathrm{to} \mathrm{the} \mathrm{system} \left(\mathrm{g}\right)}$$

(2)

$${\mathrm{CH}}_4 \mathrm{yield} ({\mathrm{Y}}_{{\mathrm{CH}}_4})=\frac{{\mathrm{CH}}_4 \mathrm{yield} \mathrm{in} \mathrm{the} \mathrm{reactor} \left(\mathrm{mmole}\right)}{\mathrm{MEK} \mathrm{fed} \mathrm{to} \mathrm{the} \mathrm{system} \left(\mathrm{g}\right)}$$

(3)

$${\mathrm{C}}_2{\mathrm{H}}_4 \mathrm{yield} ({\mathrm{Y}}_{{\mathrm{C}}_2{\mathrm{H}}_4})=\frac{{\mathrm{C}}_2{\mathrm{H}}_4 \mathrm{yield} \mathrm{in} \mathrm{the} \mathrm{reactor} \left(\mathrm{mmole}\right)}{\mathrm{MEK} \mathrm{fed} \mathrm{to} \mathrm{the} \mathrm{system} \left(\mathrm{g}\right)}$$

(4)

$$\mathrm{Fuel} \mathrm{gas} \mathrm{yield} ({\mathrm{Y}}_{\mathrm{fuel}})={\mathrm{H}}_2 \mathrm{yield}+\mathrm{CO} \mathrm{yield}+{\mathrm{CH}}_4 \mathrm{yield}+{\mathrm{C}}_2{\mathrm{H}}_4 \mathrm{yield}$$

(5)

$${\mathrm{Y}}_{{\mathrm{H}}_2}=\frac{{\mathrm{F}}_{{\mathrm{N}}_2}}{1-({\mathrm{X}}_{\mathrm{CO}}+{\mathrm{X}}_{{\mathrm{CO}}_2}+{\mathrm{X}}_{{\mathrm{CH}}_4}+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}+{\mathrm{X}}_{{\mathrm{H}}_2})}\times \frac{{\mathrm{X}}_{{\mathrm{H}}_2}}{{\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}}\times \frac{1000}{24.5}$$

(6)

$${\mathrm{Y}}_{\mathrm{CO}}=\frac{{\mathrm{F}}_{{\mathrm{N}}_2}}{1-({\mathrm{X}}_{\mathrm{CO}}+{\mathrm{X}}_{{\mathrm{CO}}_2}+{\mathrm{X}}_{{\mathrm{CH}}_4}+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}+{\mathrm{X}}_{{\mathrm{H}}_2})}\times \frac{{\mathrm{X}}_{\mathrm{CO}}}{{\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}}\times \frac{1000}{24.5}$$

(7)

$${\mathrm{Y}}_{{\mathrm{CH}}_4}=\frac{{\mathrm{F}}_{{\mathrm{N}}_2}}{1-({\mathrm{X}}_{\mathrm{CO}}+{\mathrm{X}}_{{\mathrm{CO}}_2}+{\mathrm{X}}_{{\mathrm{CH}}_4}+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}+{\mathrm{X}}_{{\mathrm{H}}_2})}\times \frac{{\mathrm{X}}_{{\mathrm{CH}}_4}}{{\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}}\times \frac{1000}{24.5}$$

(8)

$${\mathrm{Y}}_{{\mathrm{C}}_2{\mathrm{H}}_4}=\frac{{\mathrm{F}}_{{\mathrm{N}}_2}}{1-({\mathrm{X}}_{\mathrm{CO}}+{\mathrm{X}}_{{\mathrm{CO}}_2}+{\mathrm{X}}_{{\mathrm{CH}}_4}+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}+{\mathrm{X}}_{{\mathrm{H}}_2})}\times \frac{{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}}{{\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}}\times \frac{1000}{24.5}$$

(9)

$${\mathrm{Y}}_{\mathrm{i}}=\frac{{\mathrm{F}}_{{\mathrm{N}}_2}}{1-({\mathrm{X}}_{\mathrm{CO}}+{\mathrm{X}}_{{\mathrm{CO}}_2}+{\mathrm{X}}_{{\mathrm{CH}}_4}+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}+{\mathrm{X}}_{{\mathrm{H}}_2})}\times \frac{{\mathrm{X}}_{\mathrm{i}}}{{\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}}\times \frac{1000}{24.5}$$

(10)

$${\mathrm{Y}}_{\mathrm{fuel}}={ \mathrm{Y}}_{{\mathrm{H}}_2}+{\mathrm{Y}}_{\mathrm{CO}}+{\mathrm{Y}}_{{\mathrm{CH}}_4}+{\mathrm{Y}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$$

(11)

where Y (fuel gas) = Gas yield (mmole/g-MEK), Xi = The concentration of the exhaust gas (i = CO, CO₂, CH₄, C₂H₄, H₂), Y_i = The yield rate of the exhaust gas (i = CO, CO₂, CH₄, C₂H₄, H₂).

As indicated by the equations above, the mixture amount was used to calculate the calorific value, which was evaluated using the following equation to determine the heat value burned by the burner.

$$\mathrm{GHV} :{ \mathrm{X}}_{{\mathrm{H}}_2}\times 67.82+{\mathrm{X}}_{{\mathrm{CH}}_4}\times 212.48+{\mathrm{X}}_{{\mathrm{C}}_2{\mathrm{H}}_4}\times 339.64+{\mathrm{X}}_{\mathrm{CO}}\times 67.76$$

(12)

MEK undergoes cracking into approximately four molecules per one MEK molecule in the first stage. This can be further multiplied by the gas expansion factor (f_T) to determine the effective volume flow rate. Subsequently, the residence times during the process can be calculated by dividing the volume flow rate in a fixed reaction zone using the following equations.

$$Q_{Fuel}=({\mathrm{F}}_{\mathrm{MEK}}\times {\mathsf{\rho }}_{\mathrm{MEK}}÷72.11\times 24.5\times 4+{\mathrm{F}}_{{\mathrm{N}}_2})\times f_T$$

(13)

$$f_T=\frac{\left(T+298.15\right)}{298.15}$$

(14)

where ${\mathrm{F}}_{\mathrm{MEK}}$ = the MEK flow rate (mL/min), ${\mathsf{\rho }}_{\mathrm{MEK}}$ = MEK density (g/cm³), $f_T$ = gas expansion factor.

The residual carbon portion is the residual carbon divided by the total inlet of MEK.

$$\mathrm{Coke} \mathrm{yield} ({\mathrm{Y}}_{\mathrm{coke}})=\frac{\mathrm{Carbon} \mathrm{black} \mathrm{in} \mathrm{the} \mathrm{reactor} \left(\mathrm{g}\right)}{\mathrm{MEK} \mathrm{fed} \mathrm{to} \mathrm{the} \mathrm{system} \left(\mathrm{g}\right)}\times 100\%$$

(15)

3. Results and Discussion

3.1. Model of Lab-Scale System

The main goal of this study is to ensure the continuous operation of MEK in the pyrolysis process while minimizing char and tar formation that may cause blockage. To achieve this, the reactor is designed based on the mini theoretical residence time, and the heat transfer phenomena are checked by calculating the flow zone using computational fluid dynamics (CFD). The overall heat transfer coefficient is determined by referring to the mass and heat transfer data of the ASHRAE Handbook [19] and Parry’s handbooks [20] for the reactor and heat exchanger. The gas flow rate is calculated using the equation mentioned in Equation (13), and the length and diameter of the reactor are determined using various calculation methods listed in

Table 1. The calculation of the experimental machine.

Model	Thermal Cracking Unit	Cyclone-Heat Exchanger Unit
Software	SOLIDWORKS Flow Simulation	SOLIDWORKS Flow Simulation
Mesh	Total cells: 1,416,169	Total cells: 3,341,314
	Fluid cells: 628,680	Fluid cells: 2,171,307
	Solid cells: 787,489	Solid cells: 1,170,007
Boundary Conditions	Gas Inlet: 36 g/min @ 298 K	Gas Inlet: 36 g/min @ 1173 K
	Real Wall: 1173 K	Water Inlet: 30 LPM @ 298 K
	Outlet: 1bar @ 298 K	Outlet: 1bar @ 298 K
Materials	Fluid: Syn-gas	Fluid: Methane&H2O
	Solid: Steel Stainless	Solid: Steel Stainless
Result	Heating Cylinder:	Tube Side:
	Average Heat Transfer Coefficient: 2.868 W/m2/K	Average Heat Transfer Coefficient: 529.6 W/m2/K
	Average Tube Wall Temperature: 1173 K	Average Tube Wall Temperature: 294.5 K
	Average Heat Flux: 2522.88 W/m2	Average Heat Flux: −626.3 W/m2
	Spiral Deflector:	Shell Side:
	Average Heat Transfer Coefficient: 4.79 W/m2/K	Average Heat Transfer Coefficient: 493.5 W/m2/K
	Average Tube Wall Temperature: 1105 K	Average Tube Wall Temperature: 294.0 K
	Average Heat Flux: 2344 W/m2	Average Heat Flux: 488.3 W/m2

. The heat transfer properties of gas convection were analyzed using SOLIDWORKS flow simulation, which calculated the temperature and velocity responses. The reactor and cyclone-heat exchanger were evaluated separately. The boundary condition was set with a mass flow rate of approximately 36 g/min, assuming that one mole of MEK vaporized and cracked into four moles of gas molecules at high temperatures. The reactor (thermal cracking unit) calculation assumed adiabatic conditions, as the primary reaction zone was covered by the furnace. The barrel wall temperature was maintained isothermally due to a sufficient energy supply. The cyclone-heat exchanger unit calculation also assumed adiabatic conditions to ensure cooling capacity while disregarding the effects of natural convection and radiation.

In Figure 5a, the results indicate the presence of a primary endothermic zone located at the bottom, with a temperature of approximately 650 °C. This temperature level provides sufficient energy for the evaporation and cracking of the solvent. Within the furnace, the barrel wall is maintained at 900 °C, and the heat from the furnace is transferred to the central tube. However, due to the spiral flow, the temperature distribution takes on a regular shape, resulting in a temperature difference of about 100–200 °C between the barrel wall and the central tube. The heat transfer near the central tube relies on the efficiency of heat transfer through the outer barrel. Perone reported a similar heat transfer phenomenon in a tubular heat exchanger using SolidWorks flow simulation [21].

As the gas progresses towards the outlet, it undergoes counter-heat transfer with the feedstock, causing a decrease in its temperature. This energy exchange contributes to the preheating of the feedstock within the reactor. The gas velocity remains constant in the spiral section (Figure 5b), maintaining a consistent speed of approximately 1.5–2 m/s. This steady velocity promotes effective thermal diffusion and radiation, facilitating the decomposition of the feedstock.

These temperature results indicate that the gases are adequately heated during the passing zone, promoting the reproducibility of the results. Following the temperature analysis, the heat exchanger was designed using Aspen Plus Exchanger Design and Rating (EDR), and the heat transfer coefficients were calculated using empirical equations, as shown in

Table 2. Parameter and result of the experimental heat exchanger by EDR.

NO	ASPEN EDR Input/Result	Unit	Value
1	Mass flow rate	kg/s	HotSide: 0.0036; ColdSide: 0.1657
2	Inlet temperature	°C	HotSide: 870; ColdSide: 25
3	Heat exchanged	kW	8.7
4	Allowable pressure drop	bar	HotSide: 0.11; ColdSide: 0.2
5	Fouling resistance	m2·K/W	HotSide: 0.00035; ColdSide: 0.00035
6	TEMA type	--	B-E m *
7	Tube OD/Pitch/Pattern	mm	20.7/29.5/30 Triangular-Unbaffled
8	Shell ID/OD	mm	203.2/208.92
9	Tube length	mm	1000
10	Number of tubes/Tube passes	--	37/1
11	Actual/Required area ratio	--	0.95
12	Film coefficient	W/m2/K	Overall fouled: 16.7 Overall clean: 16.9

* B: Bonnet bolted or integral with tube sheet; E: One pass shell; M: Bonnet.

. Fouling factors were referred to from previous research by Paul and Amitesh [22] to account for potential thermal resistance that may increase over time. The results provided a highly credible scheme for building the facility, including amplification factors. The fouling factor, pressure drops, and actual/required area ratio were crucial in the sizing/rating project. Figure 6a presents the analysis of the cyclone-heat exchanger, illustrating its role in cooling the gas mixture. The swirling motion within the cyclone promotes heat transfer, thereby maintaining higher temperatures. With a gas velocity of approximately 15 m/s, particles are effectively captured, as depicted in Figure 6b. The gas temperature, initially ranging from 500 to 600 °C, is progressively reduced until reaching around 30–40 °C. This level of cooling is satisfactory for subsequent analysis and provides valuable insights for simulating operational conditions in future large-scale production plants.

3.2. Thermal Pyrolysis Result of Lab-Scale System

In general, increasing the reaction temperature leads to an increase in the reaction rate. However, various factors, such as the composition of the gas mixture, reaction capability, volume expansion ratio, and system stability, also vary with temperature. It is important to note that higher temperatures may result in the formation of by-products, which can reduce the desired product yield. For example, at higher temperatures, vinyl may crack into methane, hydrogen, and fixed carbon, leading to the formation of polycyclic aromatic hydrocarbons (PAHs). Therefore, determining the optimum pyrolysis temperature is crucial to achieving desired results.

Table 3. The exhaust gas compositions of lab-scale system.

Chemical Compound	Inlet (% mol/mol)	Outlet (% mol/mol)	LOD (% mol/mol)	DRE
MEK	57.06	N.D.	0.221	>99.6%
CH4	--	31.78	0.138	--
C2H4	--	1.56	0.047	--
C2H6	--	N.D.	0.316	--
C2H2	--	0.07	0.059	--
C3H8	--	N.D.	0.174	--
CO	--	20.66	0.217	--
CO2	--	0.51	0.057	--

LOD = “limit of detection”, DRE = 1 − [(C_out)/(C_in)] = 1 − [0.221/57.06] = 99.6%.

presents the components of the exhaust gas of the lab-scale system, which was analyzed by FTIR. The results indicate that no methyl ethyl ketone (MEK) remains in the gas phase, suggesting that MEK can be effectively decomposed with a residence time of over 1.6 s. The hydrocarbon gas components are primarily composed of methane (CH₄), with respective values of 31.78%, 1.56%, and 20.66% for CH₄, C₂H₄, and CO. It is noteworthy that C₂H₄ concentration decreased over time during the long-term test, likely due to secondary decomposition reactions. This suggests that the primary cracking mechanism still occurs at lower temperatures, with reactions starting above 550 °C. However, the results indicate that the system reaches a steady-state condition after a few minutes of operation.

Figure 7a illustrates the gas composition at different temperatures, where each temperature condition is the average of three repeated experiments. It can be observed that hydrogen concentration increases with higher temperatures, as the increased energy promotes the cracking of vinyl into hydrogen, methane, and fixed carbon [10]. Methane concentration remains relatively stable at around 40% with a slight change as the temperature increases in the range of 750–950 °C, as ketones decompose into methane at 550 °C. Theoretically, complete decomposition of MEK can be achieved at temperatures between 550~650 °C [23]; however, the yield rate of the undesired product, tar, cannot be ignored during these temperature regions due to the poor heat conductivity of glue. Operating at higher temperatures can result in less formation of tar. Thus, a temperature range of 750–950 °C is considered more appropriate as a basis for amplification in this pyrolysis case. CO concentration decreases with increasing temperature as higher energy promotes the transfer of carbon monoxide into fixed carbon. On the other hand, CO₂ concentration barely changes with decreasing carbon monoxide, likely due to the oxygen-containing groups on the forming carbon surfaces.

In order to investigate the correlation between MEK flow rate and reactivity, both F_MEK and F_N2 were fixed to eliminate the volume expansion effect during the tests.

Table 4. Operation parameters of varied residence time tested via experimental machine.

Case	Temperature (°C)	MEK Flow Rates (mL/min)	N2 Flow Rate (L/min)	Retention Time
(a)	900	10	0.33	13.96
(b)	900	20	0.65	6.98
(c)	900	30	1	4.65

displays the different flow rates at fixed temperatures. The residence time was calculated by dividing the gas volume (using Equation (13)) by the reaction volume. Factors such as residence time, flow field, and heat transfer are closely related to the MEK flow rate and can significantly influence the cracking process. Figure 7b illustrates the gas compositions at different residence times. The results show that H₂ and C₂H₄ exhibit opposite trends with increasing residence time. Higher residence time promotes the formation of C₂H₄, H₂, CH₄, and CO₂ [24]. Additionally, CO is converted into char and CO₂ as residence time increases, and ultimately CO₂ decreases due to its reaction with char [25]. These results are consistent with the trends observed in Figure 7, indicating that residence time and operating temperature are key factors in the overall pyrolysis reaction.

Figure 8 depicts the tar accumulation on the top of the heat exchanger during a long-term test (4 h of continuous operation), with MEK flow rates of 10, 20, and 30 mL/min. The results indicate that higher residence times, resulting from lower MEK flow rates, can cause a pipe bridge effect and reduce continuous operability due to the accumulation of PAH on the wall. The temperature of the cone zone is decreased with the decrease in MEK flow rate, resulting in the accumulation of tar and enhancing the adsorption of char [24]. It thus hinders the stable operation of the system. However, this adverse effect gradually improves with increasing MEK flow rates. Eventually, the operation with MEK flow rates of approximately 30 mL/min demonstrates relatively excellent performance, as shown in Figure 8c, where the surface of the top plate is covered with dry char, which can be more easily scraped off. The optimal flow rate increases the cone temperature, resulting in a more fluent pyrolysis process.

Figure 9a illustrates the influence of residence time on the amount of carbon, showing a consistent trend of increased char amount from 2.12 wt% up to 8.15 wt% as the residence time increased from 4.65 to 13.96 s. The top plate view in Figure 8 shows tar flat sticks accumulated on the heat exchanger’s top plate during variant operations. It was observed that longer residence time resulted in increased temperature, which was conducive to the generation of carbon and char, in line with the findings of Sanchez [26]. However, this may not be beneficial for long-term operations. Additionally, Figure 10 shows the filters after the heat exchanger at different operating temperatures, indicating a proportional relationship between tar and char formation and oil colors, as reported by OH [27].

The higher heating value was observed to be around 4.65 s of residence time, as shown in Figure 10b. This indicates that gas tends to convert into fixed carbon or low carbon chain gases, with a burning value of approximately 150 kcal/mole in this research, accounting for about 70% of the heating value of natural gas. However, tar formation posed a challenge to continuous operations, making the optimal temperature at the cyclone a critical factor. To estimate tar reduction through heat treatment, the TGA system was used. 200 mg of tar was loaded into an alumina crucible and heated at a rate of 10 °C/min under an N₂ atmosphere. Figure 11 illustrates the weight loss of the samples at different operating temperatures of 800 and 900 °C. The sample at 800 °C showed a 1 wt% solid residue, with weight loss completed at approximately 400 °C (Figure 11a). On the other hand, the sample cracked at 900 °C showed a 40 wt% solid residue, with weight loss completed at around 600 °C during the TGA test (Figure 11b). These results indicate that higher temperatures tend to promote tar formation during the cracking process, leading to the formation of carbon and high boiling point polycyclic aromatic hydrocarbons. This further confirms that temperature and residence time are key factors in retarding tar and carbon formation. The tar can potentially be re-cracked at high temperatures to form solid residues on the cone side.

3.3. Field System

The field system was designed using SolidWorks, as shown in Figure 12. The design concept of the scaled-up system closely resembled that of the lab-scale reactor, with the main modifications being the adjustment of the spiral width and reactor volume to accommodate the difference in flow rate. The spiral width was kept fixed to prevent carbon accumulation near the inner tube during thermal decomposition. A larger space height could result in lower velocity and reduced heat transfer efficiency, as the energy transferred from the outer barrel wall may be compromised [28]. The pitch of the spiral remained fixed to meet the heat transfer requirements, with a furnace heat capacity of 66 kW and the reactor constructed using SUS 310 material. The design followed ASME VIII July 2017 standards, with a maximum allowable pressure of 1.8 kg/cm² at 900 °C. The 3D model during the design process is shown in Figure 12b. The heat transfer coefficient for the liquid-gas reaction was estimated based on previous works to be in the range of 30~120 W/m²/K [20]. The kinetic data used for design were obtained from Waring and Mutter’s research [9], with frequency factor and activation energy of 1.21 × 10¹⁵ and 67.2 kcal/mol, respectively, to determine the required reaction volume. The Aspen Plus PFR model with the non-random two-liquid (NRTL) thermodynamic model was used to calculate the heat of vaporization and correct the phase equilibrium at 1.0132 bar, as shown in Figure 13a, with regression based on actual experimental data [29]. The detailed calculations of the reaction phenomena in the various parts of the reactor are shown in Figure 13b, which includes three zones: (1) MEK vaporization zone, (2) cracking zone, and (3) stabilization zone. The first 0.5m is mainly for solvent gasification, with a requirement of at least 3.6 kW of heat and a temperature of approximately 525 °C. As the temperature rises, the solvent undergoes thermal cracking in the 0.5 to 2 m stage, with the temperature reaching 700℃. The design of the reaction zone ranging from 2 to 2.5 m is primarily intended for safety spare considerations. The entire process requires approximately 24 kW of energy consumption in this theoretical calculation.

The capacity of the field system, as shown in Figure 12a, was designed to handle approximately 15 tons of MEK per month. SUS 310 material with a thickness of about 10 mm and a body height of 2300 mm was used for the reactor. The practical volume of the reactor was approximately 163 L. The residence time, which varied depending on the flow rate of MEK ranging from 100 to 450 mL/min, was calculated to be around 4.7 to 19.2 s using the previously mentioned equation (Equation (13)). The heat exchanger, which incorporated a solid-gas cyclone shown in Figure 12c. The sizing of the heat exchanger was estimated using Aspen Plus EDR. The target cooling temperature was set at approximately 45 °C for burner-friendly use. The heat-exchanger zone was composed of 163 branches of an 800 mm steel tube, providing a total heat-exchange area of approximately 13.83 m². The cyclone was positioned in front of the tube bundle to allow particles larger than 50 μm to settle in the collection box, thereby minimizing fouling effects at the heat-exchange zone and ensuring long-term operational convenience.

The entire system in the factory, as depicted in Figure 14, shows promising results in terms of meeting the required capacity of the reaction zone. However, it is important to consider engineering amplification factors to account for process instability. The char kinetic parameters related to temperature exceedance were not fully consistence with the previous design. Nevertheless, the results obtained were reliable and will provide a basis for scaling up a stable system. The pyrolysis temperature of MEK falls within the range of 540~630 °C [10]. The lab-scale reactor operated at 750℃ to ensure complete cracking of MEK. However, during the operation of the field system, it was observed that a large amount of tar product formed from glue, making 750 °C unsuitable. Therefore, the operating temperature of the field system was set at 900 °C to ensure the complete conversion of glue impurities.

The cyclone-heat exchanger used to capture micro-sized particles was effective, albeit with a low efficiency of only 1% due to the small size of the soot formation (theoretically ranging from 10 to 300 nm [30]), which made it difficult for the cyclone to capture them. However, these particles tend to accumulate due to van der Waals forces, static electricity, and tar adsorption, eventually forming bulk groups that can hinder operational continuity. The actual diameter of the fine particles, as measured by an optical microscope (Olympus CX43), was shown in Figure 15a, and it matches the theoretical values of particle diameters, averaging about 100 to 300 nm. In general, a scrubber is often used to remove tar and particles from the exhaust gas. However, this design may lead to another issue of polluted water. To address this concern, filter equipment was implemented, allowing for further recovery of fine char particles for reuse. On the other hand, the cyclone heat exchanger was successful in collecting carbon with diameters above 50 μm, as demonstrated by the 3 h operation shown in Figure 15b. Due to the efficient separation performance of the cyclone device, the collected black carbon in the collection box is not sticky, making it convenient for the suction machine to convey, as shown in Figure 15c; Figure 15d is the particles diameters of carbon from the collection box tested by optical microscope, the particles accumulated forming about 50 µm or more than collected by the cyclone, but the structure was friable to separated by tested to diameter around 7 to 14 µm.

Figure 16 shows the natural gas burner (Zu How industry Co., Ltd., Zhongpu, Taiwan, G20) running situation, which was 100~200 K kcal/h capacity, equivalent to 192 to 384 L (NTP) methane. Based on the theoretical fuel-to-air ratio of 1:10, the flow rate of MEK was set at approximately 50–200 mL/min. After filtering particles larger than 1 micron from the flue gas, the burner can be successfully ignited. Figure 16a shows the diffusion flame combustion test, where the result demonstrated smooth yellow flame combustion of the fuel gas, indicating it can be used as the primary fuel in the burner. In Figure 16b, the torch color was blue due to the high concentration of hydrogen and carbon dioxide [27], as the MEK flow rate was set at 50 mL/min. Upon increasing the MEK flow rate to 200 mL/min in Figure 16c, the torch area expanded, and a sharp torch formed with no yellow flame, indicating that complete combustion was achieved. This shows that the presence of small particles smaller than 1 μm did not affect the quality of the flue gas. This shows that the fuel gas is suitable for use in the combustion system, both as an auxiliary fuel and as a primary fuel.

4. Conclusions

The aim of this study was to treat waste MEK-containing glue by subjecting it to temperatures ranging from 750 to 950 °C. The results indicate that higher temperatures reduced tar production and improved conversion, ensuring effective mixture conversion. Analysis of the flue gas of the steady-state operation revealed that MEK was fully decomposed into CH₄, C₂H₄, C₂H₂, CO, CO₂, and H₂, with approximate concentrations of 31.78%, 1.56%, 0.07%, 20.66%, 0.51%, and 30%, respectively. The pyrolysis process resulted in residual carbon of only about 5–7 wt%.

Furthermore, the quantity of exhaust gas produced was found to be suitable for reusing as a heat source, serving as a substitute for natural gas. The optimal pyrolysis conditions established in this study were a temperature of 900 °C and a residence time of approximately 5 s in the field system. However, while the cyclone-heat exchanger proved effective in cooling and capturing carbon solids larger than 50 μm, it encountered limitations in capturing smaller soot particles due to its limited cyclone ability. Therefore, alternative methods such as water absorption may prove more effective in removing soot particles, although this could lead to increased waste in water treatment. This aspect will be further investigated in future research. The exhaust gases directed into the natural gas burner successfully generated energy, indicating their potential utilization for power and steam generation. The utilization of the pyrolysis process to eliminate waste containing methyl ethyl ketone and glue offers several benefits in the industrial sector, including a substantial reduction in treatment fees and on-site heat recovery, leading to decreased carbon emissions. Additionally, the resulting carbon black can be repurposed for the production of valuable additives, such as those used in tire manufacturing or refined steel production, further enhancing the value of the pyrolysis product.

5. Patents

Pyrolysis reactor: Apparatus for fuelizing waste organic solvent and modular reactor. R.O.C Patent No. I660780.

Cyclone-heat exchanger: separating and cooling device and fueling system using the same. R.O.C Patent No. I732645.