Introduction

The production and consumption rate of plastics have been increasing drastically throughout the past decades. The global production rose from 1.5 Mt in 1950 to 311 Mt in 2014 [1, 2]. In Singapore, plastic wastes generated reached 869,100 tons in 2014, which accounted for 11.6 % of the total solid wastes generated. Nevertheless, the recycling rate remained rather low, at only 9 % [3]. The disposal of plastic wastes by landfill and incineration creates environmental problems due to its non-biodegradable nature and emission of hazardous compounds; hence, effort should be directed towards the recycling of plastic wastes in order to minimize its negative impact on the environment. In fact, plastic wastes can be viewed as valuable resources if recycled properly.

Among different feedstock recycling approaches for plastic wastes, pyrolysis is one of the most investigated method, since it can produce liquid hydrocarbon fuel of satisfactory quality, which is also compatible with the energy infrastructure in most countries that is heavily dependent on hydrocarbon fuels. In recent years, there have been a few companies and research groups attempting to translate the plastic pyrolysis technology into industrial application. For example, Anhui Oursun Resource Technology Co., Ltd [4], Plastic2Oil [5], Moriya et al. [6], and Zmuda [7] have developed catalytic pyrolysis process for plastic materials into relatively larger scale (≥0.4 tons/h for some groups). However, these applications do not accommodate PVC-containing waste plastic feedstock, since the decomposition of PVC produces chloride compounds which can cause corrosion to the reactors and pipelines, leading to high cost for maintenance and repairing. In addition, the chloride compounds could also deactivate some types of catalyst [8–10], hence lowering the performance of the process. The treatment of PVC-containing mixed plastic waste hence necessitates manual sorting to remove PVC from plastic feedstock prior to pyrolysis process, which could make the treatment less economical. As such, pyrolysis technology appeared to be less feasible for municipal plastic waste, which inevitably contains PVC. Thermal dechlorination, on the other hand, could be a simple, yet effective alternative to manual sorting of mixed plastic waste which contains PVC. There has been research on different thermal dechlorination approaches [11–13]; however, most of them were conducted in a batch or semi-batch manner, instead of continuous process which the derived information would be useful for engineering design and up-scale industrial operation. In addition, studies on continuous dechlorination and pyrolysis process as a whole for mixed plastic wastes remain rather scarce.

In this study, a continuous pyrolysis process, which was integrated with thermal dechlorination pre-treatment, was developed for feedstock recycling of PVC-containing mixed plastic wastes. The goal for this study was to understand underlying knowledge of the feedstock recycling of mixed plastic wastes via continuous dechlorination and subsequent pyrolysis, as well as to provide useful information for further improvement of the pyrolysis process in industrial field. Continuous thermal dechlorination of plastic feedstock was conducted in a vented screw conveyor, which could provide high dechlorination efficiency owing to its good heat transfer. A type of iron restructured clay (Fe-RC) catalyst, which has been proven by our group to improve oil yield and quality in mixed plastic pyrolysis [14], was also employed in attempt to maximize oil yield of the pyrolysis process. The effect of retention time on the dechlorination efficiency as well as effect of pyrolysis conditions (temperature and catalyst dosage) on product yield and oil composition were studied. Moreover, the performance of the system over a period of continuous operation was monitored and reported.

Materials and methods

Materials

The waste PE, PP, PS, PVC, and PET with a pellet size of 3 mm were obtained from a local plastic recycle company (Winrigo (S) Pte Ltd, Singapore). Mixed plastic waste sample used in this study was composed of 42 wt% PE, 35 wt% PP, 16 wt% PS, 5 wt% PVC, and 2 wt% PET. The composition of mixed plastic wastes used in this study was established according to the ratio of municipal mixed plastic waste in one of the incineration plants in Singapore. Each type of plastic waste was characterized and the properties are presented in Table 1.

Table 1 Properties of the plastic waste
Full size table

The iron restructured clay (Fe-RC) catalyst was synthesized in-house according to the method reported in our previous work [14]. Clays and iron(III) chloride used for catalyst preparation were obtained from Alfa Aesar (USA), while sodium hydroxide was obtained from Merck (Germany).

Experimental

Dechlorination study

The experimental setup for continuous dechlorination and catalytic pyrolysis of mixed plastic wastes is shown in Fig. 1. The feeding rate of plastic feedstock into the conveyor was manipulated by changing the rotational speed of the screw between 2 and 20 rpm. The vented screw conveyor, which consists of a horizontal segment and a vertical segment, was specially designed for melting and thermal dechlorination of plastic feedstock. The inner diameter of the screw conveyor was 80 mm, while the length of heating zone for the horizontal and vertical segments of the conveyor was 800 and 200 mm, respectively. The pitch of the screw was 40 mm. The temperature of the vented screw conveyor was set at 300 °C for dechlorination. Gaseous hydrogen chloride generated from dechlorination was exhausted from the vent holes located along the wall of the conveyor, and neutralized in a HCl trapper by 2 mol/L sodium hydroxide (NaOH) solution.

Fig. 1
figure 1

Schematic of the experimental setup

Full size image

For the studies of extrusion characteristic of screw conveyor (mentioned in “Extrusion characteristic of the screw conveyor”), the screw conveyor was isolated from the pyrolysis system and dechlorinated sample was collected at the outlet of the conveyor and weighed. The feeding of the sample was controlled at a 200 g interval and the feeding time was recorded.

For studies of dechlorination efficiency (mentioned in “Dechlorination performance”), the temperature inside the stirred tank reactor was set at 475 °C and no catalyst was added for the pyrolysis process.

Pyrolysis study

After dechlorination, molten dechlorinated plastic was conveyed into the stirred tank reactor where catalytic pyrolysis took place. The temperature inside the tank reactor was set at 425–500 °C and the speed of the stirrer was fixed at 20 rpm. In each catalytic pyrolysis run, Fe-RC catalyst (powder form) was added into the reactor through the catalyst hopper when dechlorinated plastic first entered the reactor. Nitrogen flow was constantly supplied to the tank reactor at a flow rate of 1 L/min to purge out the gas and oil vapour generated from pyrolysis. Pyrolytic oil vapour was condensed and collected in oil tanks. Incondensable pyrolytic gas was collected in a 1 L Tedlar® gas sampling bag for further analysis. The frequency of data collection for pyrolytic oil, pyrolytic gas, and temperature inside the tank reactor was fixed at 10 min. The data of pyrolytic oil yield, oil composition, and pyrolytic gas composition shown in this study were collected and taken for analysis after the continuous operation had reached steady state, and each of the above result presented is the mean value of at least three runs, whereby in each run, 6 data points were recorded during the steady state. Unlike oil and gas samples, solid residue sample was collected at once after completion of each run. The yield of the residues (excluding catalyst) and oil products was calculated as weight percentage (wt%) with respect to the plastic feedstock.

Analysis

Proximate analysis of the plastic waste samples was determined according to D3172-13 ASTM standard. The elemental composition of the plastic sample was determined by an Elementar® CHNS Elemental Analyzer. The higher heating value (HHV) was determined by an IKA® C2000 basic Calorimeter System equipped with a 260 ml IKA® C5010 calorimetric bomb.

The chlorine content of NaOH solution in the HCl trapper was determined by DIOMEX ICS-1100 Ion Chromatography system. The chlorine contents in feedstock, pyrolytic oil, and residue were measured according to the D808-11 ASTM standard.

The composition of the pyrolytic oil product collected was examined using a gas chromatograph (Agilent 7890B GC system, USA), which was equipped with Agilent HP-5 MS column (length: 30 m, I.D: 0.25 mm) and coupled with a mass spectrometer (Agilent 5977A MSD, USA).

The incondensable pyrolytic gas collected was analyzed by a gas chromatograph (Agilent 7890B GC system, USA) coupled with two thermal conductivity detectors (TCD) and a flame ionization detector (FID).

Calculation of feeding rate and retention time

With reference to Fig. 2, the feeding rate (Q f,i) and extrusion rate (Q e,j), which are represented by the slope of straight line between two adjacent data points in the feeding curve and extrusion curve, can be calculated by:

$$Q_{\text{f,i}} = \frac{{M_{{{\text{f,}}\,{\text{i}} + 1}} - M_{\text{f,i}} }}{{T_{{{\text{f, i}} + 1}} - T_{\text{f,i}} }}$$
(1)
$$Q_{\text{e,j}} = \frac{{\alpha (M_{{{\text{e,j}} + 1}} - M_{\text{e,j}} )}}{{T_{{{\text{e,j}} + 1}} - T_{\text{e,j}} }}$$
(2)
$$\alpha = \frac{1}{1 - \eta }$$
(3)

where M f,i+1 and M f,i refer to the cumulative sample mass of two adjacent data points in the feeding curve, M e,j+1 and M e,j refer to the cumulative sample mass of two adjacent data points in the extrusion curve. T f,i+1, T f,i, T e,j+1, and T e,j refer to the operating times at which adjacent data points were recorded. Coefficient α is included to compensate for the mass loss throughout the extrusion process due to removal of chlorine and moisture, as well as loss of small amount of hydrocarbon (in gaseous form), from mixed plastic feedstock during dechlorination step. η represents the percentage of the mass lost from the feedstock during dechlorination step.

Fig. 2
figure 2

Feeding and extrusion characteristic of the screw conveyor operating at 300 °C with different screw speeds: a 2 rpm, b 4 rpm, c 6 rpm, d 8 rpm, e 10 rpm, and f 20 rpm

Full size image

The changes in feeding and extrusion rate over adjacent data points along the curves can be represented by:

$$\Delta Q_{\text{f,i}} = Q_{{{\text{f,i}} + 1}} - Q_{\text{f,i}}$$
(4)
$$\Delta Q_{\text{e,j}} = Q_{{{\text{e,j}} + 1}} - Q_{\text{e,j}} .$$
(5)

The operation of screw conveyor was considered to have reached steady state only when both ∆Q f,i/Qf,i and ∆Q e,j/Qe,j are lower than 0.01 within the same operating duration. The mean feeding rate (Q f) and retention time of plastic feedstock inside the screw conveyor (RT) were calculated through Ordinary Least Squares (OLS) linear fit of the feeding curve and extrusion curve during the period of steady state. According to the law of OLS, the relationship between the cumulative sample mass (M) and operating time (T) for both feeding curve and extrusion curve during steady state can be described by:

$$M = aT + b$$
(6)
$$a = \overline{M} - b\overline{T}$$
(7)
$$b = \frac{{\overline{MT} - \overline{M} \overline{T} }}{{\overline{{T^{2} }} - \overline{T}^{2} }}.$$
(8)

The factor “a” for the feeding curve (a f) was defined as the mean feeding rate of the screw conveyor (Q f), where:

$$Q_{\text{f}} = a_{\text{f}} = \overline{{M_{\text{f}} }} - b_{\text{f}} \overline{{T_{\text{f}} }} .$$
(9)

The retention time of plastic feedstock inside the screw conveyor (RT) was calculated by the difference of the x-intercept (intercept of operating time) between extrusion curve and feeding curve, where:

$${\text{RT}} = \frac{{ - b_{\text{e}} }}{{a_{\text{e}} }} - \frac{{ - b_{\text{f}} }}{{a_{\text{f}} }} = \frac{{b_{\text{f}} }}{{a_{\text{f}} }} - \frac{{b_{\text{e}} }}{{a_{\text{e}} }}.$$
(10)

The mean feeding rate and retention time over different screw speeds are presented in Fig. 3. The mean feeding rate was shown to increase from 1.12 to 3.10 kg/h in a non-linear manner when the screw speed increased from 2 to 20 rpm. This was probably due to the increased slippage of melted plastic inside the screw conveyor at higher screw speed [12].

Fig. 3
figure 3

Mean feeding rate and retention time over different screw speeds at 300 °C

Full size image

Results and discussion

Dechlorination

Extrusion characteristic of the screw conveyor

The feeding and extrusion characteristic of the screw conveyor operating at 300 °C with different screw speed is shown in Fig. 2. As an example when operating at 2 rpm (Fig. 2a), the perfect steady state of screw conveyor operation was derived after 98 min, as indicated by the non-linear relationship between the cumulative mass of samples and operating time. After an operating time of 98 min, both the feeding and extrusion curves exhibited linear pattern with a similar slope, showing that the operation has reached steady state. When higher screw speeds were selected (4, 6, 8, 10, and 20 rpm), a similar extrusion characteristic was observed with a shorter lapse to reach steady state (see Fig. 2b–f). However, it should be noted that the slightly lower slop of extrusion curve compared with that of the feeding curve was due to the losses of chlorine content, moisture, and small amount of light hydrocarbons generated during dechlorination [11, 13].

Dechlorination performance

The dechlorination efficiency and chlorine distribution in pyrolytic products over different retention time of feedstock inside screw conveyor are listed in Table 2. It should be noted that, in this study, the organic chlorine fraction in gas generated from the dechlorination process was approximated to be zero and the dechlorination efficiency was calculated solely based on the amount of inorganic Cl (HCl) trapped by the NaOH solution. The approximation was made to simplify the analysis based on the fact that the organic chlorine fraction in gas generated from the dechlorination process is normally negligible (<0.5 wt% of total chlorine in gaseous form) [15]. As shown in Table 2, the dechlorination efficiency increased with a longer retention time. High dechlorination efficiency (over 91.86 %) can be achieved when retention time was longer than 35.5 min in this setup. It is worth mentioning that the results are somewhat similar to that obtained by the other group in batch experiments at a much smaller scale (more than 200-fold smaller) [16]. This suggested that the specially designed screw conveyor used in this study is an effective, suitable type of reactor for large-scale thermal dechlorination of plastic materials.

Table 2 Dechlorination efficiency and chlorine distribution in pyrolytic products
Full size table

Nevertheless, majority of the chlorine that was not removed during the dechlorination process existed in gaseous form (1.72–19.97 wt%) after the pyrolysis process at a higher temperature, rather than remaining in pyrolytic oil (0.02–0.13 wt%) and residue (0.13–0.21 wt%). In general, chlorine remained in oil products accounted for only less than 1.2 % of the total chlorine content in all pyrolytic products (pyrolytic oil, residue, and gas). It is also worth noting that the chlorine concentrations of the pyrolytic oil obtained from all runs were in the range of 6.08–39.50 ppm, which is much lower than the allowable concentration (<100 ppm) specified for reclamation pyrolytic oil in Japan [17]. This indicated that the pyrolytic oil produced from the continuous dechlorination-pyrolysis process is safe to be used without requiring additional chlorine removal treatment.

Although a near-complete dechlorination is attainable with long retention time, the long processing time is not favoured for industrial application. Hence, a balance needs to be struck between processing time and dechlorination efficiency. In this study, a dechlorination efficiency of approximately 92 % and corresponding retention time of 35.5 min, which was achieved by selecting a screw speed of 4 rpm (see Table 2), were considered reasonable and the parameter was used for the subsequent catalytic pyrolysis studies in the following sections. According to Fig. 3, the selected parameter corresponded to a mean feeding rate of 1.65 kg/h during operation at steady state.

Effect of temperature on pyrolytic products

Pyrolysis yields

The yields of pyrolytic products over different pyrolysis temperature are shown in Fig. 4. The results for experiments without catalyst are also presented for comparison. At 425 and 450 °C, catalytic pyrolysis exhibited higher oil yields (78.37 and 75.27 wt%, respectively) and lower gas yields than non-catalytic runs, indicating that pyrolytic oil production was promoted by the Fe-RC catalyst, which was attributed to the moderate acidity of the clay-based catalyst that leads to lower tendency of over-cracking of hydrocarbons [18–22]. The results were consistent with our previous batch pyrolysis experiments which employed the same type of catalyst at a much smaller scale (10 g of feedstock in tube reactor) [14], indicating that up-scaling of the catalytic pyrolysis process using the stirred tank reactor employed in this study was successful in terms of oil yield.

Fig. 4
figure 4

Yields of pyrolytic products over different temperatures (mean feeding rate at steady state: 1.65 kg/h, Fe-RC dosage for catalytic runs: 30 g)

Full size image

As the temperature increased to 475 and 500 °C, the oil yield of catalytic pyrolysis decreased considerably. This was accompanied by an increased production of pyrolytic gas. In addition, the difference in oil yield between catalytic and non-catalytic runs became smaller at such high temperatures. At 500 °C, the oil yield of catalytic pyrolysis became virtually the same as that of non-catalytic pyrolysis. This could be due to the higher occurrence of spontaneous thermal decomposition of hydrocarbons.

The relatively high yields of solid residue obtained in both thermal and catalytic pyrolysis runs were mainly attributed to the ash content in the feedstock (7.26 wt%), instead of the carbon residue generated from the plastic materials in the feedstock. In fact, the amount of carbon residue was low in all pyrolysis runs, indicating that high conversion of the mixed plastic wastes was achieved.

Composition of pyrolytic oil

Figure 5 exhibits the carbon number distribution of pyrolytic oil products over different temperatures. The results for both non-catalytic and catalytic pyrolysis are presented for comparison. The oil products obtained from non-catalytic pyrolysis runs were a mixture of C6–C30 compounds which exhibited a relatively even carbon number distribution across different fractions. It was due to the random scission of the plastic polymers under the sole effect of temperature. On the other hand, at 425 and 450 °C, Fe-RC played a role in promoting the yield of C9–C12 and C13–C19 fraction, giving rise to relatively skewed carbon number distribution patterns. The largest increase of the C9–C19 fraction was observed at 450 °C (73.38 wt%), while in the case of non-catalytic pyrolysis, the proportion of C9–C19 fraction remained rather constant (around 56–60 %) across different pyrolysis temperature. The selectivity observed in the catalytic runs could be attributed to the moderate acidity and suitable textural property of this catalyst which led to higher tendency to crack plastic polymers into C9–C12 and C13–C19 products. It is worth noting that the C9–C12 and C13–C19 fractions are the major constituents of kerosene and diesel fuel, respectively. This shows that Fe-RC is a suitable catalyst candidate of the pyrolysis of plastic wastes for the production of kerosene and diesel fuel. Similar to the yields of oil product discussed in “Pyrolysis yields”, the effect of catalyst on pyrolytic oil composition became insignificant when the temperature was increased to 475 and 500 °C, as evidenced by a similar oil composition as compared with the non-catalytic runs.

Fig. 5
figure 5

Carbon number distribution of oil products over different temperatures (mean feeding rate at steady state: 1.65 kg/h, Fe-RC dosage for catalytic runs: 30 g)

Full size image

From this study, temperature range of 425–450 °C was considered the optimal for catalytic pyrolysis of mixed plastic waste with Fe-RC. This is in agreement with a study on a different type of clay-based catalyst which reported that high catalytic performance could be achieved at 450 °C [23].

Composition of pyrolytic gas

The molar percentage of different pyrolytic gas constituents is listed in Table 3. The pyrolytic gas products were essentially composed of H2, CO, and CO2, as well as hydrocarbons with carbon number ranging from C1 to C5. In both thermal and catalytic pyrolysis, larger amounts of methane (CH4), yet lower content of heavier hydrocarbon gas (C3–C5) were obtained with increasing temperature. This is due to more severe cracking of hydrocarbons at higher temperature. Furthermore, the amount of H2 generated increased at higher pyrolysis temperature. This was probably due to increased occurrence of aromatization reaction, which produces H2 at high temperatures [24, 25].

Table 3 Compositions of the pyrolytic gas (mol %) (mean feeding rate at steady state: 1.65 kg/h, Fe-RC dosage for catalytic runs: 30 g)
Full size table

Effect of catalyst dosage on pyrolytic oil yield

Figure 6 displays the yields of pyrolytic oil over different Fe-RC dosages. In general, at lower pyrolysis temperatures (425 and 450 °C), the oil yield increased with the increase of Fe-RC dosage from 30 to 60 g. The highest oil yield of 83.73 wt% was obtained with a Fe-RC dosage of 60 g at 450 °C and feeding rate of 1.65 kg/h in this setup. When Fe-RC dosage was continuously increased to 90 g, the oil yield was found to decrease slightly to values similar as that with 30 g Fe-RC. Further increase in Fe-RC dosage resulted in further reduction in oil yield to less than 70 wt%. The poorer oil yield at high catalyst dosage was probably attributed to the bifunctional nature of the catalyst [22]. The active sites of the catalyst was considered to be insufficient when low dosage of Fe-RC adding in, resulting in less interaction with the reactant and lower oil yield. Oppositely, the density of the active sites increased and over-cracking reactions were promoted when higher dosage of Fe-RC was added in, also resulting in poorer oil yield.

Fig. 6
figure 6

Oil yields over different Fe-RC dosages (mean feeding rate at steady state: 1.65 kg/h)

Full size image

The influence of the Fe-RC dosage on the oil yield appeared to diminish when temperature was raised to 475 °C. At 500 °C, the changes in oil yield over different Fe-RC dosage became insignificant for all catalyst dosage used. This further supports the possible scenario mentioned in “Pyrolysis yields”, whereby spontaneous thermal decomposition of hydrocarbons became dominant over catalytic cracking at high temperature. As a result, a low oil yield (<60 wt%), which is similar to that of non-catalytic decomposition, was obtained regardless of the amount of catalyst being added into the process.

Time-lapse performance of the pyrolysis process

Figure 7 demonstrates the change of pyrolytic oil yield and temperature fluctuation inside the stirred tank reactor throughout an operating duration of 230 min under the optimized conditions (target pyrolysis temperature of 450 °C and Fe-RC dosage of 60 g). During the first 30 min of operation, the actual temperature inside the reactor was maintained at the set temperature as expected and there was no pyrolytic oil generated, since the plastic feedstock was still inside the screw conveyor and had not reached the pyrolysis reactor. Rapid increase in pyrolytic oil yield, along with sharp drop of temperature inside the reactor, was observed from 40 to 110 min, indicating that dechlorinated feedstock entered the reactor and catalytic pyrolysis commenced. The reduction in temperature within this period was most likely due to the fact that thermocouple inside the reactor started to contact with the relatively “cold” dechlorinated feedstock (300 °C) which first entered the reactor. Since plastic materials generally have poor thermal conductivity, it would require a period of time for the reactor to re-establish the target temperature.

Fig. 7
figure 7

Pyrolytic oil yield and actual temperature inside the stirred tank reactor throughout operating duration of 230 min (mean feeding rate at steady state: 1.65 kg/h, target temperature: 450 °C, Fe-RC dosage: 60 g)

Full size image

The pyrolysis process was considered to reach steady state from 110 min to 180 min, as evidenced by the fact that oil yield and temperature stabilized at around 82–85 wt% and 450 °C, respectively. However, the oil yield and actual temperature began to drop when the operating duration exceeded 180 min. This is probably due to the ever-increasing residue (ash in feedstock and carbon residue generated in pyrolysis) which already fully accumulated at the bottom of the reactor began to deposit onto the side of the inner surface, resulting in poorer heat transfer from the reactor to plastic feedstock. In fact, the duration of the high-oil-yield steady state lasted for around 70 min and was even shorter than the time required for the system to attain steady state from start. This is not favourable for industrial application at large scale which demands stable, optimal operation over long operating duration.

In that regard, the residue should be discharged in a continuous or regular manner to prolong the steady-state operation. Continuous residue discharge may be achieved by either incorporating a well-controlled continuous residue discharging device at the bottom of the tank reactor, or employing a screw conveyor as the pyrolysis reactor.

Conclusion

A continuous pyrolysis system (feeding rate >1 kg/h) integrated with thermal dechlorination pre-treatment was developed for feedstock recycling of PVC-containing mixed plastic waste.

High dechlorination efficiency of over 90 % was achieved with a retention time longer than 35.5 min at 300 °C in this setup. After dechlorination pre-treatment at retention time of 15.6–50.3 min, the concentration of Cl in pyrolytic oil was found to be well below the maximum amount specified for reclamation pyrolytic oil in Japan.

At pyrolysis temperature of 425 and 450 °C, Fe-RC catalyst significantly improves the oil yield and selectivity for the C9–C12 and C13–C19 oil fraction, demonstrating that it is a suitable catalyst for the pyrolysis of mixed plastic for kerosene and diesel production. Furthermore, Fe-RC dosage of 60 g and temperature of 450 °C were found be the optimum pyrolysis conditions as it provided the highest oil yield of 83.73 %. The H2 gas production was also enhanced by Fe-RC, especially at 450 and 475 °C.

Overall, the continuous system exhibited high stability and oil yield (82–85 %) throughout the steady state, which was attained after an operating time of 110 min. However, the duration of steady state was rather short due to accumulation of residue. Modification of the pyrolysis reactor was hence necessary to allow continuous residue discharge and prolong the steady state.