Optimization of diesel engine performance and emission using waste plastic pyrolytic oil by ANN and its thermo economic assessment

doi:10.21203/rs.3.rs-2477405/v1

The current study focuses on the engine performance and emission analysis of a 4 stroke compression ignition engine powered by medical waste plastic oil (WPO) followed by their optimization study and economic analysis. Engine tests were conducted using WPO blended diesel at various proportions to acquire required data for training the ANN model, which enables better prediction for the engine performance by making use of the standard back-propagation algorithm.Considering supervised data obtained from repeated engine tests, an artificial intelligence-based model of ANN was designed to select different parameters of performance and emission as output layers, at the same time engine loading and different blending ratio of the test fuels were taken as the input layers. The ANN model was built up making use of 80% of testing outcomes for training. The ANN model forecasted engine performance and exhaust emission with regression coefficients (R) at 0.989–0.998 intervals and a mean relative error from 0.002% to 0.348%. Such results illustrated the effectiveness of the ANN model for estimating emissions and the performance of diesel engines. This study demonstrates the use of artificial neural networks (ANNs) to forecast a multi-component fuel mixture, which is novel and reduces the amount of experimental effort required to determine the engine output characteristics. Moreover, the economic viability of the use of 20WPO as an alternative to diesel was justified by thermo-economic analysis.

Medical plastic wastes

catalytic pyrolysis

performance

emission

ANN

economic analysis.

1. Study of performance and emission diesel engine fuelled with WPO blended diesel.

2. Development of an artificial intelligence-based ANN model at specific engine conditions.

3. Analysis of productivity of ANN model developer for this experiment and test the economic feasibility.

Huge demand for energy, rapid depletion of fossil fuels, a threat from environmental issues, and lack of facilities for proper disposal of municipal wastes evokes new research schemes to find out an alternative sustainable source for conventional petro fuels. Conversion of untreated waste plastics into fuel oil for CI engine by pyrolysis brings dual benefits towards resolving the challenges for energy security and waste disposal (Damodharan 2018). It also mitigates the challenges of plastic waste generation from manufacturer to bulk users in global scenario prospects. In this regard, attention towards the technique of conversion plastic wastes into fossil fuel substitutes is highly encouraged. An extensive literature study reports that waste plastic oil has been used in diesel engines because of its greater stability and drivability with no engine alteration (Mani 2011). The investigational outcomes fueling engine with WPO was quite comparable with diesel. However, it has been reported that there was a significant increase in emissions that occurred using WPO in the CI engine (Das 2022).

The summary of the results of the different experiments carried out by different researchers is described as follows. Mani et al. explored the pattern combustion, emission and performance of a 4-stroke CI engine by using 100% waste plastic oil with brake thermal efficiency compared to diesel by 75% of maximum power and higher exhaust emissions like NO_x, CO, and unburned HC for all loads (Mani 2009).A sequence of operations was conducted in the same engine by using different concentration of WPO-diesel fuel mixture to study the combustion, performance, and emission and the obtained results were reported appreciable (Mani 2011). Mitsuhara et al. also claimed that WPO obtained from different polymers, blended with diesel demonstrated identical performance and emissions.CO and NOx were also reported lower for higher blends of WPO in diesel (Mani 2009). To assess a high-speed CI engines performance and emission, Kumar et al. attempted varied combinations of diesel with WPO produced from HDPE. Their investigation results indicated that BTE remained lower for all blends irrespective of loads whereas BSFC, NO_x, CO, and unburned HC exhibited elevation at a higher concentration of WPO in fuel mixture owing to lesser calorific value and reduction in higher loads. The emission features of diesel engine using a blend of WPO and diesel was investigated by Kaimal and his colleagues. When comparing pure WPO to a blend of WPO and diesel, emissions of NO_x and smoke were reduced by up to 17.8% and 22%, respectively; however, BSFC, UHC, and CO emissions of the engine were increased (Kaimal 2016).Senthil Kumar et al. studied the significance of JME emulsion on performance and emission characteristics of a fuel mixture (10% WPO + 90% diesel) operated in a CI engine. They assured better BTE, EGT and HRR limited to 20% of JME in a 10% WPO mixture at highest engine loading conditions. BSFC was found to enhance at lower loads because of its poor heating value and decreased at higher loads irrespective of fuel blends. Reduction in emissions of HC, CO, and smoke in a mixture of JME–WPO and elevation in NOx emission were reported because of a poorer fumigation rate of the blended fuel than that of WPO (Senthil kumar 2016).Güngör et al. concluded that the performance of an engine in a multi-cylinder CI engine could be better when fuelled with WPO in diesel blends up to 5% (by volume)( Güngör 2015).The influence of reduction in injection timing from23°CA bTDC to 9°CA bTDC on engine performance was studied by Mani and Nagarajan when fuelled with WPO and revealed that thermal efficiency and smoke emission were found to increase with a decrease of NOx, CO, and HC emission under all conditions (Gnanasekharan 2016). The same group tested a WPO, water, and surfactant Span-80 emulsion and discovered that peak cylinder pressure and heat release rate were lower for the emulsion than for diesel. Furthermore, under all engine loading circumstances, an emulsion of WPO with 30% water to that of diesel resulted in a decrease in NOx and smoke emissions and an increase in HC and CO emissions. Thermal efficiency was reported higher than that of diesel by 3% (Kumar 2016). Panda et al. studied the engine performance by using pyrolytic WPO in a direct injection CI engine and claimed better engine operation up to 30% by volume of WPO to diesel. Kalargaris et al. investigated a turbocharged direct injection CI engine by using 25 to 90% by volume of WPO-diesel mixtures and declared the occurrence of an extended delay period and higher HRR owing to a lesser cetane number of WPO (Kalargaris 2017). In our earlier work, energy and exergy analysis on pyrolytic WPO in the CI engine by a varying fraction of WPO in diesel fuel. BTE was also reported slightly higher with comparatively lower BSFC.While NO_x and HC emissions were reported higher in the WPO-diesel blend to that of diesel (Das 2022). From the above literature, it is understood that numerous studies on diesel engine performance have been reported but no such efforts were made to correlate the fuel input with engine output using WPO blended diesel for optimum efficiency.

In this context, a few researchers have attempted to optimize engine performance and emission by implementing the technique of artificial intelligence by using waste cooking biodiesel (Pai 2011), Polango biodiesel (Sharma 2016), diethyl ether (Uslu 2018), and CNG with diesel in dual fuel mode (Yusaf 2012). Such a technique reported significant outcomes in terms of engine performance and emission. There was no work yet reported using WPO in this regard. So, the present work intends to use the efficacy of artificial intelligence for better prediction of performance and emission of the test engine with optimized variables in agreement to the experimental outcomes.

The present work used ANN a soft tool enabled by artificial intelligence (AI) for high accuracy in forecasting engine performances and emissions analysis using a different combination of WPO. Such Predictive modeling was applied derived from ANN search algorithm to associate the performance and emission with the input parameters such as a different concentration of WPO in test fuels and various engine loads. The formulated models accurately predicted the engine performance factors with a high accuracy congruent to experimental results. This proposed prediction model is also able to map complex correlations among different engine output and input parameters with high accuracy and consistency by using the help of a systematically trained ANN. The engine performance of a gasoline engine employing waste pomegranate ethanol mixes is predicted in this study by using ANN (Atik 2013). The performance of the engine was analyzed and evaluated using an artificial neural network model. The studies were carried out using a mixture of plastic pyrolysis oil, diesel, and ethanol in a CI engine. The correlation coefficient (R) between 0.964–0.9816 indicates that the prediction model has very good agreement (Dhande 2022). High levels of accuracy were achieved in successfully obtaining the forecast values. It explains how the engine output parameters are predicted and validated using ANN (Khandal 2021). In addition to this, a thermo-economic analysis has also been made to address the significance of waste plastic oil as a cost-effective alternative fuel to diesel without any modification in engine operation.

2.1 Characterization of pyrolysis fuel

Medical plastic wastes utilized in this experimentation were comprised of the waste syringe and saline bottles, which were procured from various hospitals. For the experiment, the used saline bottles and syringes were taken in a ratio of about 70:30 ratio by weight proportionate to the waste generation. The shredded medical plastic waste sample along with 10% Zeolite A catalyst was used as the feedstock for the pyrolysis process. The pyrolysis process was operated in a stainless-steel batch reactor as reported in our previous study assisted by a PID controller at an optimum temperature of 500°C.The vapor generated inside the reactor was condensed using a water-cooled condenser to obtain the waste plastic oil (WPO) The details about the pyrolysis reactor set up and experimental programme is mentioned in our previous work (Das 2020).

The oil produced through pyrolysis was subjected to GC–MS test to determine its chemical composition. The carbon chain structure of the waste plastic oil is summarized in Table 1 and from the table, it is clear that the oil carries major fraction of hydrocarbons with a range of C₁₁-C₁₂.In this current study of engine performance, a mixture of standard diesel with different proportions of medical waste plastic oil (WPO) has been taken as test fuels. The emulsified WPO proportions with diesel by 10, 20 and 30% (by volume) were represented as 10WPO, 20WPO, and 30WPO, respectively. Table 2 summarizes the main fuel properties of diesel collectively with plastic oil fuel. Plastic oil is having a lower density, lower flash point, and fire point that of diesel. The gross calorific value of plastic oil is marginally superior to that of diesel.

Table 1

GC-MS analysis of WPO
Carbon chain structure	Volume (%)
C₁₀	4.12
C₁₀-C₁₁	11.81
C₁₁-C₁₂	47.6
C₁₂-C₁₃	19.12
C₁₃-C₁₈	2.9

Table 2

Fuel characteristics of different fuels
Fuel Properties	Diesel	WPO	10WPO	20WPO	30WPO	ASTM standard
The density of fuel at15 ^ο C, (g/m³)	835	793	830.8	826.6	822.4	D 1298
Fuel flash point (^ο C)	52	5	47.3	51.06	37.9	D 93
Fuel fire point (^ο C)	57	9	52.2	47.4	40.8	D 97
Fuel GCV (MJ/kg)	45	46	45.1	45.2	45.3	D 240
Fuel Viscosity (cSt) @40 ^ο C	2.15	3.75	2.31	2.47	2.63	D 445
Cetane number	42	31	40.9	41.78	38.7	D976

2.2 Engine setup and Procedure

To conduct the experiment, a test engine (Kirloskar Make) having four stroke, direct injection, single cylinder water cooling assisted diesel engine integrated with an AC generator, was employed. The engine was operated with a bore diameter of 87mm, a highest power rating of 5.2kW@ 1500 rpm, and injected fuel pressure of 200 bar. Figure 1 describes the photographic view of the experimental setup. The test engine comprises of eddy current dynamometer employed for engine loading, a load cell, a fuel metering system, and dedicated sensor systems for temperature measurements at specified positions. Moreover, a real-time data acquisition system was integrated with the set up for providing precise and accurate engine performance results and an emission analyzer (AVL make) was also employed for computing the major pollutants such as CO, HC, and NO_x concentration in the exhaust gas. The trial was conducted by taking diesel as pilot fuel to run the engine. Thereafter, all the test fuels of different proportions of waste plastic oil with diesel were inducted as the primary fuels to the engine for the required measurement. The engine efficiency, fuel consumption, and exhaust emissions were measured and considered for various loads and blends of waste plastic oil. The tests were concluded by flushing out the retained test oil from the fuel induction system. The experiment was conducted at a compression ratio of 18 and injection timing of 23° b TDC (CA).Every experiment has been repeated three times and the mean was taken into consideration in order to minimize the errors and exhibits better prediction.

2.3 Error and Accuracy measurements

While experimenting, calibration of the test device is required to establish before trial to avoid the chances of errors and uncertainties during measurement. However, some inherent and observational errors and uncertainties are accountable for estimation accuracy. The amount of uncertainties associated with various parameters of test engines were calculated as illustrated in Table 3. An analysis made on uncertainty was also evaluated by following Eq. (1).

$$Y=\sqrt{{{X}_{1}}^{2}+{{X}_{2}}^{2}+{{X}_{3}}^{2}+{{X}_{4}}^{2}+{{X}_{5}}^{2}+{{X}_{6}}^{2}+{{X}_{7}}^{2}}\left(1\right)$$

$$Y=\sqrt{({1)}^{2}+{\left(0.2\right)}^{2}+{\left(1\right)}^{2}+{\left(0.2\right)}^{2}+{\left(0.2\right)}^{2}+{\left(0.2\right)}^{2}+{\left(1\right)}^{2}}$$

$$Y=\pm 1.78\%$$

Table 3

Assessments of test engine Accuracy and Uncertainties
Sl. No	Instruments	Calibration	Accuracy	Uncertainties (%)
1	Manometer	0-50mm	$\pm 1 \text{m}\text{m}$	± 1
2	Fuel meter	0-100 cc	± 0.01cc	± 1
3	Load indicator	0-100kg	± 0.01kg	± 0.2
4	RPM Meter	0-1000 rpm	± 10 rpm	± 0.1
5	Exhaust Gas Analyzer	CO: 0–10% CO ₂: 0–20% HC: 0–10,000ppm NO_x: 0-5000ppm	± 0.02 ± 0.03 ± 20ppm ± 10ppm	± 0.2 ± 0.15 ± 0.2 ± 0.2
6	AVL Smoke meter	0–31,999 mg/m³	± 0.02 mg/m³	± 0.5
7	Exhaust gas temperature indicator	0-1000°C	± 1ºC	± 0.15

2.4 Modeling of Neural Network (ANN)

ANN is a computational model designed to reduce the human mental efforts to execute intellectual functions and decision-making devoid of receiving any support. The model is well composed of three phases as input layer, an output layer, and some hidden layers. Every layer is consisting of the exact number of neurons treated as processing elements. The neurons are assembled through communication links and attributing connection weights. On the other hand, weights of connection contribute remarkably by transmitting a signal to neurons (Gnanamoorthi 2021).The efficient neural architecture is described based on neurons with their hidden layers (Kiani 2010).The log-sigmoid transfer function (logsin), regarded as the network output layer in the system, is described with a linear (purelin) ( Ghobadian 2010). The activation function represents the layers of the neural network in the system (Delucas 2001). The ANN data prediction system was constructed to represent the accuracy of a robust ANN system (Canakci 2006) .The ANN model is designed to anticipate the best performance and emission characteristics of the test engine.

ANN anticipation on six yield responses is significant for comprehension of the general combustion and emissions arrangement within the engine. ANN is especially valuable in diminishing the duration and finance required for full-scale mapping, providing ground contemplates while enormous varieties in control systems of the test engine. Figure 2 represents the multilayered neural network comprising of engine load and fuel blend as input factors where as different engine performance and emission characteristics such as BSFC, BTE, EGT, CO, NO_X and HC as output factors. The optimum setting of the ANN operation is summarized in Table 4. The function of the hidden layers is to recognize features from the input data and use such information to correlate between a given input and the exact output. The hidden layers execute computations on the weighted inputs and produce net inputs which are subsequently applied with activation functions to produce the actual output as described in Fig. 2.

In conventional methods, several experiments are to be conducted under different engine conditions by using different test fuels to assure the best outcome of results in terms of engine performance and emissions, which consumes more time and spends much money .On the other hand, researchers now prefer computer-assisted soft tools like ANN to achieve more accuracy in results with a few experimentations in less time. ANN technique becomes useful and popular to forecast the most possible outcomes of engine performance and emission under various engine conditions in terms of its fuel blend and engine load.

Table 4

Optimum settings of the ANN
ANN parameter settings for experiments
ANN Network category	Feed-forward back-propagation
Network learning function	LEARNGDM
The function used for performance	MSE
Function used in training	TRAINLM
Transfer functions	Tansig/Purelin
Neurons used	20

2.5 Economic analysis of fuels used in the engine

Such a study enables us to visualize an alternative to existing fuel in its scope of cost- competitiveness, sustainability, and market demand. It is also used as an appropriate measure to advocate the significant competency of the alternative fuel in terms of its economic viability and technical feasibility.

For a balanced condition control volume of a test engine, the fuel economic equilibrium equation can be expressed as (2), [17].

$$\sum {\dot{\text{C}}}_{\text{f}}+ {\dot{\text{C}}}_{\text{W}} ={\dot{\text{C}}}_{\text{Q}}+ \sum {\dot{\text{C}}}_{\text{g}}+\dot{\text{Z}} \left(2\right)$$

The above correlation consists of ${\dot{\text{C}}}_{\text{W}}$ and ${\dot{\text{C}}}_{\text{q}}$ which point out cost rates of work done and heat transfer as described:

$${\dot{\text{C}}}_{\text{W}}= {\text{c}}_{\text{W}}{\dot{\text{E}}}_{\text{W}} \left(3\right)$$

$${\dot{\text{C}}}_{\text{Q}}= {\text{c}}_{\text{Q}}{\dot{\text{E}}}_{\text{Q}} \left(4\right)$$

${\dot{\text{C}}}_{\text{f}}$ and ${\dot{\text{C}}}_{\text{g}}$ signify the cost rates of fuel inlet and exhaust gas outlet

$${\dot{\text{C}}}_{\text{f}}= {\text{c}}_{\text{f}}{\dot{\text{E}}}_{\text{f}}={\text{c}}_{\text{F}}{\dot{\text{E}}}_{\text{F}} \left(5\right)$$

$${\dot{\text{C}}}_{\text{g}}= {\text{c}}_{\text{g}}{\dot{\text{E}}}_{\text{g}}={\text{c}}_{\text{G}}{\dot{\text{E}}}_{\text{G}} \left(6\right)$$

In Equations (2–6),${\dot{\text{C}}}_{\text{W}}$, ${\dot{\text{C}}}_{\text{Q}}$, ${\dot{\text{C}}}_{\text{F}}$and ${\dot{\text{C}}}_{\text{G}}$ arranged in order of costsper unit exergy rate of energy, rate of heat transfer, test fuels, and emissions of exhaust gases.

$$\dot{\text{Z}}={\dot{\text{Z}}}_{\text{C}\text{I}}+{\dot{\text{Z}}}_{\text{O}\text{M}} \left(7\right)$$

Where, $\dot{\text{Z}}$denotes the summation of assets investments (${\dot{\text{Z}}}_{\text{C}\text{I}}$) with operational and maintenance (${\dot{\text{Z}}}_{\text{O}\text{M}}$) expenditures. The principal investment rate is described by yearly input of principal investment cost per number of time unit function.

$\dot{\text{Z}}$ can be computed by taking the subsequent equation (Meisami 2018) .

$$\dot{\text{Z}}=\frac{\text{Z} \text{X} \text{C}\text{R}\text{F} \text{X} {\phi }}{\text{N} \text{X} 3600} \left(8\right)$$

Where Z is taken as procurement cost of test engine in terms of US dollars with ${\phi }$ as maintenance issue, indicates the deficiency of comprehensive information, assumed to be 1.06 (United States Department of Agriculture Yearbook,2016). N is regarded as yearly engine operating hours, assumed to be (12 x 100) hours per annum. So, the capital recovery factor (CRF) can be calculated by the subsequent equations:

$$\text{C}\text{R}\text{F}=\frac{{\text{i}}_{\text{e}\text{f}\text{f}}{(1+{\text{i}}_{\text{e}\text{f}\text{f}})}^{\text{n}}}{{(1+{\text{i}}_{\text{e}\text{f}\text{f}})}^{\text{n}}-1} \left(9\right)$$

Considering n as the lifetime of the ideal engine (assumed 15 years) and${\text{i}}_{\text{e}\text{f}\text{f}}$ is successful return rate, which is represented by:

$${\text{i}}_{\text{e}\text{f}\text{f}}={\left(1+ \frac{\text{i}}{\text{p}}\right)}^{\text{p}}-1 \left(10\right)$$

Wherever, i & p explain the interest rate annually and compounding rate, in that order.

Assuming the interest rate/year as 20% i.e. (i = 0.2) and compounding rate/ month (p = 12), the successful rate of return and capital return aspects were estimated and taken as 21.94% and 23.11%.

In any thermodynamic study, the test fuel representing assets to generate power (cost of losses per unit equivalent to fuel cost),

$${\dot{\text{C}}}_{\text{Q}}= {\dot{\text{C}}}_{\text{G}}={\dot{\text{C}}}_{\text{F}} \left(11\right)$$

According to thermodynamic governing equations for exergy:

${\text{C}}_{\text{F} }$ represents cost/ unit of fuel exergy (dollar/ kJ) of power production. To determine these cost values, an average international price per unit mass of fuel for a life span of 2007–2022 (15 years) was regarded as the standard price for diesel fuel and waste plastic oil. Thus, the mean prices of the fuels are 1.34 US dollars ($) and 0.47US dollars ($) per kg for diesel, and WPO, respectively (Meisami 2018).

3.1 Performance of test engine

BTE evaluates the efficient conversion of thermal power of fuels to mechanical work output. The results obtained from experiments explained the variation of thermal efficiency of the engine by changing load as depicted in Fig. 3.In addition, thermal efficiency was measured at 25.08% and 25.91% at maximum loading conditions for pure diesel and waste plastic oil respectively. Conversely, the thermal efficiency was observed at 24.83, 25.91, and 24.46% when used fuel mixtures of different concentrations of WPO, such as 10%, 20%, and 30% added with diesel respectively, under maximum load. In addition, the thermal efficiency of different WPO-diesel mixtures was observed marginally greater than diesel up to of 20% mix together. The enhancement in BTE was noticed due to longer ignition delay at maximum load resulting high volume of fuel burnt in pre-mixed stage and reduction in heat losses (Ayodhya 2018).Moreover, by rising the WPO-diesel blend ratio, the BTE was found increased due to higher calorific value of WPO (Das 2021c).

Experimental results reveal that BSFC of various WPO-diesel blends proportionately decrease with respect to diesel and were found lowest at maximum load as depicted in Fig. 4. BSFC rates are significantly influenced by the mass flow rate of fuel and decreased with the increase of load for all fuel mixtures. BSFC of diesel fluctuates from 0.70 kg/kWh to 0.34 kg/kWh at minimum load to maximum load, respectively. For other fuel mixtures such as 10% of WPO, 20% of WPO, and 30% of WPO, it differs from 0.66 to 0.35 kg/kWh, 0.62 to 0.34 kg/kWh, and 0.69 to 0.35 kg/kWh at minimum to maximum load respectively.

At the same time, BSFC is found marginally lower by substantial increments by volume of WPO in the mixture of WPO-diesel. The higher calorific value of WPO in the fuel mixture credits for the lesser consumption of fuel in comparison with diesel.

Figure5 depicts the change of EGT with several blends of WPO at different loading states. Since additional fuel is combusted to meet up the power requirement at a higher load, EGT is increased proportionately. EGT of diesel varies from 164.15°C to 229.07°C at minimum to maximum load respectively. For other fuel mixtures such as 10% WPO, 20% WPO, and 30% WPO, it varies from152.11°C to 219.37°C, 156.22°C to 224.90°C, and 158.06°C to 226.85°C at minimum to maximum load respectively. Lower EGT of WPO as compared to diesel is owing to higher thermal efficiency imparted by the availability of potential oxygen in WPO during combustion, improved fuel spray pattern as a result of lesser density, and kinematic viscosity of fuel mixture. In addition, the EGT increases with the increase of load as the engine require an extra quantity of fuel to produce extra power necessitates sustaining the additional engine loading (Kim 2002).

3.2 Emission of test Engine

Figure 6 illustrates the variation of carbon monoxide emission with various load and fuel blends. The degree of CO discharged deviates from 0.08% at 25% load to 0.058% at full load for diesel. CO emission contrasts from 0.084–0.062%, 0.083–0.06% and 0.086–0.064% for 10% WPO, 20%WPO and 30% WPO at minimum to maximum load respectively. The results illustrate that the CO emission of WPO is higher than that of diesel at maximum load and blend. The larger emission of CO at elevated loads may be caused owing to higher fuel consumption. Inconsistent fuel mixing along with insufficient oxygen content hydrocarbon in WPO results in incomplete combustion of WPO contributing to excess CO emission.

HC for diesel varies from 19 ppm to 42 ppm for minimum to maximum load, respectively. For other fuel mixtures such as 10% WPO, 20% WPO, and 30% WPO, it varies from 20 ppm to 44 ppm ,24ppm to 42ppm and 28 ppm to 46 ppm at minimum to maximum load, respectively. The lower hydrocarbon emission found at low load is due to augmentation of inherent oxygen in fuel making the mixture more homogenous. However, higher fuel consumption and deficiency of oxygen in the fuel mixture elevated the concentration of unburned HC at higher loads. Moreover, the improper pattern of fuel spray propagations and residual gases in the combustion chamber became another reason for increase in HC emission level (Celik 2005).

Figure8 exhibits the disparity in emissions of nitrogen oxide for various fuels with respect to load. It can be understood that NO_x emission increases by using a different blend of WPO in the test engine. NO_x fluctuates from 37ppm to 201 ppm at minimum to maximum load for diesel respectively. NOx varies from 64 to 212 ppm for 10% of WPO, 61 to 220 ppm for 20% of WPO, and 66 to 228 ppm for 30% of WPO with minimum and maximum load respectively. At higher load and temperatures, NO_x level was found to increase in WPO blends containing long-chain hydrocarbon compounds owing to delays in the ignition period as compared to diesel (Celik 2005).

4.1 Prediction of engine performance and emission using ANN

Model design for ANN was prepared and executed exclusively for CI engines operated by a mixture of plastic oil with diesel at different conditions. The results were obtained from an experimental study with the ANN model in Figure 2through various algorithms. The transfer functions show the use of ANN for the prediction of BSFC, BTE, EGT, and emissions like CO, HC, and NO_x which is quite satisfactory. The model design was the realization of input data values to targets data values as indicated in Figure2. The model is simulated using MATLAB 7.2.Ina model trained with the dataset, putting 80% of the data in the training set, 10% in the validation set, and 10% in the test set is a good split to start with. The optimum split of the test, validation and train set depends upon factors such as the use case, the structure of the model, the dimension of the data, etc.

A sum of 6 random data points comprising different kinds of WPO fuel combinations was presented as unrevealed test information represented in the given Figure 2. Such trials helped to decide the viability of created ANN model (expecting necessary engine parameters within the limited scopes of data collections) employed through its turn of events.

The ANN created was observed to be effectively anticipating six specific engine-out responses precisely. Figures 9–13 portray the examination of ANN predictions with estimated information for 20 data points over six designated engine out reactions. Forecasting results for BSFC are shown in Figure 9, R was estimated at 0.98984 with normal resistance ±0.002 (kg/kWh) for all the 20 test indicators. Figure 10 explains the mostexcellent choice of forecasting target valuesof thetraining algorithm meant for BSFC. Figure 11 described the prediction of R-value for BTE obtain as 0.99857 and its mean tolerance value of±0.348 through every test point. Figure 9 illustrates the general process flow of how a neural network operates. The ANN tool can be utilized as a prediction tool in an efficient manner, with a very narrow error margin. The R value ranges from 0.98209 to 0.99744, and it shows how closely the ANN predicted performance and emission outputs match experimentally measured data.

For BSFC assessment, slope descents with strength and flexible learning rate of back propagation algorithm with purelin function which was became determinant by least MSE (0.001) and greatest regression coefficient of (0.999763) as evident from Figure 9. The ideal number of neurons was assumed as 10 for minimal MSE. The best expectation scope of the target value for the presentation qualities was seen at 16 epochs. From the investigation of performance and emission characteristics by ANN, it was understood that Leven berg Marquardt (trainlm) and gradient descent with energy along with flexible learning rate back propagation (trainingdx) that performs satisfactorily due to precision in measurement to regression coefficient.

For BTE (Figure 11), slope descents with strength and flexible learning rate of back propagation algorithm with purelin function which plays a vital role produce least MSE (0.001) and greatest regression coefficient of (0.99983) which was apparent from Figure 10. The ideal quantity of neurons was16 for which minimal MSE was observed.

The EGT, Levenberg–Marquardt (trainlm), plays a significant role in conjunction with the log sigmoidal performance function, resulting in the lowest MSE (0.001) and greatest Regression coefficient (0.99989), as shown in Figure 12.

The forecasting value of CO is shown in Figure 13, R assessment was estimated at 0.97174 and actual mean tolerance of ± 0.003 % per volume for all 20 test points. In Figure14, an R-value of 0.99254 was acquired for HCcalculation with mean tolerance of ±0.001 (ppm) through every test point. Likewise, NO_xwas anticipated precisely with an R estimation of 0.99874 with normal resistance of ± 38.13 ppm/Vol. as shown in Figure 15. The ideal figure of neurons in which insignificant MSE was seen as 12. From Figure 13, the best approval for the emission qualities of CO was seen at the 19 th epoch.

The HC (ppm), Levenberg–Marquardt (trainlm) plays a major role associated with log sigmoidal performance function which produces minimal MSE (0.002) and highest Regression coefficient (0.98975) which was observable as shown in Figure 14.

The most excellent forecasting for HC emission was monitored at 15 epochs from Figure 14. The most favourable number of neurons for minimal MSE was found to be 16.

For NOx, slope descents with strength and flexible learning rate of back propagation algorithm with purelin function which plays a vital role produce least MSE (0.001) and greatest Regression coefficient of (0.99932) which was apparent from Figure 15. The ideal quantity of neurons was 16 for which minimal MSE was observed. Feed forward back propagation algorithm was used to predict the performance and emission behavior of dual-fuel engines. ANN model also was employed for engine performance and emission prediction for different set of intake mixture temperatures and load (Prabhu 2021).

4.2 Fuel economy assessment

Cost rates of inlet fuel, power, generated losses (exergy losses to cooling water and exhaust gases), and exergy destruction for various mixtures of waste plastic oil and diesel are shown in Figure 16. The entire cost rates in the fuel economic analysis have been described in US cents per second.

The lower price of WPO compared to diesel reduces the cost of fuel mixtures while there is an increment in the WPO fraction. In addition, the brake thermal efficiency is enhanced with an enhancement of the WPO fraction in the mixture. So, the combined effect of these two reasons direct to a preliminary hike in fuel price and WPO 20 shows substantially decreases in fuel cost as compared to other blends with diesel, as shown in Figure 16.It is noticeable that the lesser fuel cost attributes lowering the cost of power production. Hence, the power production cost exhibits an equivalent tendency to inlet fuel costing as viewed in Figure 16. The main terms of engine exergy losses include heat transfer to cooling systems, exhaust emission, and exergy of destruction, of which the final one is a significant determinant for fuel economy. As a result of which, the heat dissipation cost with exhaust emission cost and the cost of exergy destruction is observed comparably lower in WPO20 than that of diesel. However, further addition of WPO in diesel shows negative effect on its overall costing.

From Figure 16, it is quite noticeable that WPO 20 shows an encouraging trend of reduction in cost rate power and different losses incurred during diesel fuel combustion than other blends.

In this study of engine performance and emission using WPO blended diesel fuel, ANN was implemented to establish a better optimization engine model selecting minimum control variables to attain better output response. Better engine performance results with 20% WPO blended diesel is observed from the experimental result to that of diesel at full engine load devoid of test setup adjustment. The HC and NO_x were proportionately increased with load in comparison to diesel, while CO emission was found to decrease with the increase of loads for all fuel mixtures and slightly elevated than diesel at maximumload.ANN modeling was applied to predict the performance and emission characteristics. For the performance and emission characteristics, the MRE was within 0.002–0.348% respectively. The overall regression coefficient for the optimal network, the performance, and emission characteristics have been found to be 0.98984–0.99857, signifying a high degree (99.85%) of correlation between ANN predicted and experimentally measured data. Therefore, the proposed ANN model may be served as a successful tool for engine performance and emission prediction subjected to the optimal fuel mixture and different operating conditions. Hence, it is quite noticeable that WPO 20 shows an encouraging trend of reduction in cost rate power and different losses incurred during diesel fuel combustion than other blends.

Ethical Approval

Not applicable

-Consent to Participate

Not applicable

-Consent to Publish

Subscription

-Authors Contributions

“All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by [Amar Kumar Das], [Achyut Kumar Panda], and [Saroj Kumar Rout]. The first draft of the manuscript was written by [Amar Kumar Das] and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.”

-Funding

“The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.”

-Competing Interests

“The authors have no relevant financial or non-financial interests to disclose.”

-Availability of data and materials

Not applicable

Atik K, Kahraman N, Ceper BİLGE (2013) Prediction of performance and emission parameters of an SI engine by using artificial neural networks,Isi Bilimi Ve Teknigi Dergisi-Journal of Thermal Science and Technology, 33(2)
Ayodhya AS, Lamani VT, Bedar P, Kumar GN (2018) Effect of exhaust gas recirculation on a CRDI engine fueled with waste plastic oil blend. Fuel 227:394–400
Canakci M, Erdil A, Arcaklioğlu E (2006) Performance and exhaust emissions of a biodiesel engine. Appl Energy 83(6):594–605
Çelik V, Arcaklioğlu E (2005) Performance maps of a diesel engine. Appl Energy 81(3):247–259
Damodharan D, Sathiyagnanam AP, Rana D, Kumar BR, Saravanan S (2018) Combined influence of injection timing and EGR on combustion, performance and emissions of DI diesel engine fueled with neat waste plastic oil. Energy Conv Manag 161:294–305
Das AK, Sahu SK, Panda AK (2022) Current status and prospects of alternate liquid transportation fuels in compression ignition engines: A critical review. Renew Sustain Energy Rev 161:112358
Das AK, Hansdah D, Mohapatra AK, Panda AK (2020) Energy, exergy and emission analysis on a DI single cylinder diesel engine using pyrolytic waste plastic oil diesel blend. J Energy Inst 93(4):1624–1633
De LA, Durán A, Carmona M, Lapuerta M (2001) Modeling diesel particulate emissions with neural networks Fuel,. 80:539–5484
Dhande DY, Choudhari CS, Gaikwad DP, Dahe KB (2022) Development of Artificial Neural Network to predict performance of Spark Ignition Engine fuelled with waste pomegranate ethanol blends. Information Processing in Agriculturehttps://doi.org/10.1016/j.inpa.2022.05.001
Ghobadian B, Rahimi H, Nikbakht AM, Najafi G, Yusaf TF (2009) Diesel engine performance and exhaust emission analysis using waste cooking biodiesel fuel with an artificial neural network. Renewable Energy 34(4):976–982
Gnanamoorthi V, Purushothaman P, Gurusamy A, Devaradjane G (2021) Prediction efficiency of artificial neural network for CRDI engine output parameters. Transp Eng 3:100041
Gnanasekaran S, Saravanan N, Ilangkumaran M (2016) Influence of injection timing on performance, emission and combustion characteristics of a DI diesel engine running on fish oil biodiesel. Energy 116:1218–1229
Güngör C, Serin H, Özcanlı M, Serin S, Aydın K (2015) Engine performance and emission characteristics of plastic oil produced from waste polyethylene and its blends with diesel fuel. Int J Green Energy 12(1):98–105
Kaimal VK, Vijayabalan P (2016) A study on synthesis of energy fuel from waste plastic and assessment of its potential as an alternative fuel for diesel engines. Waste Manag 51:91–96
Kalargaris I, Tian G, Gu S (2017) Combustion, performance and emission analysis of a DI diesel engine using plastic pyrolysis oil. Fuel Process Technol 157:108–115
Khandal SV, Gadwal SB, Raikar VA, Yunus, Khan TM, Badruddin IA (2021) An experimental-based artificial neural network performance study of common rail direct injection engine run on plastic pyrolysis oil. Int J Sustain Eng 14(2):137–146
Kiani MKD, Ghobadian B, Tavakoli T, Nikbakht AM, Najafi G (2010) Application of artificial neural networks for the prediction of performance and exhaust emissions in SI engine using ethanol-gasoline blends. Energy 35(1):65–69
Kim D, Shin S, Sohn S, Choi J, Ban B (2002) Waste plastics as supplemental fuel in the blast furnace process: improving combustion efficiencies. J Hazard Mater 94(3):213–222
Kumar R, Mishra MK, Singh SK, Kumar A (2016) Experimental evaluation of waste plastic oil and its blends on a single cylinder diesel engine. J Mech Sci Technol 30(10):4781–4789
Mani M, Subash C, Nagarajan G (2009) Performance, emission and combustion characteristics of a DI diesel engine using waste plastic oil. Appl Therm Eng 29(13):2738–2744
Mani M, Nagarajan G, Sampath S (2011) Characterisation and effect of using waste plastic oil and diesel fuel blends in compression ignition engine. Energy 36(1):212–219
Meisami F, Ajam H, Tabasizadeh M (2018) Thermo-economic analysis of diesel engine fueled with blended levels of waste cooking oil biodiesel in diesel fuel. Biofuels 9(4):503–512
Pai PS, Rao BS (2011) Artificial neural network based prediction of performance and emission characteristics of a variable compression ratio CI engine using WCO as a biodiesel at different injection timings. Appl Energy 88(7):2344–2354
Prabhu AV, Alagumalai A, Jodat A (2021) Artificial neural networks to predict the performance and emission parameters of a compression ignition engine fuelled with diesel and preheated biogas–air mixture. J Therm Anal Calorim 145(4):1935–1948
Senthil kumar, Sankaranarayanan P, G (2016) Effect of Jatropha methyl ester on waste plastic oil fueled DI diesel engine. J Energy Inst 89(4):504–512
Sharma A, Sahoo PK, Tripathi RK, Meher LC (2016) Artificial neural network-based prediction of performance and emission characteristics of CI engine using polanga as a biodiesel. Int J Ambient Energy 37(6):559–570
Uslu S, Celik MB (2018) Prediction of engine emissions and performance with artificial neural networks in a single cylinder diesel engine using diethyl ether. Eng Sci Technol Int J 21(6):1194–1201
Yusaf TF, Buttsworth DR, Saleh KH, Yousif BF (2010) CNG-diesel engine performance and exhaust emission analysis with the aid of artificial neural network. Appl Energy 87(5):1661–1669

GraphicalAbstract.png
Graphical Abstract

Optimization of diesel engine performance and emission using waste plastic pyrolytic oil by ANN and its thermo economic assessment

Status:

Journal Publication

Version 1

Abstract

Figures

The Important Highlights Of The Work Are

1. Introduction

2. Experimental

2.1 Characterization of pyrolysis fuel

2.2 Engine setup and Procedure

2.3 Error and Accuracy measurements

2.4 Modeling of Neural Network (ANN)

2.5 Economic analysis of fuels used in the engine

3. Results And Discussions

3.1 Performance of test engine

3.2 Emission of test Engine

4. Ann Prediction And Thermo-economic Assessment

5. Conclusion

Statements & Declarations

References

Supplementary Files

Status:

Journal Publication

Version 1

Sl. No	Instruments	Calibration	Accuracy	Uncertainties (%)
1	Manometer	0-50mm	\(\pm 1 \text{m}\text{m}\)	± 1
2	Fuel meter	0-100 cc	± 0.01cc	± 1
3	Load indicator	0-100kg	± 0.01kg	± 0.2
4	RPM Meter	0-1000 rpm	± 10 rpm	± 0.1
5	Exhaust Gas Analyzer	CO: 0–10% CO ₂: 0–20% HC: 0–10,000ppm NO_x: 0-5000ppm	± 0.02 ± 0.03 ± 20ppm ± 10ppm	± 0.2 ± 0.15 ± 0.2 ± 0.2
6	AVL Smoke meter	0–31,999 mg/m³	± 0.02 mg/m³	± 0.5
7	Exhaust gas temperature indicator	0-1000°C	± 1ºC	± 0.15