Next Article in Journal
An Investigation of Tourism, Economic Growth, CO2 Emissions, Trade Openness and Energy Intensity Index Nexus: Evidence for the European Union
Previous Article in Journal
Treatment of Wastewater from Soil Washing with Soluble Humic Substances Using Biochars and Activated Carbon
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Energies
Volume 16
Issue 11
10.3390/en16114312
energies-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Use of Artificial Neural Networks to Model Biomass Properties of Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) in View of Harvest Season

by
Jona Šurić
1,
Neven Voća
1,*,
Anamarija Peter
1,
Nikola Bilandžija
2,
Ivan Brandić
1,
Lato Pezo
3 and
Josip Leto
4
1
Department of Agricultural Technology, Storage and Transport, Faculty of Agriculture, University of Zagreb, Svetošimunska Cesta 25, 10000 Zagreb, Croatia
2
Department of Agricultural Engineering, Faculty of Agriculture, University of Zagreb, Svetošimunska Cesta 25, 10000 Zagreb, Croatia
3
Institute of General and Physical Chemistry, University of Belgrade, Studentski trg 12/V, 11000 Belgrade, Serbia
4
Department of Field Crops, Forage and Grassland, Faculty of Agriculture, University of Zagreb, Svetošimunska Cesta 25, 10000 Zagreb, Croatia
*
Author to whom correspondence should be addressed.
Energies 2023, 16(11), 4312; https://doi.org/10.3390/en16114312
Submission received: 5 April 2023 / Revised: 2 May 2023 / Accepted: 23 May 2023 / Published: 24 May 2023
Browse Figures
Versions Notes

Abstract

:
Miscanthus and Virginia Mallow are energy crops characterized by high yields, perenniality, and low agrotechnical requirements and have great potential for solid and liquid biofuel production. Later harvest dates result in lower yields but better-quality mass for combustion, while on the other hand, when biomass is used for biogas production, harvesting in the autumn gives better results due to lower lignin content and higher moisture content. The aim of this work was to determine not only the influence of the harvest date on the energetic properties but also how accurately artificial neural networks can predict the given parameters. The yield of dry matter in the first year of experimentation for this research was on average twice as high in spring compared to autumn for Miscanthus (40 t/ha to 20 t/ha) and for Virginia Mallow (11 t/ha to 8 t/ha). Miscanthus contained 52.62% carbon in the spring, which is also the highest percentage determined in this study, while Virginia Mallow contained 51.51% carbon. For both crops studied, delaying the harvest date had a positive effect on ash content, such that the ash content of Miscanthus in the spring was about 1.5%, while in the autumn it was 2.2%. Harvest date had a significant effect on the increase of lignin in both plants, while Miscanthus also showed an increase in cellulose from 47.42% in autumn to 53.5% in spring. Artificial neural networks used to predict higher and lower heating values showed good results with lower errors when values obtained from biomass elemental composition were used as input parameters than those obtained from proximity analysis.

1. Introduction

The benefits of growing energy crops are numerous and are evident in lowering fossil fuel prices, developing rural areas, reducing global carbon emissions, and preventing soil erosion [1]. Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) are two energy crops whose characteristics correspond to the above-mentioned positive effects on the environment. Both crops have ecological potential in terms of biodiversity conservation through sequestration, i.e., carbon storage, and have a positive impact on soil quality by preventing soil erosion [2,3]. In addition, they are characterized by high yields, cold resistance, C4 photosynthesis, perenniality, long life (15–20 years), and low agrotechnical requirements. It is also important to emphasize the possibility of growing energy crops in poor-quality soils (marginal soils) [4]. Energy crops are characterized by good tolerance to nutrient and water deficiencies, without which they still have high above-ground productivity in areas unsuitable for food production [4,5]. Miscanthus giganteus, a fast-growing grass, is a sterile hybrid and reproduces vegetation through rhizome cuttings or seedlings, which prevents invasive and uncontrolled spread [6]. Along with Miscanthus, Virginia Mallow also meets the criteria of a perennial energy crop with good energy properties, but is less known in Europe [7]. Virginia mallow, classified as a perennial energy crop belonging to the mallow family (Malvaceae), is a soft, woody herbaceous crop native to North America [7,8]. The work of Mehmood et al. [9] confirms this by mentioning a maximum yield of Virginia Mallow of 9–20 t/ha on marginal soils, similar to the yield of Miscanthus (15–19 t/ha). It is known from previous research that both Miscanthus and Virginia Mallow can be harvested once a year, i.e., from the beginning of November until the beginning of the next growing season, i.e., in March and April [10]. Harvesting in late autumn or early winter achieves the maximum yield per unit area, while spring harvesting is expected to have a lower yield due to natural drying in the field [11].
Biomass, consists of a mixture of complex chemical compounds that release a certain amount of energy during the combustion process [4]. For this reason, the chemical composition as well as the cell structure of the plant are fundamental indicators of the energy value of biomass [12]. The properties of biomass are significantly affected by the phenological stage of the plant at harvest due to variability in chemical and physical composition [13]. The proportion of ash and lignin are the two most important factors, the amounts of which significantly affect the production of stable bio-oil in which the fractions do not separate [14]. The quality and composition of the bio-oil obtained by the pyrolysis process depends on the composition of the cell wall, the amount of moisture and ash, and the content of nutrients in the biomass, which are closely related to the biological development stages of the plants [15]. Different harvest dates contribute significantly to the reduction of inorganic content (ash) in the biomass, which may mean that harvesting in the fall, when ash content is higher, can negatively affect the distribution of pyrolysis products [16]. Kiesal and Lewanowski [17] found in their study that later harvests produce lower amounts of methane. The same was found in the study of Panagiotis et al. [18], who concluded that the harvest date is one of the most important parameters in determining whether a particular biomass meets the necessary characteristics for biogas production. For example, delayed harvesting until spring leads to a decrease in methane production due to an increased amount of lignin in the biomass [19]. The same thing is confirmed by a study by Oleszek et al. [20] in which Virginia Mallow shows the potential for biogas production in the first tests from the biomass obtained in the summer harvest. During growth and development, the plant takes up varying amounts of nutrients into the tissue, which can affect the fermentation process. Thus, if the C:N ratio in the substrate is not optimal, the carbon will not be fully converted to methane, resulting in a lower content in the biogas [21,22].
According to Borkowska and Styk [23], in Virginia Mallow, biomass moisture content decreased from 40% in November to 20% in January, which certainly shortens the drying process before storage and increases the percentage of dry matter, which has a positive effect on direct combustion. According to Larsen et al. [24], the yield loss due to delayed harvesting is slightly higher for Miscanthus, ranging from 35 to 40%. When biomass is used for direct burning, delayed harvesting contributes to lower ash content as the plant sheds its leaves at the end of the growing season [8]. On the other hand, the loss of leaf mass can have a positive effect on weed control because the foliage left in the field forms a protective layer on the soil (mulch) that makes it difficult for weed shoots to sprout [25]. In addition to mulching, leaving leaves on the soil surface has a positive effect on soil organic matter. At the same time, they promote the development of beneficial microflora and prevent erosion [26]. Leaves in biomass for direct combustion increase the proportion of ash, which is a mineral residue that remains after the combustion of biomass. It is alkaline in nature, which means that it lowers the melting point, affecting the thermal expansion that contributes to the formation of slag in smelting furnaces [16]. Considering the ability of ash to settle on boilers, it is desirable to have low ash levels that prevent clogging of the system [27].
Carbon, hydrogen, and oxygen are the major elemental components of biomass-derived fuel, and their relative proportions contribute significantly to the fuel’s heating value. Carbon in biomass is about 50% of the dry mass and is present in organic compounds with oxygen, which means more oxygen and volatiles compared to coal [28]. Direct combustion releases CO2, a compound that has a harmful effect on the environment. However, the advantage when CO2 is produced from biomass, compared to that released by anthropogenic impacts, is the carbon sequestration that occurs when plants absorb some of the atmospheric carbon into biomass through photosynthesis and release the other part back into the atmosphere through respiration [29]. Nitrogen and sulfur are elements that should also be present in biofuel in the smallest possible quantities. Nitrogen does not burn, nor does it participate in the production of thermal energy, while on the other hand it has a negative effect on the activity of the elements with which it combines. The most important thing is that its combustion produces undesirable NOx oxides, which have a negative impact on the environment [4].
Lignocellulose material consists of three polymers: Lignin, cellulose, and hemicellulose, which play a key role in determining optimal biomass energy [30]. If the plant contains more lignin, it is more suitable for direct combustion, i.e., the production of electricity and thermal energy [31], while a higher content of cellulose and hemicellulose favors the production of liquid biofuels [32]. The primary wall is characterized by a stretchy and thin but firm structure, which does not need to be as flexible after growth is complete. Thus, sometimes the primary wall persists without major changes, but more often a rigid secondary cell wall is formed by the deposition of new layers within the old. The most common additional polymer in the secondary wall is lignin [33]. Cellulose and hemicellulose contain more oxygen compared to lignin, resulting in a lower calorific value. Therefore, biomass with higher lignin content is more favorable for direct combustion [34]. The higher (HHV) and lower (LHV) heating values are another important parameter for determining the quality of biomass. The HHV indicates the amount of energy released during complete combustion, including gas cooling and water condensation [35]. The LHV indicates the net energy value of the biomass excluding the heat energy remaining in the condensed water vapor contained in the gases [36].
Therefore, it can be concluded that the determination of the optimal harvest date is closely related to the purpose for which the cultivated biomass is to be used. Since the exact harvesting date, whether it is biomass for direct combustion or biogas, is difficult to determine, the incorporation of modern digital solutions in the form of artificial neural networks can be one of the possible solutions to the above problem. Using artificial neural networks (ANN) as a tool to predict some energy values can certainly help to make a final decision. In the field of machine learning, ANN models have recently been increasingly used as a tool for predicting the energetic properties of biomass [37,38]. The ability to consider a large amount of data connected by nonlinear links characteristic of the input data is the main advantage of the above models [39]. In addition to the basic structure of input, hidden, and output layers, a learning algorithm is used to predict the desired output values. By comparing empirically determined and calculated values, the error of the model is determined [40].
Based on the statistical processing of the results obtained in this study and the use of an artificial neural network, the accuracy of the neural network in predicting the results was tested in comparison with the results obtained by laboratory analyses. Thus, the idea was to investigate the possibility of using ANN models to estimate the heating value of biomass based on input parameters obtained from energy characteristics and structural and elemental analyses. In this research, eight artificial neural network models are developed based on the above input parameters, four of which are used as supervised forms of learning (available input and output data) and four are user-defined predictive models to determine the capabilities of ANN in modelling energy values by comparing real and calculated data.
The main objective of this work is to determine the influence of harvest date (spring and autumn) on yield and biomass chemical properties of Miscanthus and Virginia Mallow during two growing seasons.

2. Materials and Methods

2.1. Locations of the Experimental Fields

Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) were grown in the experimental fields at the University of Zagreb Faculty of Agriculture, but in two different geographical locations. The first location is a Miscanthus field established in April 2011 by planting rhizomes in a row of 1 m on the experimental field of the Center for Grass Production on the northern slopes of the Medvednica Mountains (elevation 650 m, 45°92′71″ N 15°97′36″ S). The second location of the experimental field was at “Maksimir”, which is located directly next to the Faculty of Agriculture and is intended exclusively for teaching work and scientific research related to production technology (elevation 123 m, 45°49′48″ N 16°01′19″ E). On the mentioned field, Virginia Mallow was planted in May 2017 by planting seedlings 0.75 m apart between rows. No agricultural activities, particularly fertilization and irrigation, were conducted at either location during the growing season. Biomass samples for analysis were collected in autumn 2019, spring 2020, autumn 2020, and spring 2021, indicating that this is a mature plantation (with Miscanthus in the ninth and tenth years of cultivation and Virginia Mallow in the third and fourth years of cultivation).

2.2. Harvest Dates of Miscanthus & Virginia Mallow

In 2019, 2020, and 2021, biomass harvest was conducted in two periods (autumn and spring) at two studied sites of Miscanthus and Virginia Mallow cultivation. The autumn harvest occurs at the end of the growing season in October or November, 10–15 days after the onset of frosts. The spring harvest takes place before new shoots emerge, from March to May, depending on the climatic conditions in each study year. Table 1 shows the exact dates of Miscanthus and Virginia Mallow biomass sampling for the spring and autumn harvest periods over two years.

2.3. Meteorological Conditions during the Two Years of the Experiment

The first year of the experiment began in April 2019 and lasted until March 2020, during which time the area of the experimental site on Medvednica had an average annual temperature of 8.9 °C and 110.7 mm of recorded precipitation. The second research year lasted from April 2020 to March 2021, when very similar climatic conditions prevailed: the average air temperature was 9.1 °C, while an average of 110.4 mm of precipitation fell [41].
In contrast to the Medvednica experimental site, which is located at a much higher altitude, the Maksimir experimental site recorded much higher temperature values and less precipitation for the same period. In the first year of the experiment, the average temperature was about 13 °C and 81.8 mm of precipitation, while in the second year, the average temperature was half a degree lower and was 12.5 °C, with slightly more recorded precipitation of almost 84 mm. Figure 1 shows agroclimatic conditions for the Medvednica experimental fields and Figure 2 for the Maksimir experimental fields. Both graphs contain data with average values of daily temperatures and precipitation per month.

2.4. Determination of Yield

Biomass yield (t/ha) was measured over two years and determined by cutting plants from 10 m2 of each plot at a height of 5 cm above the ground using a chainsaw for a hedge and then weighing them on a digital hanging scale (KERN—HBC50K100). After weighing the harvested biomass, sub-samples of approximately 2 kg of chopped mass, were collected, weighed, and dried at 60 °C for 48 h [42]. After drying, the samples were reweighed and converted to dry matter yield in t/ha.

2.5. Energy Characteristics & Calorific Value

After biomass dried in a dryer at 60 °C for 48 h, samples were ground using a blender (Retsch Grindomix GM 300, Haan, Germany) and then crushed in a laboratory mill (IKA Analysentechnik GmbH, Staufen, Germany). After milling the sample, the particle distribution was determined according to a modified standard method using a sieve shaker (Retsch AS 200, Retsch GmbH, Haan, Germany). For the laboratory analyses, 150 g of a dry sample with a particle size of 250 μm to 1000 μm was used, which was then analyzed by standard methods.
The percentage of dry matter is determined by the drying process using a Memmert laboratory dryer (Memmert model 30-1060, Memmert GmbH + Co. KG, Schwabach, Germany). Approximately 1 g of the sample is weighed into a glass container, which is then brought to a constant mass in a drying oven heated to 105 °C. The sample is then placed in a desiccator, where it cools down to room temperature. By following the AOAC [43] protocol and calculating the difference before and after drying, the moisture content is obtained.
Ash content was determined by burning the biomass samples in a muffle furnace (Nabertherm Controller B170, Lilienthal, Germany) at a temperature of 550 °C to mass constancy. First, about 1 g of the sample was weighed into a porcelain pot, which was then placed in the muffle furnace. The ash content in the samples was determined by determining the difference between the mass of the sample before and after combustion according to the standard method HRN EN ISO 18122:2015 [44].
Coke content was determined by burning a biomass sample in a muffle furnace (Nabertherm, New Castle, DE, USA) at a temperature of 900 °C for 5 min. After combustion, the samples were transferred to a desiccator for cooling. The coke content was determined by determining the mass difference before and after combustion according to the standard method ISO EN HRN 15148:2009 [45]. The content of volatile substances and fixed carbon was mathematically calculated from the difference. All laboratory procedures were performed in three replicates, and data were expressed as means of dry matter. The higher heating value (HHV) was determined according to the standard method ISO HRN EN 14918:2010 [46] in an adiabatic calorimeter (IKA C200 Analysentechnik GmbH, Staufen, Germany). The adiabatic calorimeter works on the principle of temperature difference due to combustion of the sample in a bomb filled with oxygen. Next, 0.5 g to 1.0 g of the sample was weighed into a quartz vessel and the vessel was placed in a calorimetric bomb, which was then filled with oxygen under a pressure of 30 bar. The filled and sealed bomb was placed in a calorimeter filled with water. The higher heating value was determined using the IKA C200 software package, and the lower heating value (LHV) was calculated. The calorific value is expressed in MJ/kg on a dry basis.

2.6. Determination of Elemental Composition

The content of total carbon, hydrogen, nitrogen and sulfur was determined by the dry combustion method on a CHNS analyzer (Elementar Analysensysteme GmbH, Langenselbold, Germany) in accordance with the methods for carbon, hydrogen, nitrogen [47] and sulfur [48]. The procedure was carried out by burning the sample in the presence of tungsten (VI) oxide (WO3) (catalyst) in a stream of oxygen at 1100 °C. Samples were prepared by making a container from foil and weighing into it 150 mg of WO3 and 50 mg of the sample. The foil was then tightly packed and weighed, not counting the mass of the foil and the mass of the WO3. The prepared samples were placed in the instrument carousel and the analysis was started. After the analysis was completed, the results for each sample were read and the amount of oxygen was determined by calculation.

2.7. Determination of Structural Composition

The amount of cellulose, hemicellulose, and lignin [49] was determined using a fiber analyzer (ANKOM 2000 Fiber Analyzer, Macedon, NY, USA).

2.8. Statistical Analysis and Data Processing

TIBCO Statistica (13.3.0) data analysis software, was used for statistical analysis (TIBCO Software Inc, Palo Alto, CA, USA) [50]. Analysis of variance (ANOVA) and Tukey’s post hoc HSD (Honest Significant Difference) test were used to assess the differences between the means of the observable variables. The movement of the additional data (multidimensionality) in conjunction with the observed variables is revealed by principal component analysis (PCA).

2.9. Artificial Neural Network (ANN) Modeling

The data used to train the ANN model are split into 70% for training and 30% for testing. The input and output variables were identified after splitting the data. The first two ANNs were created with input data from structural analyses (% of Cellulose, Hemicellulose and Lignin) and energy characteristics (percentage of dry matter (DM), ash, coke, fixed carbon (FC) and volatile matter (VM)), while the other two networks were created with data from final analysis (percentage of C, H, N, S and O). The initial values for all models created were HHV (higher heating value) and LHV (lower heating value). To estimate HHV and LHV, all models were developed as supervised models and tailored predictive models using previously defined weighting factors and thresholds. A total of eight models were built (four supervised and four non-supervised custom prediction models). The output value (Y) of the model ANN is represented in the equation, where W1,2 stands for the weighting coefficients, f for the activation (transfer) function, B1,2 for the bias, and X for the input data of the model (Equation (1)) [51]:
Y = f 1 ( W 2 ⋅ f 2 ( W 1 ⋅ X + B 1 ) + B 2 )
To determine the optimal set of samples (input variables) in the model, Yoon’s global sensitivity method [52] was performed to determine the relative importance of the input variables on the output (Equation (2)) HHV and LHV:
R I i j ( % ) = ∑ k = 0 n ( w i k ⋅ w k j ) ∑ i = 0 m | ∑ k = 0 n ( w i k ⋅ w k j ) | ⋅ 100 %
w—weighting coefficient of ANN model, i—input variable, j—output variable, k—hidden neuron, n—number of hidden neurons, m—number of input variables.

3. Results

3.1. Biomass Productivity

From the data presented in Table 2 on the yield per unit area (t/ha) of Miscanthus and Virginia Mallow, it appears that it is significantly dependent on the time of harvest. Namely, significant differences were found between the autumn and spring harvests for both crops; for Miscanthus, a significant difference was also found between years, while for Virginia Mallow, the autumn harvest of the first year was statistically equal to the second year. The yield of Virginia Mallow in the first experimental year did not differ from the yield in the second experimental year, so the average autumn yield was 11 t/ha, and the spring yield was 7 t/ha. The lowest yield of Miscanthus was recorded in the spring of the first experimental year (20.88 t/ha), while the yield in the second experimental year was 35% higher during the same period. Comparing the yields in autumn and spring years, there is a significant increase in the second experimental year. Such a trend can be explained by the pre climatic conditions for the Miscanthus growing area shown in Figure 1. Namely, in the first year of cultivation, extremely low rainfall was recorded in the summer months from June to August, and an average of 3–4 degrees’ higher temperature than in the same period of the second year.

3.2. Energy Characteristics & Calorific Value

Energy characteristics in this study included dry matter, ash, coke, fixed carbon, and volatile matter content. Table 3 shows that for Miscanthus, dry matter content was significantly affected by harvest timing. For example, the dry matter biomass content of Miscanthus was as much as twice as high in the first research year in the spring as in the autumn.
Table 3 also shows that for Virginia Mallow, the proportion of dry matter in the first research year was almost 45% higher in spring than in autumn, while in the second research year, the highest dry matter content was recorded in spring (87.40%), which is 70% more than in autumn.
The results of this study show a significant difference in the amount of ash in relation to the autumn and spring harvest of Miscanthus, by almost 50%. Ash content in Virginia Mallow biomass was not significantly different in the first research year, averaging 2.8%, but a significantly higher ash content of 4.25% was measured in the autumn of the second year of research.
In this research, it can be seen that harvest season had a significant effect on the amount of coke and fixed carbon in both Miscanthus and Virginia Mallow. Coke content decreased in both plants studied during spring harvest compared to the autumn harvest. For example, Virginia Mallow had as much as 40% less coke in the spring of the second research year than in the autumn, while Miscanthus had only 3% less coke. The lowest coke levels found in this study were for Virginia Mallow (9.86%) and Miscanthus (11.68%). The trend of the amount of fixed carbon of the spring crop relative to the autumn harvest in Miscanthus was reversed for Miscanthus compared to coke, so that it was 9.42% in autumn and 10.14% in the spring. It is interesting to note that the fixed carbon content in the spring of the first research year and the autumn of the second year of research was not significantly different for Miscanthus, which contributed to the highest fixed carbon content in the spring of the second research year of 10.68%. The delay of harvest date had a significant effect with an increase in volatile substances, so that Virginia Mallow had as high as 85.35% of volatile substances in spring, which is also the highest result in this study.
The average LHV of Miscanthus in this study was 17.52 MJ/kg, and that of Virginia Mallow was 16.31 MJ/kg. Considering the results obtained in the first experimental year of the study, where LHV is statistically the same in both harvests, while in the second experimental year there are differences in autumn and spring, it cannot be concluded with certainty that harvest time is a key factor for the reported values. HHV did not differ significantly from harvest date only in the second experimental year of Miscanthus cultivation, averaging 18.6 MJ/kg, whereas in the first experimental year for the same crop, the delay in harvest date contributed to an increase in HHV from 17.73 MJ/kg in the autumn to 17.83 MJ/kg in the spring.
Looking at the values obtained for Virginia Mallow for the same parameter, a negative trend is observed, which means that in this case, the delay in harvest contributed to a decrease in the higher heating value.

3.3. Elemental Composition

From Table 4, it can be concluded that the season of harvest had a significant effect on the number of elements present in the biomass of Miscanthus and Virginia Mallow. For nitrogen, for example, lower levels were measured in the spring than in the autumn for both crops observed. For example, Miscanthus had the lowest nitrogen content (0.08%) in the spring of the first experimental year, while Virginia Mallow had three times less nitrogen (0.19%) in the spring than in the autumn (0.59%) in the same year. In another experimental year, the values in spring are also lower than in autumn, but the difference is smaller. As for carbon, its percentage increased with the delay of the harvest date, so that in the case of Miscanthus, the highest value was 52.62% in the spring of the first experimental year. In the case of Virginia Mallow, an increase from 49.42% in the autumn to 51.51% in the spring was also observed.
Sulfur and oxygen are the only elements that were not significantly affected by harvest date in the second experimental year when we consider only Miscanthus. Thus, the average percentage of sulfur was 0.19% and that of oxygen was 43%.
Higher hydrogen levels were found in both crops in the first experimental year of the study. Delaying harvest until spring in Virginia Mallow increased the hydrogen content from 6.14% in the autumn to 6.19% in the spring. In contrast to hydrogen, oxygen levels were higher in both Miscanthus and Virginia Mallow in the second experimental year of this study. A significant increase in oxygen was observed in both Miscanthus and Virginia Mallow with respect to the year of cultivation.
Principal component analysis (PCA) was carried out in order to present the share of an individual element with regard to the harvest date as vividly as possible. The analysis was successfully used to classify and separate different samples in factor space (based on the proportion of each element). The detection of similarities in the studied parameters is based on experimental values and classified according to their results in the factor space of the coordinate system. The results of the elemental composition of Miscanthus (M) and Virginia Mallow (VM), i.e., the measured parameters (descriptors), are shown graphically in Figure 3 using the PCA. The graph shows the successful separation of 8 samples ((M1)-Miscanthus1-2019(Autumn), (M2)-Misanthus2-2020(Spring), (M3)-Misanthus3-2020(Autumn), (M4)-Miscanthus4-2021(Spring), (VM1)-Virginia Mallow1-2019(Autumn), (VM2)-Virginia Mallow2-2020(Spring), (VM3)-Virginia Mallow3-2020(Autumn), and (VM4)-Virginia Mallow4-2021(Spring)). The most positive influence on the PC1 coordinate was observed for O (based on correlations, 41.91%) and S content (8.13%), while the most negative influence was observed for H (38.84%) and C content (10.80%). The most positive influence on the PC2 coordinate was observed for N content (38.14%), while the most pronounced negative impact on the PC2 coordinate calculation was observed for C (36.84%) and S content (22.61%). The most pronounced differences between samples were observed according the PC1 coordinate, which showed that VM3 (44.03%), VM4 (45.23%), M3 (43%), and M4 (42.7%) were characterized with increased O content, while samples VM1 (6.14%), VM2 (6.19%), M1 (6.07%), and M2 (6.12%) were characterized by increased H content. The most pronounced N content was observed for VM1 (0.59%) and VM3 (0.39%) samples, while the most C and S content were noticed for M2 (52.62%; 0.1%) and M4 (51.6%; 0.17%) samples.

3.4. Structural Composition

Shifting the harvest period resulted in a 12% increase in cellulose content in Miscanthus biomass in the spring compared to the autumn. In the same crop, the increase in cellulose was also recorded in the second year of research from 52 to 53%. On the other hand, the later harvest of Virginia Mallow resulted in a lower amount of cellulose in the first research year, while in the spring of the second research year, it had the highest cellulose content recorded in this study at almost 57% (Table 5).
Regarding hemicellulose, Miscanthus contained a lower amount of hemicellulose in the spring than in the autumn in both experimental years, while in Virginia Mallow, the delayed harvest resulted in a 35% increase in hemicellulose in the spring compared to autumn.
Lignin is the only one of the three parameters observed when it comes to cell wall structure that showed an increase in both Miscanthus and Virginia Mallow with a later harvest. Values for lignin in the spring were 10 to 15% higher compared to the autumn, and the highest lignin content was observed in Virginia Mallow in the first experimental year in the spring (16.8%), while the lowest value was recorded in Miscanthus in autumn, also in the first year of research (13.08%).

3.5. Artificial Neural Networks Modeling

Table 6 and Table 7 show the weight coefficients and biases of the input and output layers for the developed model ANN to estimate LHV and HHV with respect to characteristic analyses; these weight coefficients and biases determine the strength and polarity of the connections between nodes and the activation thresholds, which in turn determine the learning ability and predictive performance of the network.
Table 8 shows the performance of the developed ANN-MLP models in relation to the characteristic input parameters.

4. Discussion

4.1. Biomass Productivity

From the results of this study, it can be concluded that harvest date significantly affects biomass yield.
The yield of dry matter and, accordingly, the moisture content depend to a great extent on the harvest date, so with this knowledge, it is necessary to determine for what purpose the biomass produced is to be used.
When growing energy crops, it is very challenging to determine the best time to harvest and ensure that the biomass remains suitable for biofuel production. Accordingly, ref. [17] found that for direct combustion, it is important to pay attention to yield loss and maximum utilization of nutrients. It was found that yield is significantly lower in spring than in autumn, which can be explained by biomass losses due to leaf fall in winter, which reduces the amount of biomass [8].
Yield losses at later harvest were found in the study by Larsen et al. [24], who found a lower yield of about 40% for Miscanthus.
In the study by Jablonowski et al. [53], significantly higher yields of Virginia Mallow were recorded, averaging 25 t/ha, but delayed harvesting until spring resulted in lower values, so yield was then 15 t/ha. And according to Meehan et al. [54], yield of Miscanthus drops from January to April, so the winter harvest dropped from 24 t/ha to 12 t/ha in the spring. Similar results were obtained by Godin et al. [12], where the biomass yield of Miscanthus was 14 t/ha in the autumn and 9 t/ha in the spring. It is interesting to note that the yields of Miscanthus in both studies are significantly lower than in this study.
Thus, we can conclude that yield can be significantly affected by climatic conditions, i.e., extreme temperatures or precipitation, and the age of the plant, which should not be ignored [55,56]. Climatic conditions are confirmed by the results of the study by Clifton-Brown et al. [57], which found that postponing harvest to spring resulted in an average 50% yield reduction, although yields varied greatly by country. For example, yields in Sweden or Finland were 13 t/ha in the fall and 9 t/ha in the spring, while countries with warmer climates had much higher yields of over 20 t/ha in the fall and about 16 t/ha in the spring.

4.2. Energy Characteristics & Calorific Value

Harvesting in the spring results in lower moisture content, which is a result of natural drying in the field, and ultimately lower yield [58]. The reduction in moisture content increases the dry matter content, which leads to an increase in biomass quality when the biomass is used for direct combustion and significantly reduces the investment in the processes of drying, storage, and transportation [9]. Authors Oleszek et al. [20] believe that the high dry matter content of Virginia Mallow stems is of exceptional value, creating the potential for direct burning immediately after harvest. The same authors conclude that significant savings can be realized from biomass with relatively low moisture during harvest through a simplified harvesting and shredding process in the form of long-term storage without the risk of spoilage.
Dry matter content is an important factor determining biomass potential for energy use (net energy potential), calorific value, combustion efficiency, and combustion temperature [59]. By shifting the harvest from autumn to spring, there is a reduction in moisture due to natural drying in the field, which is desirable in the case of burst combustion because the high water content significantly hinders carbon combustion [60]. Stolarski et al. [61] also observed significant water loss of about 40% in Miscanthus and about 30% in Virginia Mallow from November to April. Another study conducted at Virginia Mallow found the highest percentage of dry matter at the end of the growing season, i.e., during the biomass harvest in January, when more than 70% of the dry matter was recorded. The authors linked the increase in dry matter to the loss of leaves and the relative increase in lignin and cellulose in the barn [53].
According to Larsen et al. [24] recorded a slightly smaller loss, on average, from 41.2% of dry matter in the autumn harvest to 73.9% of dry matter in the spring harvest. The amount of dry matter in a plant is related to the agro-ecological conditions of the environment in which it grows, and is primarily related to rainfall and temperature [62].
Postponing harvest has a significant effect on the proportion of ash in the biomass. The decrease in the ash fraction in the biomass harvested in spring is due to the loss of leaf mass that occurs towards the end of the growing season, and since leaves have a high content of minerals and ash, they are certainly not a desirable segment in direct combustion [63]. The same was confirmed by Nazli et al. [64] who found that the biomass contains a lower percentage of ash during the winter months, when the plant sheds its leaves, which favors the process of direct combustion. In the study by Šiaudinis et al. [65], the average ash content in autumn was 6%, which was higher than the highest value recorded in this study.
Coke and fixed carbon are desirable characteristics of biomass as they represent the amount of energy released during combustion [66].
Another important parameter determined in the energy analysis is the volatile matter, which consists of condensing water vapor and other gases that are released when biomass is exposed to high temperatures, leaving solid carbon behind [66]. The number of volatile substances in the biomass affects the reactivity of the fuel. According to Cavalaglio et al. [67], an increase in the amount of volatile matter in biomass leads to an increase in adiabatic combustion temperature, i.e., faster complete combustion. On the other hand, volatiles in biogas production digesters are an important parameter for evaluating the rate of anaerobic digestion [68]. Krička et al. [31] recorded a similar result for the volatile matter content of Virginia Mallow of 84.90%. In the same study, the volatile matter content for Miscanthus was 88.76%, which is a slightly higher result than that obtained in this study. However, in the study by Babich et al. [69], the volatile matter content in the same crop ranged from 73.6 to 73.9%, which is much lower than the value obtained in this study.
Meehan et al. [54] investigated the effect of shifting the harvest date from winter to spring on the LHV of Miscanthus and found a positive trend. For example, the LHV was about 10 MJ/kg in January and about 12 MJ/kg in March, which is certainly a positive trend in the case of direct burning. Authors Stolarski et al. [70] studied Virginia Mallow cultivation and measured the HHV and LHV of biomass harvested in March. The obtained results showed slightly higher values for both parameters (HHV = 19.2 MJ/kg, HHV = 14.9 MJ/kg). In other literature studies, similar values for HHV were found in the range of 16–17 MJ/kg, which was also the case in this research [31,71].

4.3. Elemental Composition

The nitrogen content of the fuel should be as low as possible to reduce combustion costs. According to the standard CENTS 14961:2005 [72], the maximum nitrogen content in biomass fuels is 0.7%. In other studies, conducted on Virginia Mallow, the average nitrogen content was between 0.2 and 0.5% [73,74], while in the work of Pokój et al. [75], a much higher percentage was determined, where the nitrogen content was almost 2%. The timing of harvest reduced the nitrogen content in Miscanthus, so in this study, it was only 0.08% in spring. Meehan et al. [54] monitored nitrogen percentage from January to March and recorded an increase from month to month, so that the nitrogen percentage in spring was 0.31%, which was found in September in our study.
Both carbon and hydrogen oxidize to form H2O and CO2. The presence of both elements contributes positively to the heating value of biomass [76]. According to the literature, the percentage of carbon in Virginia Mallow varies between 41 and 44% [55,73,75] and in Miscanthus around 50% [52], which corresponds to direct combustion of a satisfactory value. As for hydrogen, Šiaudinis et al. [65] found 5.61% in the biomass of Virginia Mallow harvested in autumn, while Stolarski et al. [70] found 5.38% hydrogen in the same crop but in spring. In this study, hydrogen was also found to decrease due to the delay in harvest date, in contrast to carbon, which leads to an increase in the aforementioned element when harvested later. Oxygen is an undesirable fuel component. It does not burn, but only participates in combustion, which means that oxygenated compounds are not combustible and have a lower calorific value. Both [65] and [77] determined oxygen levels of 41.94% and 49.86%, respectively, for Virginia Mallow in the summer, which is also consistent with the values determined in this study, but only in the second year. In the first year, both Miscanthus and Virginia Mallow had an oxygen content of only around 25%, which was certainly due to the high temperatures and low rainfall during the summer months. Sulfur, in combination with oxygen, is involved in the formation of sulfur oxides SOx, which can be very corrosive and cause problems in boilers [76]. The same authors state that up to 40–70% of the sulfur from biomass is contained in the ash. It is also important to note that the amount of sulfur in biomass is extremely low (less than 0.2%), so the amount of harmful sulfur oxides is almost negligible. This is confirmed by the literature, which indicates that the sulfur content in Virginia Mallow averages between 0.02 and 0.05% [27,55] and in Miscanthus 0.1 [78].

4.4. Structural Composition

From the results of this study, it can be concluded that the composition of lignocellulose is highly dependent on the time of harvest. The capacity of biomass for methane production is strongly influenced by the content of lignin, hemicellulose, and cellulose, i.e., the amount of carbohydrates, proteins, and fats. The richer the biomass in fibers, especially lignin, the worse the methane production [79]. The same conclusion is reached by Fernandes et al. [80], who indicate that Miscanthus should be harvested for use in biogas plants while it is still green, i.e., in the fall, certainly before winter, since delayed harvesting increases lignin content and prolongs fermentation time. In contrast, biogas production is maximized when the biomass contains adequate amounts of sugar, protein, and fat. Therefore, the highest yields are usually obtained by harvesting during the summer months [73]. As a given plant grows and develops, i.e., ages, the lignin content increases, causing the protein content to decrease [79]. Bilandžija et al. [8] results for Virginia Mallow showed a lignin content of 19.88% in autumn and 25.45% in spring. The authors of Cerazy-Waliszewska et al. [81] note that spring biomass of Miscanthus increases the content of lignin, hemicellulose, and, especially, cellulose. Zanetti et al. [82] found that hemicellulose content in Miscanthus decreased with increasing harvest delay, which was due to an increase in cellulose content and lignin deposition. The same result for Miscanthus was found in the study by Managold et al. [83], where hemicellulose was lower at the later harvest. In the study by Bilandžija et al. [8], it was found that there was no difference in the hemicellulose content of Virginia Mallow between the autumn and spring harvests. Comparing the variability of the literature results with the differences in this study, it can be interpreted that in addition to harvest date, climatic conditions and soil type influence the composition of lignocellulose biomass [84]

4.5. Artificial Neural Network Model

The first model developed to predict HHV and LHV was based on the input values from the structural composition and energy characteristics (Cel, Hemcel, Lig, DM, Ash, Coke, FC, and VM). The model was developed with an 8-9-2 structure (the numbers indicate the number of artificial neurons in the input, hidden and output layers). To determine the ability of the model in training and testing, as well as the errors of the model, the coefficient of determination (R2) was calculated for the above parameters, which was 0.62 for training and 0.44 for testing, while the error of the model in training was 0.08 and 0.07 in testing. The second model developed was based on the input values of C, H, N, S, and O and had a 5-3-2 structure. The model showed higher performance in predicting the significant values of HHV and LHV, as evidenced by a high coefficient of determination for testing and training (R2 = 1.00) and a low level of error (0.00).
Table 9 shows the calculated statistical parameters that provide information about the capabilities of the ANN models in estimating the energy values (HHV and LHV) of Miscanthus and Virginia Mallow with respect to specific input parameters in the model.
Table 9 shows the statistical performance test of the developed ANN models for estimating HHV and LHV based on the input data of specific analyses (structural composition, energy characteristics and elemental composition). The first two models of ANN (ANN 1 and 2) were developed as supervised forms, while ANN 3 and 4 were developed as custom prediction models where “new” data were implemented into the system. ANN 1 and 2 have a relatively low level of error based on the parameters X2 (0.02), RMSE (0.15 and 0.14), MBE (0.01 and −0.01), while the coefficient of determination determines the relationship to be 0.58 and 0.59, respectively. Models 3 and 4 (user-defined prediction) had slightly higher error parameters but a higher coefficient of determination (0.76 and 0.60). ANN Models 5 and 6 were also developed as supervised forms of the model using the input data from the ultimate analysis. For both models, the statistical error parameters X2, RMSE, MBE, MPE, and SSE were 0.00, and the coefficient of determination showed a high correlation in modelling the energetic properties, where R2 was 1.00. The developed ANN models 7 and 8 were used as custom predictive models. Their error level was higher and the coefficient of determination was 0.78 and 0.62.
In order to determine the influence of input variables on the output values predicted by the model, a global sensitivity analysis can help identify the most influential input factors and their interactions, providing insights into the model’s behavior and leading to design changes (Figure 4).
Figure 4 shows the relative importance of the input variables on the output values of HHV and LHV. In the models based on the input data from the structural composition and energy characteristics (Figure 4a,b), the increase in energetic properties was affected by the variables in the following percentages: Cel (13.21%), Hem (−26.33%), Lig (−18.16%), DM (−6.54%), ASH (14.44%), Coke (5.17%), FC (0.09%) and VM (−16.06%). In the ANN models presented in (c) and (d), the change in HHV and LHV was influenced by the decrease in N (−0.01%), C (−41.17%), S (−0.78%), H (−2.86%), and O (−55.18%).

5. Conclusions

Harvest timing significantly affected the energy properties of Miscanthus and Virginia Mallow. Shifting harvest to spring compared to autumn resulted in significant differences in the energy parameters studied. Regarding the negative impact of nitrogen oxides on the environment and the atmosphere, this study showed a significant decrease in nitrogen content in both observed crops by delaying the harvest, which is certainly a positive influence. In addition to nitrogen reduction, spring harvesting resulted in higher levels of fixed carbon, volatiles, and even a 2-fold increase in dry matter, but also lower levels of ash, coke, nitrogen, and sulfur. The increases obtained in the spring harvest, together with the reduction of nitrogen content, contribute positively to the characteristics of biomass for direct combustion with less negative impact on the environment.
On the other hand, in the second year of research, Miscanthus showed no statistically significant difference by delaying the harvest date, except for the higher heating value, oxygen content, and hemicellulose content. In the same research year, Virginia Mallow showed no significant difference, except in cellulose content. It is interesting to note that in Virginia Mallow, several statistically equal parameters were determined in the first year (lower heating value, ash content, fixed carbon content, and sulfur content), while in Miscanthus, only the lower heating value remained the same regardless of harvest date. Given the lignocellulose composition and the increase in cellulose content at spring harvest, Virginia Mallow and Miscanthus represent potential for liquid biofuel production. Higher levels of direct combustion and lignin content are also desirable and were found in the spring compared to the autumn. Although Virginia Mallow had the highest lignin content at 16.8%, area yields should be considered before final selection of an energy crop for biofuel production of any kind. At the same time, increasing the lignin content in biogas production is not desirable because of the difficulty of methane formation.
Using the developed ANN models showed promising results for the estimation of HHV and LHV. The model based on elemental analysis data (ANN 2) outperformed the models based on the structural composition and energy characteristics data. The high R2 value of 1.00 for ANN 2 indicates a strong correlation between the input variables and the output values. These models can be used as reliable tools to predict HHV and LHV, which are crucial parameters in energy-related industries.
The results obtained for Miscanthus and Virginia Mallow illustrate the broad possibility of using perennial energy crops for the production of “green” renewable energy. It is important to clarify before harvesting for what purposes the harvested biomass will be used, because delaying the harvest date will have a positive effect on biomass used for energy production through direct combustion, while harvesting in the fall will result in biomass that will be more suitable for biogas production in terms of energy composition.

Author Contributions

Conceptualization, J.Š., N.B., A.P. and I.B.; methodology, A.P., L.P. and J.L.; validation, J.Š., A.P., N.V. and N.B.; formal analysis, L.P., I.B. and A.P.; writing—original draft preparation, J.Š. and N.B.; writing—review and editing, L.P., N.V., A.P, J.L. and I.B.; visualization, I.B. and J.L.; supervision, L.P., N.V. and N.B.; project administration, N.V. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Croatian Science Foundation (HRZZ) under project No. IP-2018-01-7472, “Sludge management via energy crops production”.

Data Availability Statement

Data generated during the study can be obtained by the authors of this study.

Acknowledgments

This research was cofounded by the Croatian Science Foundation (HRZZ) within the project “Young Researchers’ Career Development Project—Training of Doctoral Students” (DOK-2021-02), co-financed by the European Union, under the OP “Efficient Human Resources 2014–2020” from the ESF funds.

Conflicts of Interest

The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analysis, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

References

  1. Amjith, L.R.; Bavanish, B. A review on biomass and wind as renewable energy for sustainable environment. Chemosphere 2022, 293, 133579. [Google Scholar] [CrossRef] [PubMed]
  2. Popp, J.; Lakner, Z.; Harangi-Rákos, M.; Fári, M. The effect of bioenergy expansion: Food, energy, and environment. Renew. Sustain. Energy Rev. 2014, 32, 559–578. [Google Scholar] [CrossRef]
  3. Bilandžija, D.; Stuparić, R.; Galić, M.; Zgorelec, Ž.; Leto, J.; Bilandžija, N. Carbon Balance of Miscanthus Biomass from Rhizomes and Seedlings. Agronomy 2022, 12, 1426. [Google Scholar] [CrossRef]
  4. Voća, N.; Leto, J.; Karažija, T.; Bilandžija, N.; Peter, A.; Kutnjak, H.; Šurić, J.; Poljak, M. Energy Properties and Biomass Yield of Miscanthus × giganteus Fertilized by Municipal Sewage Sludge. Molecules 2021, 26, 4371. [Google Scholar] [CrossRef] [PubMed]
  5. Jezowski, S.; Mos, M.; Buckby, S.; Cerazy-Waliszewska, J.; Owczarzak, W.; Mocek, A.; McCalmont, J.P. Establishment, growth and yield potential of the perennial grass Miscanthus × giganteus on degraded coal mine soils. Front. Plant Sci. 2017, 8, 726. [Google Scholar] [CrossRef]
  6. Pidlisnyuk, V.; Mamirova, A.; Pranaw, K.; Shapoval, P.Y.; Trögl, J.; Nurzhanova, A. Potential role of plant growth-promoting bacteria in Miscanthus × giganteus phytotechnology applied to the trace elements contaminated soils. Int. Biodeterior. Biodegrad. 2020, 155, 105103. [Google Scholar] [CrossRef]
  7. Siwek, H.; Włodarczyk, M.; Mozdzer, E.; Bury, M.; Kitczak, T. Chemical Composition and Biogas Formation potential of Sida hermaphrodita and Silphium perfoliatum. Appl. Sci. 2019, 9, 4016. [Google Scholar] [CrossRef]
  8. Bilandžija, N.; Krička, T.; Matin, A.; Leto, J.; Grubor, M. Effect of Harvest Season on the Fuel Properties of Sida hermaphrodita (L.) Rusby Biomass as Solid Biofuel. Energies 2018, 11, 3398. [Google Scholar] [CrossRef]
  9. Mehmood, M.A.; Ibrahim, M.; Rashid, U.; Nawaz, M.; Ali, S.; Hussain, A.; Gull, M. Biomass production for bioenergy using marginal lands. Sustain. Prod. Consum. 2017, 9, 3–21. [Google Scholar] [CrossRef]
  10. Bilandžija, N.; Jurišić, V.; Voća, N.; Leto, J.; Matin, A.; Sito, S.; Krička, T. Combustion properties of Miscanthus × giganteus biomass—Optimization of harvest time. J. Energy Inst. 2017, 90, 528–533. [Google Scholar] [CrossRef]
  11. Zub, H.W.; Arnoult, S.; Brancourt-Hulmel, M. Key traits for biomass production identified in different Miscanthus species at two harvest dates. Biomass Bioenergy 2011, 35, 637–651. [Google Scholar] [CrossRef]
  12. Godin, B.; Lamaudière, S.; Agneessens, R.; Schmit, T.; Goffart, J.P.; Stilmant, D.; Gerin, P.A.; Delcarte, J. Chemical characteristics and biofuels potentials of various plant biomasses: Influence of the harvesting date. J. Sci. Food Agric. 2013, 93, 3216–3224. [Google Scholar] [CrossRef] [PubMed]
  13. Meserszmit, M.; Chrabąszcz, M.; Chylińska, M.; Szymańska-Chargot, M.; Trojanowska-Olichwer, A.; Kącki, Z. The effect of harvest date and the chemical characteristics of biomass from Molinia meadows on methane yield. Biomass Bioenergy 2019, 130, 105391. [Google Scholar] [CrossRef]
  14. Fahmi, R.; Bridgwater, A.V.; Donnison, I.; Yates, N.; Jones, J.M. The effect of lignin and inorganic species in biomass on pyrolysis oil yields, quality and stability. Fuel 2008, 87, 1230–1240. [Google Scholar] [CrossRef]
  15. Hodgson, E.M.; Nowakowski, D.J.; Shield, I.; Riche, A.; Bridgwater, A.V.; Clifton-Brown, J.C.; Donnison, I.S. Variation in Miscanthus chemical composition and implications for conversion by pyrolysis and thermo-chemical bio-refining for fuels and chemicals. Bioresour. Technol. 2011, 102, 3411–3418. [Google Scholar] [CrossRef] [PubMed]
  16. Hodgson, E.M.; Fahmi, R.; Yates, N.; Barraclough, T.; Shield, I.; Allison, G.; Bridgwater, A.V.; Donnison, I.S. Miscanthus as a feedstock for fast-pyrolysis: Does agronomic treatment affect quality? Bioresour. Technol. 2010, 101, 6185–6191. [Google Scholar] [CrossRef]
  17. Kiesel, A.; Lewandowski, I. Miscanthus as biogas substrate. In Proceedings of the Conference Paper on the 23rd European Biomass Conference and Exhibition, Viena, Austria, 1–4 June 2014. [Google Scholar]
  18. Panagiotis, T.; Panagiotis, G.K.; Angelidaki, I. Anaerobic mono—And Co-digestion of mechanically pretreated meadow grass for biogas production. Energy Fuel. 2015, 29, 4005–4010. [Google Scholar]
  19. Prochnow, A.; Heiermann, M.; Plöchl, M.; Linke, B.; Idler, C.; Amon, T.; Hobbs, P.J. Bioenergy from permanent grassland—A review: 1. Biogas. Bioresour. Technol. 2009, 100, 4931–4944. [Google Scholar] [CrossRef]
  20. Oleszek, M.; Matyka, M.; Lalak, J.; Tys, J.; Paprota, E. Characterization of Sida hermaphrodita as a feedstock for anaerobic digestion process. J. Food Agric. Environ. 2013, 11, 1839–1841. [Google Scholar]
  21. Weizhang, Z.; Zhongzhi, Z.; Yijing, L.; Wei, Q.; Meng, X.; Min, Z. Biogas productivity by co-digesting Taihu blue algae with corn straw as an external carbon source. Bioresour. Technol. 2012, 114, 281–286. [Google Scholar]
  22. Wang, X.J.; Yang, G.H.; Feng, Y.Z.; Ren, G.X.; Han, X.H. Optimizing feeding composition and carbon–nitrogen ratios for improved methane yield during anaerobic co-digestion of dairy, chicken manure and wheat straw. Bioresour. Technol. 2012, 120, 78–83. [Google Scholar] [CrossRef] [PubMed]
  23. Borkowska, H.; Styk, B. Virginia Fanpetals (Sida hermaphrodita L. Rusby): Cultivation and Utilization Monograph; University of Life Sciences: Lublin, Poland, 2006; Volume 69. [Google Scholar]
  24. Larsen, S.U.; Jørgensen, U.; Kjeldsen, J.B.; Lærke, P.E. Long-term Miscanthus Yields Influenced by Location, Genotype, Row Distance, Fertilization and Harvest Season. BioEnergy Res. 2014, 7, 620–635. [Google Scholar] [CrossRef]
  25. Christian, D.G.; Haase, E.; Schwarz, H.; Dalianis, C.; Clifton-Brown, J.C.; Cosentino, S. Agronomy of Miscanthus. In Miscanthus for Energy and Fibre; Jones, M.B., Walsh, M., Eds.; James & James (Science Publishers) Ltd.: London, UK, 2001; pp. 21–45. [Google Scholar]
  26. El-Beltagi, H.S.; Basit, A.; Mohamed, H.I.; Ali, I.; Ullah, S.; Kamel, E.A.R.; Shalaby, T.A.; Ramadan, K.M.A.; Alkhateeb, A.A.; Ghazzawy, H.S. Mulching as a Sustainable Water and Soil Saving Practice in Agriculture: A Review. Agronomy 2022, 12, 1881. [Google Scholar] [CrossRef]
  27. Von Gehren, P.; Gansberger, M.; Pichler, W.; Weigl, M.; Feldmeier, S.; Wopienka, E.; Bochmann, G. A practical field trial to assess the potential of Sida hermaphrodita as a versatile, perennial bioenergy crop for Central Europe. Biomass Bioenergy 2018, 122, 99–108. [Google Scholar] [CrossRef]
  28. Jenkins, B.M.; Baxter, L.L.; Miles, T.R. Combustion properties of biomass. Fuel Proc. Technol. 1998, 54, 17–46. [Google Scholar] [CrossRef]
  29. Patil, P.; Kumar, A.K. Biological Carbon Sequestration Through Fruit Crops (Perennial crops–natural “sponges” for absorbing carbon dioxide from atmosphere). Plant. Arch. 2017, 17, 1041–1046. [Google Scholar]
  30. Bajpai, P. Structure of Lignocellulosic Biomass. In Pretreatment of Lignocellulosic Biomass for Biofuel Production; Springer: Singapore, 2016; pp. 7–12. [Google Scholar]
  31. Krička, T.; Matin, A.; Bilandžija, N.; Jurišić, V.; Antonović, A.; Voća, N.; Grubor, M. Biomass valorisation of Arundo donax L., Miscanthus × giganteus and Sida hermaphrodita for biofuel production. Int. Agrophys. 2017, 31, 575–581. [Google Scholar] [CrossRef]
  32. Jönsson, L.J.; Alriksson, B.; Nilvebrant, N.O. Bioconversion of lignocellulose: Inhibitors and detoxification. Biotechnol. Biofuels 2013, 6, 16. [Google Scholar] [CrossRef]
  33. Alberts, B.; Johnson, A.; Lewis, J.; Raff, M.; Roberts, K.; Walter, P. Molecular Biology of the Cell, 4th ed.; Garland Science: New York, NY, USA, 2002; The Plant Cell Wall. Available online: https://www.ncbi.nlm.nih.gov/books/NBK26928 (accessed on 22 May 2023).
  34. Lewandowski, I.; Clifton-Brown, J.C.; Andersson, B.; Basch, G.; Christian, D.G.; Jorgensen, U.; Jones, M.B.; Riche, A.B.; Schwarz, K.U.; Tayebi, K. Enviroment and harvest time affect the combustion qualities of Miskantus genotypes. Agron. J. 2003, 95, 1274–1280. [Google Scholar] [CrossRef]
  35. Demirbas, A. Relationships between Heating Value and Lignin, Moisture, Ash and Extractive Contents of Biomass Fuels. Energy Explor. Exploit. 2002, 20, 105–111. [Google Scholar] [CrossRef]
  36. Howaniec, N.; Smoliński, A. Steam gasification of energy crops of high cultivation potential in Poland to hydrogen-rich gas. Int. J. Hydrog. Energy 2011, 36, 2038–2043. [Google Scholar] [CrossRef]
  37. Giwa, S.O.; Adekomaya, S.O.; Adama, K.O.; Mukaila, M.O. Prediction of selected biodiesel fuel properties using artificial neural network. Front. Energy 2015, 9, 433–445. [Google Scholar] [CrossRef]
  38. Dashti, A.; Noushabadi, A.S.; Raji, M.; Razmi, A.; Ceylan, S.; Mohammadi, A.H. Estimation of biomass higher heating value (HHV) based on the proximate analysis: Smart modeling and correlation. Fuel 2019, 257, 115931. [Google Scholar] [CrossRef]
  39. Pattanayak, S.; Loha, C.; Hauchhum, L.; Sailo, L. Application of MLP-ANN models for estimating the higher heating value of bamboo biomass. Biomass Convers. Biorefin. 2021, 11, 2499–2508. [Google Scholar] [CrossRef]
  40. Kartal, F.; Özveren, U. A deep learning approach for prediction of syngas lower heating value from CFB gasifier in Aspen plus. Energy 2020, 209, 118457. [Google Scholar] [CrossRef]
  41. Croatian Meteorological and Hydrological Service. Available online: https://meteo.hr/index_en.php (accessed on 1 November 2022).
  42. ISO 18134-2:2017; Solid Biofuels—Determination of Moisture Content—Oven Dry Method—Part 2: Total Moisture—Simplified Method. ISO: Geneva, Switzerland, 2017.
  43. AOAC. Official Methods of Analysis, 16th ed.; Association of Official Analytical Chemists: Washington, DC, USA, 1995. [Google Scholar]
  44. EN ISO 18122:2015; Solid Biofuels—Determination of Ash Content. ISO: Geneva, Switzerland, 2015.
  45. CEN/TS 15148:2009; Solid Biofuels—Determination of the Content of Volatile Matter. CEN: Brussels, Belgium, 2009.
  46. CEN/TS 14918:2005; Solid Biofuels—Method for the Determination of Calorific Value. CEN: Brussels, Belgium, 2005.
  47. EN 15104:2011; Solid Biofuels—Determination of Total Content of Carbon, Hydrogen and Nitrogen—Instrumental Methods. SIS: Stockholm, Sweden, 2011.
  48. EN 15289:2011; Solid Biofuels—Determination of Total Content of Sulfur and Chlorine. ISO: Geneva, Switzerland, 2011.
  49. Van Soest, P.J.; Robertson, J.B. Analysis of Forages and Fibrous Foods; Cornell University: Ithaca, NY, USA, 1985; p. 202. [Google Scholar]
  50. TIBCO Statistica, v. 13.3.0; TIBCO Software Inc.: Palo Alto, CA, USA, 2017. Available online: https://www.tibco.com/products/tibco-statistica(accessed on 22 May 2023).
  51. Pezo, L.L.; Ćurčić, B.L.; Filipović, V.S.; Nićetin, M.R.; Koprivica, G.B.; Mišljenović, N.M.; Lević, L.B. Artificial neural network model of pork meat cubes osmotic dehydratation. Hem. Ind. 2013, 67, 465–475. [Google Scholar] [CrossRef]
  52. Yoon, Y.; Swales, G.; Margavio, T.M. A Comparison of Discriminant Analysis versus Artificial Neural Networks. J. Oper. Res. Soc. 1993, 44, 51–60. [Google Scholar] [CrossRef]
  53. Jablonowski, N.D.; Kollmann, T.; Nabel, M.; Damm, T.; Klose, H.; Müller, M.; Bläsing, M.; Seebold, S.; Krafft, S.; Kuperjans, I.; et al. Valorization of Sida (Sida hermaphrodita) biomass for multiple energy purposes. GCB Bioenergy Bioprod. Sustain. Bioecon. 2016, 9, 202–214. [Google Scholar] [CrossRef]
  54. Meehan, P.G.; Finnan, J.M.; Mc Donnell, K.P. The effect of harvest date and harvest method on the combustion characteristics of Miscanthus × giganteus. GCB-Bioenergy Bioprod. Sustain. Bioecon. 2012, 5, 487–496. [Google Scholar] [CrossRef]
  55. Slepetys, J.; Kadziuliene, Z.; Sarunaite, L.; Tilvikiene, V.; Kryzeviciene, A. Biomass potential of plants grown for bioenergy production. In Growing and Processing Technologies of Energy Crops, Proceedings of the International Scientific Conference Renewable Energy and Energy Eciency, Jelgava, Latvia, 28–30 May 2012; University of Agriculture: Jelgava, Latvia, 2012; pp. 66–72. [Google Scholar]
  56. Borkowska, H.; Molas, R. Two extremely different crops, Salix and Sida, as sources of renewable bioenergy. Biomass Bioenergy 2011, 46, 234–240. [Google Scholar] [CrossRef]
  57. Clifton-Brown, J.; Stampfl, P.; Jones, M.P. Miscanthus biomass production for energy in Europe and its potential contribution to decreasing fossil fuel carbon emissions. Glob. Change Biol. 2004, 10, 509–518. [Google Scholar] [CrossRef]
  58. Battaglia, M.; Fike, J.; Fike, W.; Sadeghpour, A.; Diatta, A. Miscanthus × giganteus biomass yield and quality in the Virginia Piedmont. Grassl. Sci. 2019, 65, 233–240. [Google Scholar] [CrossRef]
  59. Parmar, K. Biomass—An Overview on Composition Characteristics and Properties. IRA-JAS 2017, 7, 42–51. [Google Scholar] [CrossRef]
  60. Grandesso, E.; Gullett, B.; Touati, A.; Tabor, D. Effect of moisture, charge size, and chlorine concentration on PCDD/F emissions from simulated open burning of forest biomass. Environ. Sci. Technol. 2011, 45, 3887–3894. [Google Scholar] [CrossRef] [PubMed]
  61. Stolarski, M.J.; Krzyżaniak, M.; Śnieg, M.; Słomińska, E.; Piórkowski, M.; Filipkowski, R. Thermophysical and chemical properties of perennial energy crops depending on harvest period. Int. Agrophys. 2014, 28, 201–211. [Google Scholar] [CrossRef]
  62. Assefa, A.; Debella, A. Review on dry matter production and partitioning as affected by different environmental conditions. Int. J. Adv. Res. Biol. Sci. 2020, 7, 37–46. [Google Scholar]
  63. Baxter, X.C.; Darvell, L.I.; Jones, J.M.; Barraclough, T.; Yates, N.E.; Shield, I. Study of Miscanthus × giganteus ash composition—Variation with agronomy and assessment method. Fuel 2012, 95, 50–62. [Google Scholar] [CrossRef]
  64. Nazli, R.I.; Tansi, V.; Öztürk, H.H.; Kusvuran, A. Miscanthus, switchgrass, giant reed, and bulbous canary grass as potential bioenergy crops in a semi-arid Mediterranean environment. Ind. Crops Prod. 2018, 125, 9–23. [Google Scholar] [CrossRef]
  65. Šiaudinis, G.; Jasinskas, A.; Šarauskis, E.; Steponavičius, D.; Karčauskienė, D.; Liaudanskienė, I. The assessment of Virginia mallow (Sida hermaphrodita Rusby) and cup plant (Silphium perfoliatum L.) productivity, physico-mechanical properties and energy expenses. Energy 2015, 93, 606–612. [Google Scholar] [CrossRef]
  66. García, R.; Pizarro, C.; Lavín, A.G.; Bueno, J.L. Characterization of Spanish biomass wastes for energy use. Bior. Technol. 2012, 103, 249–258. [Google Scholar] [CrossRef]
  67. Cavalaglio, G.; Cotana, F.; Nicolini, A.; Coccia, V.; Petrozzi, A.; Alessandro Formica, A.; Bertini, A. Characterization of Various Biomass Feedstock Suitable for Small-Scale Energy Plants as Preliminary Activity of Biocheaper Project. Sustainability. 2020, 12, 6678. [Google Scholar] [CrossRef]
  68. Cai, J.; He, Y.; Yu, X.; Banks, S.W.; Yang, Y.; Zhang, X.; Yu, Y.; Liu, R.; Bridgwater, A.V. Review of physicochemical properties and analytical characterization of lignocellulosic biomass. Renew. Sustain. Energy Rev. 2017, 76, 309–322. [Google Scholar] [CrossRef]
  69. Babich, O.O.; Krieger, O.V.; Chupakhin, E.G.; Kozlova, O.V. Miscanthus plants processing in fuel, energy, chemical and microbiological industries. Foods Raw Mater. 2019, 7, 403–411. [Google Scholar] [CrossRef]
  70. Stolarski, M.J.; Snieg, M.; Krzyzaniak, M.; Tworkowski, J.; Szczukowski, S.; Graban, Ł.; Lajszner, W. Short rotation coppices, grasses and other herbaceous crops: Biomass properties versus 26 genotypes and harvest time. Ind. Crops Prod. 2018, 119, 22–32. [Google Scholar] [CrossRef]
  71. Šurić, J.; Brandić, I.; Peter, A.; Bilandžija, N.; Leto, J.; Karažija, T.; Kutnjak, H.; Poljak, M.; Voća, N. Wastewater Sewage Sludge Management via Production of the Energy Crop Virginia Mallow. Agronomy 2022, 12, 1578. [Google Scholar] [CrossRef]
  72. European Committee for Standardization. Solid Biofuels—Fuel Specifications and Classes; CEN: Brussels, Belgium, 2005. [Google Scholar]
  73. Rusanowska, P.; Zielinski, M.; Dudek, M.; Debowski, M. Mechanical pretreatment of lignocellulosic biomass for methane fermentation in innovative reactor with cage mixing system. J. Ecol. Eng. 2018, 19, 219–224. [Google Scholar] [CrossRef]
  74. Dudek, M.; Rusanowska, P.; Zielinski, M.; Debowski, M. Influence of ultrasonic disintegration on eciency of methane fermentation of Sida hermaphrodita silage. J. Ecol. Eng. 2018, 19, 128–134. [Google Scholar] [CrossRef]
  75. Pokój, T.; Bułkowska, K.; Gusiatin, Z.M.; Klimiuk, E.; Jankowski, K.J. Semi-continuous anaerobic digestion of diferent silage crops: VFAS formation, methane yield from fiber and non-fiber components and digestate composition. Bioresour. Technol. 2015, 190, 201–210. [Google Scholar] [CrossRef]
  76. Obernberger, I.; Brunner, T.; Bärnthaler, G. Chemical properties of solid biofuels—Significance and impact. Biomass Bioenerg. 2006, 30, 973–982. [Google Scholar] [CrossRef]
  77. Kron, I.; Porvaz, P.; Kráľová-Hricindová, A.; Tóth, Š.; Sarvaš, J.; Polák, M. Green harvests of three perennial energy crops and their chemical composition. IJAER 2017, 3, 2870–2883. [Google Scholar]
  78. Lewandowski, I.; Kicherer, A. Combustion quality of biomass: Practical relevance and experiments to modify the biomass quality of Miscanthus × giganteus. Eur. J. Agron. 1997, 6, 163–177. [Google Scholar] [CrossRef]
  79. Cumplido-Marin, L.; Graves, A.R.; Burgess, P.J.; Morhart, C.; Paris, P.; Jablonowski, N.D.; Facciotto, G.; Bury, M.; Martens, R.; Nahm, M. Two Novel Energy Crops: Sida hermaphrodita (L.) Rusby and Silphium perfoliatum L.—State of Knowledge. Agronomy 2020, 10, 928. [Google Scholar] [CrossRef]
  80. Fernandes, T.V.; Bos, G.J.; Zeeman, G.; Sanders, J.P.; van Lier, J.B. Effects of thermo-chemical pre-treatment on anaerobic biodegradability and hydrolysis of lignocellulosic biomass. Bior. Technol. 2009, 100, 2575–2579. [Google Scholar] [CrossRef]
  81. Cerazy-Waliszewska, J.; Jeżowski, S.; Łysakowski, P.; Waliszewska, B.; Zborowska, M.; Sobańska, K.; Ślusarkiewicz-Jarzina, A.; Białas, W.; Pniewski, T. Potential of bioethanol production from biomass of various Miscanthus genotypes cultivated in three-year plantations in west-central Poland. Ind. Crops Prod. 2019, 141, 111790. [Google Scholar] [CrossRef]
  82. Zanetti, F.; Scordia, D.; Calcagno, S.; Acciai, M.; Grasso, A.; Cosentino, S.L.; Monti, A. Trade-off between harvest date and lignocellulosic crop choice for advanced biofuel production in the Mediterranean area. Ind. Crops Prod. 2019, 138, 111439. [Google Scholar] [CrossRef]
  83. Mangold, A.; Lewandowski, I.; Möhring, J.; Clifton-Brown, J.; Krzyżak, J.; Mos, M.; Pogrzeba, M.; Kiesel, A. Harvest date and leaf:stem ratio determine methane hectare yield of miscanthus biomass. GCB-Bioenergy Bioprod. Sustain. Bioecon. 2019, 11, 21–33. [Google Scholar] [CrossRef]
  84. Haberzettl, J.; Hilgert, P.; Von Cossel, M.A. Critical Review on Lignocellulosic Biomass Yield Modeling and the Bioenergy Potential from Marginal Land. Agronomy 2021, 11, 2397. [Google Scholar] [CrossRef]
Energies 16 04312 g001 550
Figure 1. Climate diagram with data from the meteorological station ‘Medvednica’ for the period of April (6) to March (9) (Croatian Meteorological and Hydrological Service, 2022).
Figure 1. Climate diagram with data from the meteorological station ‘Medvednica’ for the period of April (6) to March (9) (Croatian Meteorological and Hydrological Service, 2022).
Energies 16 04312 g001
Energies 16 04312 g002 550
Figure 2. Climate diagram with data from the meteorological station ‘Maksimir’ for the period of April (6) to March (9) (Croatian Meteorological and Hydrological Service, 2022).
Figure 2. Climate diagram with data from the meteorological station ‘Maksimir’ for the period of April (6) to March (9) (Croatian Meteorological and Hydrological Service, 2022).
Energies 16 04312 g002
Energies 16 04312 g003 550
Figure 3. Biplot diagram of the elemental composition of Miscanthus (red lines) and Virginia Mallow (blue lines) as a function of harvest period.
Figure 3. Biplot diagram of the elemental composition of Miscanthus (red lines) and Virginia Mallow (blue lines) as a function of harvest period.
Energies 16 04312 g003
Energies 16 04312 g004 550
Figure 4. Global sensitivity of the ANN based on the Yoon method (a) ANN for estimating the HHV based on the structural composition and energy characteristics, (b) ANN for estimating the LHV based on the structural composition and energy characteristics, (c) ANN for estimating the HHV based on elemental composition, (d) ANN for estimating the LHV based on the elemental composition.
Figure 4. Global sensitivity of the ANN based on the Yoon method (a) ANN for estimating the HHV based on the structural composition and energy characteristics, (b) ANN for estimating the LHV based on the structural composition and energy characteristics, (c) ANN for estimating the HHV based on elemental composition, (d) ANN for estimating the LHV based on the elemental composition.
Energies 16 04312 g004
Table
Table 1. Biomass sample dates for analysis of energetic properties after different harvest dates.
Table 1. Biomass sample dates for analysis of energetic properties after different harvest dates.
YearHarvest PeriodMiscanthusVirginia Mallow
1Autumn 20197 November15 November
1Spring 202019 March5 March
2Autumn 20202 November28 December
2Spring 202124 March2 March
Table
Table 2. Dry matter yield of Miscanthus and Virginia Mallow in autumn and spring harvest in two experimental years.
Table 2. Dry matter yield of Miscanthus and Virginia Mallow in autumn and spring harvest in two experimental years.
SampleMiscanthus
Vegetation SeasonI. (2019–2020)II. (2020–2021)
Harvest SeasonAutumnSpringAutumnSpring
DMY (t/ha)39.09 ± 5.26 c20.88 ± 2.15 a45.37 ± 7.89 d28.23 ± 5.08 b
SampleVirginia Mallow
DMY (t/ha)11.49 ± 1.89 b7.51 ± 2.31 a10.23 ± 1.8 b7.81 ± 1.89 a
DMY—dry matter yield. The means in the same column, with different lowercase superscripts, are statistically different (p < 0.05), according to Tukey’s HSD test.
Table
Table 3. Energy characteristics and calorific value of Miscanthus and Virginia Mallow with respect to the harvest period in two experimental years.
Table 3. Energy characteristics and calorific value of Miscanthus and Virginia Mallow with respect to the harvest period in two experimental years.
SampleMiscanthus
Vegetation SeasonI. (2019–2020)II. (2020–2021)
Harvest SeasonAutumnSpringAutumnSpring
DM (%)40.51 ± 1.44 a85.15 ± 2.85 c47.63 ± 0.78 b84.64 ± 2.58 c
ASH (%)2.28 ± 0.31 b1.55 ± 0.27 a2.21 ± 0.3 b1.49 ± 0.16 a
COKE (%)11.7 ± 1.11 b11.68 ± 0.74 ab12.59 ± 0.94 c12.16 ± 1.46 bc
FC (%)9.42 ± 1.11 a10.14 ± 0.76 b10.39 ± 0.92 b10.68 ± 1.46 c
VM (%)80.34 ± 1.21 a83.42 ± 1.24 c80.59 ± 0.91 a82.44 ± 1.39 b
HHV (MJ/kg)17.73 ± 0.34 a17.83 ± 0.39 ab18.51 ± 0.46 c18.7 ± 0.3 c
LHV (MJ/kg)16.41 ± 0.34 a16.49 ± 0.38 a17.28 ± 0.49 b17.53 ± 0.29 c
SampleVirginia Mallow
Vegetation seasonI. (2019–2020)II. (2020–2021)
Harvest seasonAutumnSpringAutumnSpring
DM (%)55.71 ± 5.43 b79.59 ± 2.35 c51.35 ± 0.89 a87.4 ± 0.29 d
ASH (%)2.82 ± 0.45 a2.85 ± 0.45 a4.25 ± 0.96 b3.09 ± 0.41 a
COKE (%)10.99 ± 1.06 bc10.79 ± 0.75 b13.77 ± 1.16 d9.86 ± 0.92 a
FC (%)8.17 ± 1.1 b8.15 ± 0.79 b9.74 ± 1.44 c6.78 ± 0.9 a
VM (%)79.98 ± 1.05 b82.08 ± 1.4 c77.26 ± 1.25 a85.35 ± 0.85 d
HHV (MJ/kg)17.48 ± 0.52 ab17.26 ± 0.29 a18.2 ± 0.15 d17.57 ± 0.19 bc
LHV (MJ/kg)16.14 ± 0.52 a15.91 ± 0.29 a16.8 ± 0.16 c16.39 ± 0.19 b
DM—Dry matter; FC—Fixed Carbon; VM—Volatile matter; HHV—Higher heating value; LHV—Lower heating value; The means in the same column, with different lowercase superscripts, are statistically different (p < 0.05), according to Tukey’s HSD test.
Table
Table 4. Elemental composition of the investigated Miscanthus and Virginia Mallow biomass in two experimental years and a different harvest time.
Table 4. Elemental composition of the investigated Miscanthus and Virginia Mallow biomass in two experimental years and a different harvest time.
SampleMiscanthus
Vegetation SeasonI. (2019–2020)II. (2020–2021)
Harvest SeasonAutumnSpringAutumnSpring
N (%)0.29 ± 0.05 c0.08 ± 0.09 a0.37 ± 0.1 d0.15 ± 0.11 b
C (%)50.69 ± 0.72 a52.62 ± 0.37 d50.8 ± 0.23 b51.6 ± 0.47 c
S (%)0.07 ± 0.01 a0.1 ± 0.02 b0.21 ± 0.05 c0.17 ± 0.18 c
H (%)6.07 ± 0.06 c6.12 ± 0.08 c5.76 ± 0.08 b5.4 ± 0.13 a
O (%)26.78 ± 0.83 b24.69 ± 0.61 a43 ± 0.92 c42.7 ± 0.66 c
SampleVirginia Mallow
Vegetation seasonI. (2019–2020)II. (2020–2021)
Harvest seasonAutumnSpringAutumnSpring
N (%)0.59 ± 0.05 d0.19 ± 0.05 a0.39 ± 0.09 c0.21 ± 0.05 b
C (%)49.42 ± 0.31 a51.51 ± 0.4 c49.65 ± 0.51 b49.12 ± 0.54 a
S (%)0.06 ± 0.01 a0.07 ± 0.03 a0.03 ± 0.01 a0.09 ± 0.04 b
H (%)6.14 ± 0.14 c6.19 ± 0.08 d5.74 ± 0.07 b5.38 ± 0.12 a
O (%)28.26 ± 0.58 b26.35 ± 0.53 a44.03 ± 0.52 c45.23 ± 0.57 d
N—Nitrogen; C—Carbon; S—Sulfur; H—Hydrogen; O—Oxygen. The means in the same column, with different lowercase superscripts, are statistically different (p < 0.05), according to Tukey’s HSD test.
Table
Table 5. Cell Wall Composition of Miscanthus and Virginia Mallow in two harvest periods in two experimental years.
Table 5. Cell Wall Composition of Miscanthus and Virginia Mallow in two harvest periods in two experimental years.
SampleMiscanthus
Vegetation SeasonI. (2019–2020)II. (2020–2021)
Harvest SeasonAutumnSpringAutumnSpring
Cel (%)47.42 ± 1.36 a53.5 ± 1.17 c52.18 ± 1 b53.88 ± 1.41 c
Hem (%)24.33 ± 2.17 b23.58 ± 1.79 a26.01 ± 1.45 d24.82 ± 2.89 c
Lig (%)13.08 ± 0.67 a14.54 ± 1 bc14.18 ± 0.92 b16.09 ± 1.51 d
SampleVirginia Mallow
Vegetation seasonI. (2019–2020)II. (2020–2021)
Harvest seasonAutumnSpringAutumnSpring
Cel (%)56.5 ± 2.23 b55.46 ± 1.85 a56.84 ± 2.03 c56.93 ± 0.78 d
Hem (%)14.97 ± 2.88 b20.11 ± 1.56 c12.09 ± 1.67 a14.61 ± 1.55 b
Lig (%)15.3 ± 2 a16.8 ± 1.14 c15.13 ± 0.89 a15.89 ± 0.89 b
Cel—Cellulose; Hem—Hemicellulose; Lig—Lignin; The means in the same column, with different lowercase superscripts, are statistically different (p < 0.05), according to Tukey’s HSD test.
Table
Table 6. Weight coefficients and biases of input and output layer for ANN model based on structural composition and energy characteristics.
Table 6. Weight coefficients and biases of input and output layer for ANN model based on structural composition and energy characteristics.
Input LayerOutput Layer
WeightBiasWeightBias
CelHemLigDMAshCokeFCVMHHVLHVHHVLHV
3.622.362.451.122.47−5.26−5.36−1.72−2.380.911.00−0.25−0.27
3.64−2.106.202.276.67−1.74−4.06−4.50−2.47−0.92−0.88
−0.67−5.11−6.48−1.05−3.925.075.971.274.550.470.62
10.47−3.33−1.662.354.81−3.69−3.501.498.285.796.14
−0.81−2.66−0.19−0.39−0.521.001.08−1.610.360.030.13
−0.70−9.81−8.86−4.08−1.934.814.88−0.475.11−0.46−0.31
−1.83−4.18−1.31−1.73−0.393.082.50−2.510.51−0.76−0.82
−0.97−7.47−5.81−1.970.012.092.13−1.484.671.371.22
1.1513.018.205.90−5.28−6.79−3.0711.610.05−6.58−6.91
Cel—Cellulose; Hem—Hemicellulose; Lig—Lignin; DM—Dry matter; FC—Fixed Carbon; VM—Volatile matter; HHV—Higher heating value; LHV—Lower heating value.
Table
Table 7. Weight coefficients and biases of input and output layer for ANN model based on structural composition.
Table 7. Weight coefficients and biases of input and output layer for ANN model based on structural composition.
Input LayerOutput Layer
WeightBiasWeightBias
NCSHOHHVLHVHHVLHV
1.330.65−2.512.840.58−0.880.510.520.420.42
−1.133.642.21−10.725.13−0.25−0.53−0.54
−2.470.064.68−14.010.633.980.520.53
HHV—Higher heating value; LHV—Lower heating value.
Table
Table 8. Model performance.
Table 8. Model performance.
InputOutputNet. NameTraining Perf.Test Perf.Training ErrorTest ErrorTraining AlgorithmError FunctionHidden ActivationOutput Activation
cel-hem-lig-DM-Ash-Coke-FC-VMHHV LHVMLP 8-9-20.620.440.080.07BFGS 50SOSTanhLogistic
N-C-S-H-OMLP 5-3-21.001.000.000.00BFGS 256SOSIdentityIdentity
Cel—Cellulose; Hem—Hemicellulose; Lig—Lignin; DM—Dry matter; FC—Fixed Carbon; VM—Volatile matter; HHV—Higher heating value; LHV—Lower heating value; MLP—Multi layer. perceptron.
Table
Table 9. Statistical test “Goodness of fit”.
Table 9. Statistical test “Goodness of fit”.
ModelInputOutputTypeΧ2RMSEMBEMPESSER2SkewKurtSDVar
ANN1Cel Hem Lig DM Ash Coke FC VMHHV/0.020.150.010.458.450.58−0.02−0.330.220.05
ANN2LHV/0.020.14−0.010.488.210.590.09−0.270.220.05
ANN3HHVC.P.1.531.23−0.725.4271.030.760.02−1.931.001.00
ANN4LHVC.P.1.711.30−0.696.5187.370.600.01−1.951.111.23
ANN5N C S H OHHV/0.000.000.000.000.001.0012.50160.040.000.00
ANN6LHV/0.000.000.000.000.001.00−0.061.600.000.00
ANN7HHVC.P.268.7816.2816.2891.082.490.78−0.14−0.920.190.04
ANN8LHVC.P.266.3816.2116.2197.701.930.620.02−0.820.160.03
ANN—Artificial neural network; Cel—Cellulose; Hem—Hemicellulose; Lig—Lignin; DM—Dry matter; FC—Fixed Carbon; VM—Volatile matter; HHV—Higher heating value; LHV—Lower heating value; C.P.—Custom prediction.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Šurić, J.; Voća, N.; Peter, A.; Bilandžija, N.; Brandić, I.; Pezo, L.; Leto, J. Use of Artificial Neural Networks to Model Biomass Properties of Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) in View of Harvest Season. Energies 2023, 16, 4312. https://doi.org/10.3390/en16114312

AMA Style

Šurić J, Voća N, Peter A, Bilandžija N, Brandić I, Pezo L, Leto J. Use of Artificial Neural Networks to Model Biomass Properties of Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) in View of Harvest Season. Energies. 2023; 16(11):4312. https://doi.org/10.3390/en16114312

Chicago/Turabian Style

Šurić, Jona, Neven Voća, Anamarija Peter, Nikola Bilandžija, Ivan Brandić, Lato Pezo, and Josip Leto. 2023. "Use of Artificial Neural Networks to Model Biomass Properties of Miscanthus (Miscanthus × giganteus) and Virginia Mallow (Sida hermaphrodita L.) in View of Harvest Season" Energies 16, no. 11: 4312. https://doi.org/10.3390/en16114312

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop