Classification of waste wood categories according to the best reuse using FT-NIR spectroscopy and chemometrics

May combine two claims.

The classification threshold was set at 16 or 10 correctly classified scans out of 32 (50% and around 30%) because there is high variability in the material.

“We have decided to set a classification threshold of 16 or 10 correctly classified scans (out of 32), i.e. 50% and around 30% of the total number of scans.It is imperative for the success of the classification that we allow for a lower threshold than 100%, as there is a high variability in the material [15,16].”

• cites [15] Huron, 2017, An extensive characterization of various treated waste wood for assessment of suitability with combustion process, Fuel sourceNot checked — no full text of the source
• cites [16] Edo, 2016, Assessment of chemical and material contamination in waste wood fuels – a case study ranging over nine years, Waste Manag. sourceSupported

May combine two claims.

Only a few grams of material are used for lab analysis compared to tons of waste wood handled, leading to reliability problems due to huge variability in material composition.

“Only a few grams of material are used for the lab analysis compared to the tons of waste wood handled, leading to well-known problems of reliability of the results due to the huge variability in material composition [19,20].”

• cites [19] Mancini, 2021, Near infrared technique as a tool for the rapid assessment of waste wood quality for energy applications, Renew. Energy sourcePartly supported

  The Introduction and §2.1 describe waste wood as highly heterogeneous and emphasize representative sampling; §4 also refers to sampling problems with such material. But §2.2 reports using around 300 g for moisture analysis, and the paper does not state that only a few grams are analyzed relative to tons handled or that this makes laboratory results unreliable. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [20] Mancini, 2021, Study of variability of waste wood samples collected in a panel board industry, Front. Chem. sourcePartly supported

  The Results and Discussion section (‘Deciding the Best Sampling Procedure’; page unavailable) says that one subsample is insufficient and that multiple subsamples and scans provide reliable estimates for a heterogeneous waste-wood lot. The Methods section (‘Collection of Waste Wood Samples’; page unavailable) describes 10 L samples, but the paper does not state that only a few grams are analyzed compared with tons handled. It supports the concern about variability and reliability, not the claimed quantities. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Some studies have characterized waste wood composition to determine suitability for combustion and to check product quality and safety associated with recycled products.

“Some studies have already been carried out on the characterization of waste wood composition to determine the suitability for combustion [15][16][17], and also to check the product quality and safety associated with recycled products [13,18,4].”

• cites [15][16][17]Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [13] Bergeron, 2014, Assessment of the coherence of the Swiss waste wood management, Resour. Conserv. Recycl. sourceNot checked — no full text of the source
• cites [18] VisNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [4] Faraca, 2019, Resource quality of wood waste: the importance of physical and chemical impurities in wood waste for recycling, Waste Manag. sourcePartly supported

  The paper’s title and the keywords listed in the indexed Abstract (no page given) identify a study of physical and chemical impurities in wood waste for recycling. The passages retrieved do not state that it assessed the quality or safety of recycled products. The combustion part of the citing sentence has a separate citation group and is not attributed to this source. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

In Germany the third class is sent for incineration.

“On the contrary, in Germany the third class is sent for incineration [9].”

• cites [9] Faraca, 2019, Dynamic accounting of greenhouse gas emissions from cascading utilisation of wood waste, Sci. Total Environ. sourceCould not verify

  The supplied passages and retrieved text identify the paper but do not address how Germany treats a third class of wood waste. Searches for Germany, class III, and incineration found no passage that verifies the claim. (LLM, conf=0.73)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Edo et al. assessed the chemical and physical characteristics of 500 waste wood samples collected at a co-combustion facility in Sweden.

“Edo et al. assessed the chemical and physical characteristics of 500 waste wood samples collected at a co-combustion facility in Sweden.”

• cites [16] Edo, 2016, Assessment of chemical and material contamination in waste wood fuels – a case study ranging over nine years, Waste Manag. sourceSupported

May combine two claims.

Other authors demonstrated that contaminant and concentration levels vary significantly according to wood waste type and source, confirming the high variability of the material and the necessity of its characterization before material sorting.

“Other authors demonstrated that contaminant and concentration levels vary significantly according to wood waste type and source, confirming the high variability of the material and the necessity of its characterization before material sorting [9].”

• cites [9] Faraca, 2019, Dynamic accounting of greenhouse gas emissions from cascading utilisation of wood waste, Sci. Total Environ. sourcePartly supported

  Split into 2 atoms — 0 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 1 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Knowledge of the chemical composition of waste wood is essential for ensuring a fuel with higher quality and lower potential for the formation of pollutants when combusted.

“The knowledge of the chemical composition of such material is essential for ensuring a fuel with higher quality and lower potential for the formation of pollutants when combusted [16].”

• cites [16] Edo, 2016, Assessment of chemical and material contamination in waste wood fuels – a case study ranging over nine years, Waste Manag. sourceSupported

May combine two claims.

The best PLS-DA model used 5 LVs and second derivative, achieving a misclassification error of 19.66%.

“The best model was developed using 5 LVs and second derivative as pre-processing, achieving a misclassification error of 19.66%.”

May combine two claims.

The percentage of non-overlapping spectral area is highest between virgin and treated wood (52.2%) and similar between disposal and virgin (38.3%) and disposal and treated wood (34.7%).

“The percentage of non-overlapping area is the highest between virgin and treated wood (52.2%) and similar between disposal and virgin (38.3%) and disposal and treated wood (34.7%).”

Probably not a substantial claim: may not pass S3 (load-bearing).

MSC and offset correction gave similar PCA results.

“As a note, it can be mentioned that we also have computed the PCA testing different pre-processing methods, where both Multiplicative Scatter Correction (MSC) and offset correction gave similar results (results not shown).”

May combine two claims.

The samples with positive PC1 and negative PC2 belong to fruit boxes (PB type 12), floor (FL type 7), and solid wood furniture (FT type 2), i.e., moderate treated wood.

“These latter samples belong to the categories of fruit boxes (PBtype 12), floor (FLtype 7) and solid wood furniture (FTtype 2), in other words, the moderate treated wood.”

May combine two claims.

The first loading shows main bands at 1422, 1468, 1906, 1922, 2274, and 2326 nm; the second loading at 1706, 1742, 1886, 2226, and 2292 nm.

“In fact, in Fig. 3D it is possible to observe that the first loading shows the main bands at 1422, 1468, 1906, 1922, 2274 and 2326 nm, and the second loading at 1706, 1742, 1886, 2226 and 2292 nm.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Disposal wood samples are spread in the PCA score space.

“Finally, the disposal wood samples are spread in the PCA score space.”

May combine two claims.

VW has two narrow peaks at 2352 and 2362 nm; DW has one peak at 2362 nm; TW has the same peak at 2362 nm and a smaller one at 2350 nm.

“In addition, the VW curve has two narrow peaks at 2352 and 2362 nm, whereas DW presents just one peak at 2362 nm and TW the same peak at 2362 nm and a smaller at 2350 nm.”

May combine two claims.

The PCA score plot with confidence ellipses shows groupings: virgin wood in the upper right, treated wood in the lower left of PC1 and PC2.

“The virgin wood samples are located in the upper right corner of the score plot, while the treated wood samples are located in the lower left corner of PC1 and PC2.”

May combine two claims.

The first two loadings cover 62.67% of the explained variance and detect most characteristic spectral bands.

“The first two loadings were considered as they cover 62.67% of the explained variance and they are able to detect most of the characteristic spectral bands”

May combine two claims.

DW and VW curves have a broader minimum around 2380 nm, while TW has a sharper minimum at 2388 nm.

“Furthermore, DW and VW curves have a broader minimum at around 2380 nm, while TW has a sharper minimum at 2388 nm.”

May combine two claims.

The treated wood samples are mostly located in the negative part of PC1 while virgin wood samples are in the positive direction.

“the treated wood samples are mostly located in the negative part of PC1 while the virgin wood samples are to the positive direction.”

May combine two claims.

Impregnated wood is placed in the area of virgin wood, while painted wood is almost at the centre of the PCA score plot.

“Impregnated wood (IW) is placed in the area of the virgin wood, while the painted wood (PW) is almost at the centre of the PCA score plot.”

May combine two claims.

In area II (1950-2020 nm), the TW curve deviates from DW and VW at 1956 and 1972 nm and the TW peak at 2010 nm is shifted.

“there is a deviation in the TW curve from the DW and VW curves at 1956 and 1972 nm and a shift in TW for the peak at 2010 nm.”

May combine two claims.

DW and VW lines have a minimum at 1470 nm, while the TW minimum is moved to 1488 nm.

“In addition, it can be noted that DW and VW lines have a minimum at 1470 nm, while in the TW line this minimum is moved forward to 1488 nm.”

May combine two claims.

Disposal and virgin wood spectra largely overlap, with differences in the disposal wood spectrum at 2140, 2314, and 2352 nm.

“The disposal and virgin wood curves are to a large extent overlapping, but with some differences in the disposal wood curve located at 2140 nm, 2314 and 2352 nm”

May combine two claims.

In area III (2300-2390 nm), TW and DW lines have a minimum and maximum at 2314 and 2322 nm, while the VW line is smooth.

“A minimum and a maximum can be observed in the TW and DW lines at 2314 and 2322 nm, while the VW line is smooth.”

May combine two claims.

Removing the moderate treated wood subcategories improves both VW-SIMCA and TW-SIMCA models; in validation, VW-SIMCA sample-level misclassification error decreased from 10.00% to 6.67% and TW-SIMCA sample-level misclassification error decreased from 13.33% to 3.33%.

“Both models improve; in the validation phase the misclassification error of VW-SIMCA model (sample level) decreased from 10.00% to 6.67% and for TW-SIMCA model (sample level) the misclassification error decreased from 13.33% to 3.33%.”

May combine two claims.

Impregnated wood is located close to the virgin wood category while painted wood is more spread.

“We found that the impregnated wood is located close to the virgin wood category while the painted wood is more spread.”

May combine two claims.

The percentage of correctly classified disposal wood (samples never classified in VW nor TW classes) has a misclassification rate of 0% in calibration and 10% in validation.

“We obtained a misclassification rate of 0% and 10% in the calibration and validation phases, respectively.”

May combine two claims.

In the mean spectra, IW and virgin wood samples are very similar, while PW and treated wood samples have a similar trend.

“It can be observed that IW and virgin wood samples are very similar, while PW and treated wood samples have a similar trend.”

May combine two claims.

In the PCA space, some DW scans are similar to VW scans, while others are more spread.

“Some DW scans are similar to VW scans in the PCA space, while others are more spread in the scores space.”

May combine two claims.

The developed SIMCA models are consistent and robust.

“The results clearly demonstrate the consistency and robustness of the developed models”

May combine two claims.

Variable selection was applied for improving the classification performance and decreasing the model complexity with good results for VW-SIMCA model, decreasing the misclassification from 10.00% to 3.33%, while no improvement was seen for the TW-SIMCA model.

“Variable selection was applied for improving the classification performance and decreasing the model complexity with good results for VW-SIMCA model, decreasing the misclassification from 10.00% to 3.33%, while no improvement was seen for the TW-SIMCA model.”

May combine two claims.

Two SIMCA models were developed to recognize the virgin wood samples and the treated wood samples, resulting in a validated misclassification error of 10.0% and 13.3% respectively.

“Two SIMCA models were developed to recognize the virgin wood samples and the treated wood samples resulting in a validated misclassification error of 10.0% and 13.3% respectively.”

May combine two claims.

The disposal wood samples can be recognized as the samples neither belonging to the virgin wood or the treated wood classes, and gave a misclassification error of 10.0%.

“The disposal wood samples can be recognized as the samples neither belonging to the virgin wood or the treated wood classes, and gave a misclassification error of 10.0%.”

May combine two claims.

The TW-SIMCA model with variable selection does not improve; in validation, misclassification error increases from 13.33% to 16.67%.

“The model does not improve; in the validation, the misclassification error of the model increases from 13.33% to 16.67%.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The PCA results confirm the small differences between DW and VW found in section 3.1.

“This nicely confirms the findings of the small differences between DW and VW found in section 3.1.”

May combine two claims.

Bands associated with virgin waste wood are normally linked to C-bonds, since lignin and hemicellulose have a higher concentration in virgin wood compared to treated wood.

“The bands associated with virgin waste wood are normally linked to C-bonds, since the lignin and hemicellulose have a higher concentration in the virgin wood, compared to treated wood.”

May combine two claims.

Averaging NIR scans diminishes differences between classes, because chemical differences between samples are not observed homogenously across a sample but in specific areas, so chemical details are found in specific scans, not at the average level.

“In fact, averaging the NIR scans diminishes the differences between the classes, as the chemical differences between the samples are not observed homogenously across a sample, but rather in specific areas, and thus the chemical details are found in the specific scans, and not at the average level.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The test set remained fixed for all computations.

“The test set remained fixed for all the computations.”

Classified as probably not a substantial claim.

Keeping all 32 sub-scans improves representativeness of heterogeneous samples.

“Instead of automatically averaging across these 32 scans, we decided to keep all 32 sub-scans in order to have an improved representativeness of the heterogeneous samples.”

May combine two claims.

The first loading is mostly related to differences between virgin and treated wood, while the second loading shows differences between treated and disposal wood.

“The first loading is mostly related to differences between virgin and treated wood samples, while the second loading shows differences between the treated wood and disposal wood samples.”

Classified as probably not a substantial claim. May combine two claims. The quote only partly states this claim.

The training set (n = 87) and test set (n = 30) were split by constrained random selection ensuring all waste wood categories and subcategories were well distributed.

“The training set (n = 87 samples) was used to develop the classification model and the test set (n = 30 samples) to validate the model and check its classification performance with unknown samples.”

• Also stated in: § 2.4. Multivariate data analysis · plain

  “The selection was constrained to ensure that all waste wood categories and sub-categories are well distributed both in the training and test sets.”

Classified as probably not a substantial claim.

PLS-DA models were validated using venetian blind cross-validation (5 segments, keeping scans of the same replicate together) and an external test set.

“The models were validated both using venetian blind-cross validation (5 segments, keeping scans of the same replicate together in the same segment) and the external test set.”

May combine two claims.

Samples with negative PC1 have a higher degree of glue (mainly composite wood), while samples with positive PC1 and negative PC2 have lower contamination/higher presence of virgin wood.

“The samples with negative values for PC1 have a higher degree of glue (they are mainly composite wood), while the ones placed in the bottom right part of the score plot (positive PC1 and negative PC2 values) present a lower degree of contamination/ higher presence of virgin wood.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The sampling procedure is far from how classification could be implemented in industry.

“We are well aware that this is far from how a classification could be implemented in the industry, as the whole sampling procedure in itself is an important issue.”

Classified as probably not a substantial claim. May combine two claims.

Ten different SIMCA models were computed for each category reducing the training set by around 30% and using the same test set to validate classification results and ensure consistency and robustness.

“Finally, in order to validate the classification results, and ensure consistency and robustness, 10 different SIMCA models were computed for each category reducing the training set by around 30% and using the same samples as before as a test set.”

Classified as probably not a substantial claim.

The best SIMCA model was selected based on the misclassification error in the calibration set.

“all the classification results have been stored and the best model has been selected based on the misclassification error in the calibration set.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The two upper classes are suitable for recycling, the third class may also be used for particleboard production as long as not exceeding thresholds for selected chemical elements, and the last (A4) is for disposal because of hazardous material.

“The two upper classes are suitable for recycling, the third class (treated wood and wood-based panels) may also be used for particleboard production as long as not exceeding thresholds for selected chemical elements and the last (A4) for disposal of because of hazardous material, due to being impregnated with preservatives or painted.”

May combine two claims.

NIRS is a good candidate and is already widely used in other recycling processes, but with a limited number of studies.

“A good candidate is near infrared spectroscopy (NIRS) [21] which is already widely used in other recycling processes, but with a limited number of studies [22,23].”

• cites [21] Blanco, 2002, NIR spectroscopy: a rapid-response analytical tool, TrAC, Trends Anal. Chem. sourceNot checked — no full text of the source
• cites [22] Gundupalli, 2017, A review on automated sorting of source-separated municipal solid waste for recycling, Waste Manag. sourceNot checked — no full text of the source
• cites [23] Lesar, 2018, Quality assessment of recycled wood with and without non-wooden materials from selected recycling companies in Europe, Waste Manag. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

The recycling or reuse of waste wood could be maximized if the waste wood is properly sorted and handled based on quality and characteristics.

“As a consequence, it is evident that the recycling or reuse of the material could be maximized if the waste wood is properly sorted and handled based on the quality and characteristics of the material itself.”

• AssumptionRests on an assumption: Waste wood composition is the key point for deciding the best-suited applications.
• AssumptionRests on an assumption: Proper sorting based on the quality grade of the material is essential to ensure a high-quality, clean and safe recycling loop.

Probably not a substantial claim: may not pass S3 (load-bearing).

Danish classification is similar to the German one.

“Regarding Denmark the classification is similar to the German one.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Glued and painted wood can be used as biofuel in Germany and the United Kingdom but is not allowed in Italy.

“As an example, glued and painted wood can be used as biofuel in Germany and United Kingdom but it is not allowed in Italy.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Finnish classification is similar to the British one, but with some limitations for the energy use of these materials.

“Finnish classification is similar to the British one, but with some limitations for the energy use of these materials [14].”

• cites [14] Garcia, 2017, State-of-the-art of waste wood supply chain in Germany and selected European countries, Waste Manag. sourceNot checked — no full text of the source

May combine two claims.

The highest standard deviation values indicate similarity of some waste wood sub-categories: VW with some DW samples (mainly impregnated wood), and moderate treated wood with virgin wood.

“This probably indicates the similarity of some waste wood sub-categories; VW with some DW samples (mainly impregnated wood), and the moderate treated wood with virgin wood”

May combine two claims.

Landfill is neither sustainable nor cost-efficient compared to reuse.

“In general, the end-point for these waste wood materials is the landfill, but it is neither a sustainable nor a cost-efficient option compared to reuse [8].”

• cites [8] Islam, 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Post-consumer waste wood is divided into four quality classes (A1 to A4).

“The post-consumer waste wood is divided into four quality classes (A1 to A4).”

May combine two claims.

The main challenges for waste wood classification are (i) partial overlap between DW (more specifically impregnated wood) and VW categories and (ii) high variability among different treated wood sub-categories, with moderate treated wood closer to virgin wood category.

“The main challenges for the waste wood classification are i) the partial overlap between DW (more specifically impregnated wood) and VW categories and ii) the high variability among the different treated wood sub-categories, with the moderate treated wood closer to the virgin wood category.”

• AssumptionRests on an assumption: DW spectra are very heterogeneous because the painted or impregnated part of the wood is located only in the external portion of the sample while the core part is made of virgin wood.

Probably not a substantial claim: may not pass S3 (load-bearing).

Countries such as Finland, Italy, Germany and the United Kingdom have defined quality indicators for the classification of waste wood in several categories based on quality and related applications.

“Countries such as Finland, Italy, Germany and United Kingdom have defined the quality indicators for the classification of waste wood in several categories on the basis of their quality and related applications [14].”

• cites [14] Garcia, 2017, State-of-the-art of waste wood supply chain in Germany and selected European countries, Waste Manag. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Recently some authors examined the possibility to use NIR spectroscopy for sorting waste wood material from Amazonian species to improve the efficiency of the carbonization process and consequently charcoal production.

“Recently some authors examined the possibility to use NIR spectroscopy for sorting waste wood material from Amazonian species to improve the efficiency of the carbonization process and consequently charcoal production [42].”

• cites [42] Lima, 2022, Classifying waste wood from Amazonian species by near-infrared spectroscopy (NIRS) to improve charcoal production, Renew. Energy sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

An extensive characterization of properties relevant for combustion was performed on two samples of waste wood and one sample of natural wood for comparison revealing the differences in the combustion reactor.

“An extensive characterization of properties relevant for combustion was performed on two samples of waste wood and one sample of natural wood for comparison revealing the differences in the combustion reactor [17].”

• cites [17] Gehrmann, 2020, Waste wood characterization and combustion behaviour in pilot lab scale, J. Energy Inst. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The application of spectroscopic techniques combined with chemometrics is highly innovative and already applied in other sorting facilities.

“The application of spectroscopic techniques combined with chemometrics is highly innovative and already applied in other sorting facilities [22,23].”

• cites [22] Gundupalli, 2017, A review on automated sorting of source-separated municipal solid waste for recycling, Waste Manag. sourceNot checked — no full text of the source
• cites [23] Lesar, 2018, Quality assessment of recycled wood with and without non-wooden materials from selected recycling companies in Europe, Waste Manag. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

Gehrmann et al. aimed at testing the suitability of waste wood as feedstock in an industrial combustion unit.

“Similarly, Gehrmann et al. aimed at testing the suitability of waste wood as feedstock in an industrial combustion unit.”

• cites [17] Gehrmann, 2020, Waste wood characterization and combustion behaviour in pilot lab scale, J. Energy Inst. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Huron et al. confirmed the heterogeneity of treated wood material, with properties quite different among the different classes of the waste wood characterized.

“In their study Huron et al. confirmed the heterogeneity of treated wood material, with properties quite different among the different classes of the waste wood characterized.”

• cites [15] Huron, 2017, An extensive characterization of various treated waste wood for assessment of suitability with combustion process, Fuel sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Online NIR sorting techniques have the potential benefit of scanning all the material in a rapid and low-cost way, thus improving the wood separation performance.

“Online NIR sorting techniques have the potential benefit of 'scanning' all the material in a rapid and low-cost way, thus improving the wood separation performance.”

• AssumptionRests on an assumption: The current separation of waste wood is mainly based on visual, mechanical, magnetic or gravity sifting techniques.
• AssumptionRests on an assumption: The recycling potential of wood-based material could be expanded and one of the main obstacles is the lack of efficient sorting of the material.

Classified as probably not a substantial claim.

NIR spectroscopy is able to increase the waste wood separation performance.

“NIR spectroscopy is able to increase the waste wood separation performance.”

• AssumptionRests on an assumption: The current separation of waste wood is mainly based on visual, mechanical, magnetic or gravity sifting techniques.
• AssumptionRests on an assumption: The recycling potential of wood-based material could be expanded and one of the main obstacles is the lack of efficient sorting of the material.