|
more information, since each method has its own strengths and
<br />weaknesses (Reynolds, 2011). Furthermore, since unused raw
<br />material batches were also dumped during production years
<br />(Uddh-Söderberg et al., 2019), there could be a risk of mistaking
<br />the dumped batches for glass hotspots as they are expected to have
<br />similar properties. However, such was not encountered in the cur-
<br />rent investigation. Regardless of the limitations, a high level of con-
<br />fidence in the results was also displayed at Alsterfors dumpsite for
<br />the near-surface materials which was the target of the investiga-
<br />tion since glass dumps are inherently shallow. Confidence in the
<br />results was also evidenced by the matching inverted profiles at
<br />intersections and agreement of data in the 3-D visualisation in
<br />Fig. 5d.
<br />3.1.3. Hotspot material particle size distribution
<br />The particle sizes of the hotspot samples (S1, S2, S3 and S4) was
<br />obtained as shown in Fig. 7. By means, the course fraction (CF) was
<br />dominating (38.3 ± 13.8%) followed by the medium (31.3 ± 9.5%)
<br />and fine (28.3 ± 5.4%) fractions as presented in Fig. 7a. According
<br />to a One-way ANOVA (p < 0.05) the differences among the means
<br />were not statistically significant. The high standard deviations
<br />can be attributed to influence from unusually bigger particles in
<br />a size category especially in coarser categories (Kaartinen et al.,
<br />2013). From previous landfill mining (LFM) investigations, CF have
<br />ranged between 24 and 59.2%, medium fraction (MF) between 21.8
<br />and 29.9%, and FF between 14.8 and 73.6%, depending on study set-
<br />tings, waste type and age (Hogland et al., 2004; Jani et al., 2016;
<br />Mönkäre et al., 2016; Mutafela et al., 2019, 2020). A challenge with
<br />comparing different study findings is that each study site is unique
<br />due to heterogeneity. However, particle sizes of different studies
<br />could be compared using PSD regardless of the study settings.
<br />Fig. 7b shows PSD results based on particles passing through a
<br />sieve mesh. The observed 45 ± 9% for particles <20 mm, for
<br />instance, is comparable with the 45 ± 7% at Kuopio MSW landfill,
<br />while the observed 44% for particles <4 mm is also comparable
<br />with the 44%–57% range at the same landfill in Finland
<br />(Kaartinen et al., 2013; Mönkäre et al., 2016).
<br />PSD influences moisture content, which in turn influences resis-
<br />tivity distribution in subsurface structures as explained in
<br />Section 3.1.5. The bigger the particle sizes, the lower the water
<br />retention capacity, and vice versa. In terms of materials recycling,
<br />excavated material particle sizes are important in identification
<br />of optimum material sorting techniques (Spooren et al., 2013)
<br />and relevant waste processing technologies (Jani et al., 2016).
<br />Furthermore, PSD will be key for future recycling-oriented LFM
<br />activities for separation of FF into exploitable resources to avoid
<br />or minimise materials re-landfilling (Parrodi et al., 2018).
<br />3.1.4. Hotspot material waste composition
<br />Results of hand-sorting of the material fractions are shown in
<br />Fig. 7c. Glass fraction was dominating (87.2%) followed by soil
<br />(6.6%), demolition (4.1%), residual (1.3%) and lastly organic (0.8%)
<br />fractions. Unusually high standard deviations were observed in
<br />some fractions due to the long ranges in results. For instance,
<br />demolition fraction ranged between 0 and 61.5% while soil ranged
<br />between 0 and 44.3%. A One-way ANOVA (p < 0.05) of the results
<br />indicated that there was a statistically significant difference among
<br />the means. According to Turkey’s multiple comparisons test (95%
<br />confidence interval), glass differed significantly from other frac-
<br />tions, while there were no statistically significant differences
<br />among all other means. The results clearly indicate the presence
<br />of glass hotspots as confirmed from hand-sorting of the material
<br />fractions from the four samples. The results further suggest that
<br />while there are sections in the dump with mixed material fractions
<br />as established through ERT models and confirmed by TP excava-
<br />tions, some sections (hotspots) contain materials that could be
<br />readily available for recycling processes if excavated carefully.
<br />However, these results do not suggest a general pattern among dif-
<br />ferent glass dumps, as this may depend on each factory’s disposal
<br />practice. The high resistivities registered in the ERT models around
<br />initially suspected hotspots are thus confirmed by the results
<br />(87.2% glass).
<br />3.1.5. Moisture content
<br />Moisture content varied between the two sites (Alsterfors and
<br />Madesjö) as well as material types (glass hotspots and soil) at
<br />Alsterfors. It was highest in Alsterfors soil samples at 24.9 ± 4.3%
<br />followed by Alsterfors glass hotspot samples at 5.7 ± 1.6% and
<br />lastly Madesjö glass samples at 3.1 ± 0.6%. Moisture content in
<br />landfills is influenced by waste composition, type, properties and
<br />local climatic conditions (Hull et al., 2005). The observed differ-
<br />ences can thus be explained by these factors, including drainage
<br />characteristics (Quaghebeur et al., 2013). Moisture content of exca-
<br />vated waste is an important factor in evaluation of valorisation and
<br />treatment alternatives (Hull et al., 2005; Quaghebeur et al., 2013).
<br />As such, it has been extensively studied in different LFM studies
<br />where it ranged between 22 and 66%, which is comparable to the
<br />Alsterfors soil part in this study (Bhatnagar et al., 2017; Hogland
<br />Fig. 7.(a) Distribution of fraction categories from all samples (n = 4); (b) particle size distribution averages presented in passed mass (%); (c) waste fractions of excavated
<br />hotspot materials.
<br />222 R.N. Mutafela et al./Waste Management 106 (2020) 213–225
|