4. CLASSIFICATION AND SEPARATION
4.4. Performance of separation and classification units for RDF/SRF production
4.4.6. Air-flow (or pneumatic) separation
1452
Air-flow separators (or air classifiers, AC) are typically present in RDF/SRF production lines
1453
of MBT plants. Air classifiers have long been established in industrial applications, such as
1454
agriculture and minerals processing, where they are used to separate components from dry
1455
mixtures61, 63, 80. In solid waste management (SWM) they were applied as a key part of
1456
conventional RDF production plants, operated initially on MSW and later commercial or
1457
source-separated waste62. Expectations for AC performance were initially high but a phase of
1458
scepticism followed in the 1990s. This can be attributed to off-the-shelf applications of ACs
1459
proven in other industrial operations, but not adapted or optimised to waste, combined with
1460
unrealistic expectations (e.g. separation of organic from inorganic items, despite their similar
1461
densimetric properties)61, 63, 153
. Currently the confidence in the effectiveness of ACs has
1462
been re-established in practice65.
1463
Within MBT plants, ACs are mainly used for concentrating the high CV combustible
1464
fraction in their low-gravity product65. Other specialised uses include the separation of a
1465
high-plastic film and paper fraction for subsequent material recovery, and for the removal of
1466
plastic from waste intended for landfill disposal in Germany, where legislative upper limits
1467
apply on the CV of landfilled material65. Application of AC for compost product refinement,
1468
with emphasis on the removal of plastics, has recently been considered, with limited
1469
success154. Timmel65reported a typical throughput rate of ACs after the preceding
1470
classification at less than 15 Mg h-1.
1471
Shapiro and Galperin80provided a thorough overview of modern classification
1472
applications, including operation principles, features and performance parameters. However,
1473
their emphasis was not on waste separation, but on particle size separation applications.
1474
Timmel65focused on residual and commercial waste treatment and an older RDF-production
1475
related overview can be found in Hasselriis61TABLE 8 provides relevant data from Timmel65
1476
and other publications.
1477
1478
<<Table 8>>
1479
1480 1481
In typical configurations, separation is based on the differences in inertial (such as
1482
density) and aerodynamic properties (such as size and shape, i.e. measured as granulometric
1483
properties) of the in-feed particles. Air flows through the in-feed waste mixture causing
1484
high-gravity waste particles (constituting the reject) to either fall freely or to be deflected
1485
towards different chutes or conveyors. The low-gravity particles (being the extract) are either
1486
carried away with the off-gasses, to be concentrated downstream in cyclones or fabric filters,
1487
or are deposited on spacious settling chambers. Up to 70% of the classifying air can be
re-1488
circulated, in cross-flow designs48. Within mining processing, separation occurs according to
1489
particle size80, however, in waste treatment the density-dominant separation is more
1490
appropriate and efficient117, 122, 155
. Other sophisticated types of ACs have been developed
1491
that incorporate additional material properties, such as elastic behaviour65. In residual and/or
1492
commercial waste separation, only gravity separators are used, and so far, centrifugal
1493
separators have not been introduced. Cross-flow separators prevail, in which the classifying
1494
air flows perpendicular to the waste and deflects the particles at various distances65(FIGURE
1495
14).
1496
1497
<<Figure 14>>
1498
1499
The performance of ACs depends on the particular design, the mode of operation and the
1500
desirable65, 80: (1) sufficiently narrow particle size ranges in the in-feed; (2) constant, and if
1502
possible, isolated feed of the individual particles; (3) well-defined and stable air-flow and
1503
reduced turbulence; (4) pneumatic conveying through pipelines applied to the low-gravity
1504
material; (5) separation of the low-gravity material from the classifying air; and (6) repeated
1505
cleaning of all fractions.
1506
Hasselriis61and Everett and Peirce117summarised the research that preceded the
1507
development of pulsed air classification. Bartlett156showed that the performance of a zig-zag
1508
air classifier is compromised at high moisture content of the input, and the amount of
1509
adsorbent materials present in the input was identified as an important parameter. The main
1510
effect was on paper density and agglomeration, although plastics were also affected and
1511
reported to the low-gravity product. The composition of the feed, such as the paper-glass
1512
ratio, is also important157.
1513
Both first principles and empirical modelling of the performance of air classifiers has
1514
been attempted, particularly outside waste management. For example, Wang et al.120used
1515
computational fluid dynamics (CFD) simulation of cross-flow AC performance for size
1516
classification and Klumpar114examined performance optimisation of air classification in
1517
closed circuits with grinding. There is little research that is directly relevant to waste sorting.
1518
However, the principles for density-dominant separation through pulsed air classification are
1519
discussed in Vesilind122and Everett and Peirce117. Validation of the air classifier unit
1520
operation of the GRAB99, 100computer model using data from UK RDF plants showed
1521
adequate results for the raw mixed waste at that time, but different coefficients would be
1522
necessary for pulverised waste101. Parameters used were air flow, particle size and density,
1523
shape, and coefficient of variation. He et al.155showed that non-waste simulation of airflow
1524
patterns within passive pulsing air classifiers can raise total effectiveness by 6-8% compared
1525
with conventional ACs. Biddulph and Connor158used effective diffusivity to model and
1526
evaluate the performance of low-gravity and high-gravity products for different duct designs
1527
of ACs, operated at high values of air/solid ratio, reporting better performance for lower
1528
values.
1529
The exact performance of air-separators has to be evaluated by pilot tests, as accurate
1530
design calculations are thought to be impossible because of the problems associated with the
1531
granulometric description of waste particles65. The selection criteria for the appropriate
air-1532
separation equipment include waste composition, particle size of waste stream to be sorted,
1533
required throughput rate and required performance65.
1534
Rotter et al.49presented a large scale comparative study on configurations of separation
1535
and classification equipment for SRF production for residual waste. This study provided
1536
insights into the material flow management performance of ACs. AC unit performance was
1537
among the top performing ballistic separation processes, which include air knife and
1538
crosswise. They achieved high enrichment in lower heating value (LHV) because of the high
1539
plastics percentage. However, this led to a high Cl content. Additionally, failure to
1540
incorporate the wet components into the SRF caused a high enrichment of cadmium (Cd).
1541
These results indicate that for the purpose of mechanical post-treatment of biodried output,
1542
air-classification may perform closer to ballistic separation both in terms of yield and Cl
1543
content, as it would be less difficult to incorporate the paper, card and textile fractions.
1544
TABLE 9 reviews results on air classification performance.
1545
1546
<<Table 9>>
1547
1548