Results

Homogeneity tests were performed on all the samples received for the tests. Under the tests, 1 g test portion from each sample bottle were fed into a thermo-gravimetric analyser under oxidizing atmosphere. The temperature was raised gradually to 550 °C at rate of 5 K/min. The rate of weight loss with time and temperature represent a characteristic behaviour of each SRF. 34 shows the real time curves for demolition wood, rubber tyre, and dried sledge.

Figure 34: Homogeneity test of the SRF samples

The real time curve showed that dried sledge has high degree of homogeneity than the others. Wood and tyre samples were less resilient in their homogeneity but the later was the worse of them all.

Ash Content

Experiments to test the ruggedness/robustness of the method were carried out by varying conditions of each test-run for samples of the same type. The heating rate between 5 k/min – 20 k/min, the sample weight 1 g, 2 g, oven air change 5 – 10 air changes/min, and the sample bottom area loading of 0.1 – 0.2 g/cm2 _{were the parameters adjusted. Figure 35 illustrate the temperature profile of the oven during} one of the experiments.

Figure 35: Temperature profile for the determination of ash content

The samples were gradually heated to 250 °C; it stays for 60 min before it is finally heated to 550 °C all at a heating rate of 5 k/min. It stays at the final temperature for 120 min. At the end of several experimental runs on different SRF at varying conditions, visual inspections of the resulting ashes revealed no sign of incomplete incineration. Depending on the type of SRF used, in which some contained threads of copper wire (shredded rubber tyre), the presence of copper oxides with distinguishing black colour was obvious. Additionally, a chemical analysis of the ashes to determine the presence of organic carbon revealed that the ashes contained up to 3 % carbon of which they were all of

Figure 36: Different types of SRF and their corresponding ashes: QRA2 (rubber tyre), QRB (demolition wood), and QRC (dried sledge)

Table 6 shows the statistical evaluation of the results for the ash content. Principally, the standard deviation about the mean depends much on the size of the particle and the homogeneity of the sample.

Table 6: Statistical evaluation of the ash content results QR-C <1 mm <6 mm <1 mm <6 mm <1 mm <6 mm pulver <1 mm <6 mm Median 22,31 20,38 23,44 22,40 3,02 3,98 60,42 19,79 18,20 Mean 23,57 21,47 22,89 21,24 3,08 3,88 60,45 20,17 18,48 Deviation 4,60 4,44 2,45 4,22 0,24 0,90 0,26 0,95 1,32 Max-Min. 14,71 31,41 7,04 13,66 0,86 3,38 0,86 3,39 3,76 QR-E Statistics QR-A QR-A2 QR-B

The results show that smaller particle sizes, pulverised and < 1 mm, are more reproducible than larger sizes. The shredded rubber tyres (QRA/2) showed large deviations about the mean. It can be attributed to the highly inhomogeneous nature of the sample which is composed of rubber, metal wires, and textile. This type of SRF is always problematic and very difficult to realise a representative test portion.

Ash melting behaviour

Different shapes and sizes of the sample moulds were used for the ruggedness/robustness tests. The target was to identify the best shape and size that will clearly show the characteristic features during the ash melting process. Three different moulds were used: Cube (3 mm x 3 mm), Cylinder (h = 3 mm, d = 3 mm; and h= 5 mm, d = 5 mm). The phases in the ash melting process are illustrated in Figure 37.

Figure 37: Phases in the ash melting process (original shape = shape and size at 550 °C)

At the end of several runs, it was realized that the side edges of the cube were not visible and it was difficult to ascertain whether deformation has started or not. The cylinder test piece with: h = 3 mm; d = 3 mm was found to be too small in size and it makes it difficult to clearly see the beginning of deformation. However, the cylinder with dimensions h= 5 mm; d = 5 mm were large enough to recognise any phenomenal changes.

Volatile content

Single determination of the volatile content was employed i.e., one sample at a time. A test portion 1 ± 0.1 g is weighed with a silica crucible and lid. The lid ensures that no air is allowed into the crucible. The cold set-up is then inserted into a preheated oven with its temperature maintained at 900 ± 10 °C. The test portion remains in the oven for 420 ± 5 s. The oven, after the insertion of the cold set-up is able to regain its initial temperature (900±10) °C in less than 60 seconds.Figure 38: illustrates the temperature profile of the oven during the experiments.

600 650 700 750 800 850 900 950 0 100 200 300 400 500 600 T i m e , se c

Sample posit ion t emperat ure Temperat ure; oven cent re

Figure 38: Temperature profile at the sample position in the oven.

The oven temperature drops very quickly to temperatures around 750 °C when the oven door is opened to place the cold stand and its crucible. Once the door is closed, it takes about a minute to recover the lost heat and to reach the initial temperature. The thermocouple which is placed at the centre of the oven usually records a peak temperature when massive de-volatilization occurs, mostly within the first 100 s after inserting the cold probe. This phenomenon believed to be exothermic reaction of the released volatiles does not have any significant effect on the experiment since the temperature at the probe position is unaffected.

Table 7: Statistical evaluation of the content of volatile matter QR-C <1 mm <6 mm <1 mm <6 mm <1 mm <6 mm pulver <1 mm <6 mm Median 52,61 55,08 54,45 55,03 72,75 73,93 26,99 79,29 80,01 Mean 53,51 54,84 54,88 55,32 72,69 73,67 27,01 79,15 79,84 Deviation 2,60 2,65 1,51 3,24 0,36 1,54 0,28 0,60 0,67 Max-Min. 10,216 10,978 6,97 13,047 1,3611 8,0919 1,3729 2,3593 3,1042 Statistics

QR-A QR-A2 QR-B QR-E

The results reveal a trend which shows that smaller particle sizes, pulverised and < 1 mm, are more reproducible than larger sizes. Again the shredded rubber tyres (QRA/2) showed large deviations about the mean. The robustness of the method can be said to be very much dependent on the range of particle sizes and for that matter the homogeneity of the sample.

Calorific Value

The statistical summary of the calorific values obtained for the reference materials are presented in Table 8.

Table 8: Statistical summary of calorific values

Reference materials QR-A QR-A2 QR-C QR-E

All samples Average 32.67 31.74 11.87 26.62 Standard deviation 1.47 0.96 0.05 0.39 Max-Min 5.70 4.52 0.13 1.75 Center points Average 32.52 31.77 11.86 26.62 Standard deviation 1.36 0.55 0.05 0.28 Max-Min 3.87 1.29 0.12 0.78

For all the values reported, only that of dried sludge (QR-C) passed the repeatability criteria specified in the TS, which states that the results of duplicate determinations, carried out in the same laboratory by the same operator with the same apparatus within a short interval of time on the same analysis sample, should not differ by more than 0.2 MJ/kg [1]. There were large variation in the results for the shredded rubber tyre, QR-A and QR-A2. These reference materials were highly inhomogeneous and difficult to handle because they contain metallic threads.

Moisture content

Three parameters, sample amount, drying time and drying temperature were varied according to a fractionated factorial designs8. The difference between particle sizes could not be examined because moisture content is a property that is not inherent in the material but dependent on previous handling and the different sizes has passed different preparation steps. Thus, all drying of samples were under equal atmosphere and equal temperature was done simultaneously.

The statistical summary of the moisture content results is presented in Table 8. The responses of the shredded rubber tyre samples (QR-A and QR-A2) are not normally distributed. The distributions generally are skewed to the right, which means majority of the values are very low.

Table 9: Statistical summary of the moisture content results

Reference materials QR-A<1 QR-A<6 QR-A2<1 QR-A2<6 QR-C QR-E<1 QR-E<6

All samples Average 0.92 0.88 1.30 1.36 1.49 2.04 4.72 Standard deviation 0.33 0.07 0.46 0.44 1.12 0.37 1.06 Max-Min 1.22 2.03 1.78 1.41 4.76 1.41 3.13 Center points Average 0.71 0.88 1.37 1.53 1.58 2.04 5.46 Standard deviation 0.14 0.07 0.19 0.33 0.06 0.01 0.50 Max-Min 0.27 0.12 0.39 0.64 0.11 0.02 0.97

The experiments were also conducted under nitrogen atmosphere and the use of lid (Table 10). Table 10: Moisture content under N2 atmosphere and lid

With N2 & Lid Without N2 & Lid

Average St.dev. Average St.dev. t-test QR-A 0.869 0.039 0.786 0.028 Sign.diff. QR-E 4.755 0.315 4.281 0.303 Not sign.diff.

The use of nitrogen and increasing sample size lowers the measured value of moisture, whereas the use of lid and increasing temperature gives higher values. These effects are reasonable; nitrogen atmosphere may help prevent an oxidation of the material that could lead to an over-estimation of moisture level. Larger sample size may slow down the drying process but also reduce the oxidation. High temperature forces more moisture to leave the sample and may also increase oxidation. And the use of lid avoids re-entry of the moisture. Also the dryness time, has significant effects in two of the test series. As expected, it leads to higher measured moisture levels. QR-E < 1mm, paper/plastic mixture and QR-C (dried sledge), seems to be the most easily influenced by variation of parameters. These are also the reference materials where influences from interaction factors are most significant.

Mechanical durability of pellets and briquettes

There were no wide distributions of the results. The durability levels were very close to 100%, showing good qualities of the pellets but making the eventual effects of parameter variations more difficult to evaluate.

The ruggedness/robustness tests were based on two parameters, the weight and the number of rotations at a 95 % significance level. Table 11 shows the results from the durability test conducted for the three different types of pellets.

Table 11: Durability results of pellets

5 mm pellets 8 mm pellets 16 mm pellets

Average 99,37 99,23 99.02

St.dev. 0,50 0,24 0.41

Max 99,93 99,84 99.93

Min 97,64 98,75 97.84

The durability results from the three different pellets showed a fairly normal distribution with the exception of the 5 mm, but the logarithm of the difference between 100% and the durability values (lg(100-Dur)) made a more normal distribution. The results were essentially the same. The variation of the main parameters within the range of the test did not cause any significant changes in durability and fines. For the 8 mm pellets, it is expected the lowering of number of rotations may cause significant effects of the durability. Two parameters did show significant effects on a 95% level on the 16 mm pellets; the numbers of rotations and the sieve size. Increasing the number of rotation and the larger the sieve size according to the model adopted will averagely yield smaller values of durability. Figure 39 a, b, & c shows the durability of the pellets against the number of rotation.

a. 5 mm pellets b. 8 mm pellets

c. 16 mm pellets

Figure 39: Durability of the pellets vs. the number of rotation

One problem in transferring the standard method of durability of biomass and other materials to SRF is also the difference in material consistency. SRF-pellets is often made of shredded material, and whereas biofuel-pellets mostly disintegrate into smaller particles, SRF are broken into foils. In such situations, a circular standardized sieve will not be appropriate for the SRF. These foils can be expected to cause problems in managements of pellets, as such an alternative way of performing durability tests or different sieve design may be considered.

Bulk density

The ruggedness test was performed using one SRF sample, two container sizes and two operators to perform the test. The test was performed with a test portion of about 100 l. Bulk densities were measured after 1, 2, 3, 4 and 5 drops. This was repeated five times. A graphical representation of the results is shown in Figure 40.

Bulk density (BD) vs. container size (L= large 50 l, S=small 5 l), number of drops (1…5) and operator (J, E),

Fuel type, SRF 1 (paper, plastic, wood mix)

100,0 110,0 120,0 130,0 140,0 150,0 160,0 170,0 180,0 1st 2nd 3rd 4th 5th Number of drops BD k g /m 3 BD J (kg/m3), S BD J (kg/m3), L BD E (kg/m3), S BD E (kg/m3), L

Figure 40: Bulk density of SRF as a function of container size and the number of drops

Larger container gave systematically about 10 % higher values for the bulk density than smaller container. Between-operators the coefficient of variability (CV) was 3 % and 6 % for large and small containers respectively. It is thus recommended that only larger (50 l) container is used for SRF bulk density measurements. It is very important to consider moisture content due to wide range of SRF bulk densities. Repeatability variation was between 0.4 - 2.5 %. The method was sensitive towards the compressibility effect. Additionally, significant differences between the number of drops were observed in most of the cases when calculated from the average values of repeats. Averages were compared using t-test9. This was more evident for large container due to lower standard deviation between repeats.

Density of pellets and briquettes

In the test procedure, 10 repeats were made. The coefficient of variation (CV) was analysed for the repeated experiments to study the variations in the results. Biomass pellets have higher density of about 1.3 g/cm3, versus 1.0 g/cm3 for SRF pellets. Coefficient of variation was low (3-8 %) in all cases and even though it decreased when number of repeats was increased, only a slight effect on the average value of density was observed. This indicates that the method was robust against number of repeats. During the immersion in water, pellets will rapidly absorb water and swell. There were systematic differences between results based on dimension measurement and immersion method. The reason for this difference may be the swelling of sample specimen. However, when the average differences were tested using t-test method, there were no significant differences between the results at 95 % confidence level. Thus the methods were comparable.

Bridging properties of SRF

Three different SRFs namely wood chips, fluff, and PET, were used for the tests of bridging properties. Two different particle sizes have been selected for the wood chips, i.e. 0.4 - 2.0 mm and 0.4 - 4.75 mm.

9

EURACHEM/CITAC Guide, Quantifying Uncertainty in Analytical Measurement, Second Edition,

Figure 41 A reports the shear stress as a function of the consolidation pressure in semi-log diagram for the three different SRF materials. Each SRF exhibits an increasing dependence of shear stress on the consolidation pressure tending to an asymptotic line on the right side of the diagram. The larger the consolidation pressure, the more intense the penetration between SRF particles and also the higher the mechanical resistance of the consolidated sample.

A - Shear stress versus consolidation pressure for

different SRFs B - Shear stress versus consolidation pressure for different particle sizes of wood chips

C - Shear stress versus consolidation pressure for two heights of consolidated sample (wood chips

0.4-4.75 mm)

D - shear stress versus moisture content at constant consolidation pressure of 71 kPa (wood

chips 0.4-4.75 mm Figure 41: Shear stress

The highest tendency to bridging behaviour is noted for fluff, followed by wood chips. This finding can be explained by considering the largely irregular shape of fluff particles and fibres, tending to form bridges and bonds between fluff agglomerates. The low moisture content of PET as well as the lubricating effect of plastic materials could explain the very low values of shear stress measured for PET.

Figure 41shows the dependence of the shear stress on the particle size for wood chips. As expected, larger particles promote the bridging tendency by virtue of more intense bonds created in the sample at high consolidation pressure.

The relative high influence of the bed height on the shear stress, in particular at low consolidation pressure, is not completely understood, it could be attributed to the constrains induced by the confinement in the limited volume of the tester.

Figure 41 reports the dependence of the shear stress on the moisture content of wood chips (0.4-4.75 mm particle size) at constant consolidation pressure. The tests have been done by changing the moisture content with addition of distilled water to the sample of fresh wood chips and homogenisation in

a shaker. The material has been used once in order to have well controlled test conditions. The dependence of the shear stress on moisture content is not straightforward: the marked increase on the left side can be explained by the establishment of inter-particle bonds due to the presence of the water (e.g. Van der Walls forces). At higher moisture content the lubricant effect played by the water counterbalances the cohesive forces and becomes predominant on the right side of the diagram.