Assuming uniform density, this can then provide an estimated mass of the particle allowing for the production of an estimated mass distribution of the sample for both particle length and width. The particle size distributions for the three grades of aggregate are presented in Fig. 1 while bulk density and particle size distribution parameters are presented in Table 2. The mean aspect ratio is the unweighted average value of a particles length divided by width for the population and provided a numerically comparable value of particle elongation while the interquartile range of mass may be used to compare the spread of distributions.Flexural tests were conducted at 28 days after casting by means of a three point bending test over a span of 300 mm. Tests were conducted at a constant displacement of 3 mm per minute on an Instron 50 KN testing frame with inbuilt instrumentation and large diameter dowel supports were used to minimise any local crushing, Fig. 2. Each variation was tested in two directions: the load applied parallel to the direction of casting force and with the load applied perpendicular to the direction of casting force by rotating the specimen 90 about the major axis, Fig. 2. All data was collected using inbuilt instrumentation at a sampling rate of 10 Hz. Each test was repeated three times. Compressive tests were conducted immediately following the flexural tests at to provide 28 day values in both cases. One half of the specimen was reduced to a 150 mm cube prior to testing by using a band saw fitted with a fine blade to minimise damage. All the tests were carried out on an Intron 50 KN testing frame using the same test parameters and in the same loading direction as the flexural tests, Figs. 2 and 3. As the compressive failure modes of the material are known to be different in the differing testing directions, a parameter that is universally applicable to both conditions is required in order to compare the results. In this case failure of the material is considered to occur at a point of rupture,mobile grow system defined as when the instantaneous stiffness falls to 25% of its recorded maximum based on a 20 point moving average.Thermal conductivity tests were conducted after a minimum of 28 days and after oven drying of the specimens at 105 for 48 h.
All tests were conducted using a Fox 600 heat flow meter at a temperature gradient of 10–30 C and in the orientations indicated in Fig. 3. The specimens were wrapped in a single layer of Clingfilm to protect the machine and limit moisture incursion.Two dimensional image analysis of the internal structure was conducted on 150 mm square slices taken from each of the flexural specimens after testing. The method used was developed in previous work by the authors and fully detailed elsewhere. Six slices were produced in each case in planes perpendicular to the direction of compressive loading,. The slices were encased in a blue casting resin prior to being sanded to reveal a cross section for analysis. The resin has the effect of both stabilising the face, that may be fragile, as well as improving the contrast of voids in the images. Imaging was conducted of the cross sections using a flatbed scanner at a resolution of 1200dpi providing a pixel size of 0.0213 mm square. Enhancement and analysis of the images was conducted in several stages using the software ImageJ. A 10px median filter was first applied to all images to remove noise and smooth outlying pixels by replacing each pixel with the median value of those within the specified radius, the selection of which was based on previous work. Following this, a series of colour threshold filters were used to produce binary images of the air, binder and aggregate and measure their perspective proportions visible at this scale. To assess orientation, the binary images of the aggregates were enhanced with three iterations of a binary opening algorithm to help segregate adjacent particles and analysed using the inbuilt particle analysis tool. The particle analysis tool identifies and measures the discrete binary objects visible within an image, including the length, width and orientation of a fitted ellipse of the same second moments and area. To provide an indication of the overall orientation of the material, orientations of each particle for the full population of all 6 images were combined into a frequency distribution. Based on preceding work where a sensitivity study into the impact of the processing was conducted, the process was controlled with values used based on the proceeding study.The compressive rupture stress, flexural strength, thermal conductivity and particle orientation distribution for specimens tested with differing binder ratios are presented in Fig. 4a, b, c and d respectively. In the parallel direction of loading a strong positive correlation is observed between binder content and the three assessed properties: compressive rupture stress, peak flexural stress and thermal conductivity. In each of these cases the impact of the binder ratio was found to be of greater magnitude compared to the natural variation found in similar specimens indicating the significance of the hemp to binder ratio in determining these properties in this direction.
These findings are in agreement with the previous findings of others who also observed a similar correlation for tests in this direction. In the perpendicular direction of loading, a positive correlation to binder content is again seen for flexural strength and thermal conductivity. The compressive rupture stress is also seen to have a positive correlation to binder ratio between the ratios of 1:1.8 and 1:2.2 however it is not observed for the higher 1:2.6 binder ratio where there is no significant difference from 1:2.2 and a perception of a slight decrease. In all the results a clear and significant difference can be seen in all three properties between the loading directions, which is in line with results of others. There are no known existing studies that consider directly the impact of hemp to binder ratio on perpendicular performance of the material for these results to be compared to. It is observed that the distributions of particle orientations in these two directions are of noticeably differing form: an even distribution imaged in the parallel direction compared to a swayed distribution imaged in the perpendicular direction. The material may be considered to have no preferential orientation in planes perpendicular to initial casting compaction and orientated in parallel planes. In the perpendicular direction this sway of orientation is observed to be greatest in the low binder ratio specimens compared to the higher binder ratio specimens. The degree of orientation therefore appears to be inversely proportional to binder content however the trend is only slight and may not be significant in the reflection of the natural variance observed in the parallel direction imaging. Fig. 5 presents the average stress strain plots from the three specimens of material tested of each binder ratio in both parallel and perpendicular compression and flexure. Fig. 5 reiterates many of the findings observed in Fig. 4 but gives additional insight into the failure modes exhibited. It is noted that in compression the failure mode occurring in loading parallel to the casting compaction is of a change in stiffness and high ductility associated with the failure of the binder structure and subsequent densification of the material. In the perpendicular direction of loading the failure mode is more brittle with a clearly defined peak. In flexure it is noticed that the direction of loading has little to no bearing on the failure mode or stiffness however the form of the plots do imply that binder ratio may have an impact on both the compressive and flexural stiffness.
The compressive rupture stress, flexural strength, thermal conductivity and particle orientation distribution for specimens tested with differing grades of hemp shiv are presented in Fig. 6a, b, c and d respectively. In both the perpendicular and parallel directions there is no correlation between the particle size of hemp aggregates and either the compressive rupture stress, flexural strength or thermal conductivity. Previous studies, often considering only two grades of aggregate, have found both a positive and negative correlation between particle sizes and various physical properties and so in this respect the results can be seen to broadly be in line with previous work. There is however still a distinct and significant difference in both the compressive rupture strength and flexural strength obtained from differing grades of aggregate used: the medium grade is observed to consistently have both the lowest compressive rupture strength and flexural strength in both testing directions. The thermal conductivity in the perpendicular direction was found to be approximately 20% higher than in the parallel direction but again this is independent of grade of aggregate. For all grades considered, the particle orientation distribution is again observed to be even imaged in the parallel to compaction direction and swayed in the perpendicular orientation. In the perpendicular orientation the sway of the distribution is found most pronounced in the coarse grade and least in the fine grade indicating a possible correlation between shiv grade and degree of particle orientation in the material. Fig. 7 presents the average stress strain plots from the three specimens of material tested of shiv grade in both parallel and perpendicular compression and flexure. From Fig. 7 the same difference in failure mode between parallel and perpendicular compressive loading is noticed as in Fig. 5 indicating that this may be independent of both constituent ratio and particle size distribution; again the failure mode in flexure is observed to be consistent in both directions of loading. It can be inferred from Fig. 7, as was observed in Fig. 5,mobile vertical rack that the material has a greater stiffness when loaded perpendicular to initial casting compaction, both in flexure and in compression. In compression it appears that the fine grade of shiv provides the highest stiffness although in general the grade of shiv seems to have limited correlation to this property.
In flexure it is observed that the medium grade of shiv provided the lowest stiffness as well as strength and a general trend between stiffness and strength seems to occur.In the imaging parallel to the direction of casting compaction, the particle orientation distribution was consistently found to be even across all variations of binder content and hemp grade. In contrast, in the perpendicular direction the distribution was consistently found to be swayed towards a horizontal alignment. This is attributed to the compaction applied during the casting process directing the elongated particles of hemp towards stratified planes transverse to compaction. This observation is in line with previous work and also indicated that the process occurs irrespective of binder content or aggregate grade. It can be assumed that all observations of anisotropic properties, present across all specimens, are as a result of this orientated structure.The degree of orientation can be assessed by how prominent the curve of the graph is for the particle orientation distribution in the perpendicular to compaction imaging direction. In the case of binder content an increasing ratio of binder is observed to seemingly reduce the level of particle orientation. It is questionable however if this trend is significant or just natural variation, the extent of which may be indicated in the results from parallel imaging. In addition there are limited explanations for such an occurrence, the most probable being an increased binder content increasing the separation between particles and limiting the effect of compaction in rotating them. In the case of aggregate grade, a fifiner grade was found to also produce a perceived reduction in the level of particle orientation. It can be seen, , that the mean aspect ratio of the shiv particles is also a product of the shiv grade and thus a finer grade can be considered to produce not only smaller particles but also more rounded ones. This reduction of the aspect ratio is almost certainly likely to lessen the extent to which particles are rotated under compaction and thus offers explanation of the perceived lower degree of orientation.