CAPÍTULO IV. MARCO METODOLÓGICO
4.5. Operacionalización
As is explained above, total porosity (ηt) was obtained from different techniques:
• pycnometer analysis;
• Thin-sections images analysis;
• Medical tomographies images analysis; • X-ray tomographies images analysis.
Weathered/altered volcanic rocks from Solfatara, Ischia and Bolsena exhibit a wide range of
ηt values increasing from 0.69 % for fresh lava to 57.67 % for unwelded ignimbrite, which
also presents the lowest value of bulk weight γ (9.60 KN/m3). Figure 4.8a shows a direct relationship between total porosity and alteration grade of samples.
In general, ηt increases with weathering grade for all rock series (SLA, SPRA, IGT and
BoPRA). Results from X-ray tomography and pycnometer tests reveal that ηt increases
progressively with weathering grade with some minor changes in lava series, where values of SLA3 increase drastically and values from SLA5 present a small reduction. Lava series show the highest increase in porosity with alteration grade, from 0.69 % obtained with medical CT
57 image analysis to 31.5 % obtained with X-ray micro-tomography images. Differences in the results from X-ray micro-tomography and medical tomographies can result from the different resolution and from the consequent averaging. In particular, results reported in Figure 4.8a are obtained starting from estimated density values from which total porosity is computed by knowing the specific weight of the solid phase. In addition, for the porosity values computed from X-ray medical tomography, the density values for different slices are influenced by the coarser resolution with respect to micro CT and the adopted specific weight value of the solid phase used in the calculations. At the same time, medical CT allows to evaluate the total porosity on large rock core samples so it is able to provide a porosity value at a larger scale, where bigger pores are included. Differences could also depend on the required manual thresholding process and image sharpening, which in turn could depend on the modes of acquisition and the heterogeneous nature of the rock.
After all, image analysis represents a rapid and precise method to obtain pore structure (e.g. area, volume, shape, frequency, and spatial distribution). Results suggest an easy individual identification and quantification of pores; moreover, 3D reconstruction of its structure is available, which in turn represents a great tool in explaining pore structure evolution during mechanical tests (e.g. uniaxial and triaxial tests). Small changes as reduction in pore-sizes, closure of small fractures and mode of sample-failure could be explained easily.
Effective porosity (ηe) was obtained from bulk-specific weight measurements and by mercury
intrusion porosimetry. Values increase from 6 % for fresh lava obtained with bulk-specific weight measurements to 65 % to unwelded ignimbrite obtained with mercury intrusion porosimetry. Figure 4.8b shows direct relationship between ηe and alteration grade of
samples. Results reveal that ηe increases progressively with weathering grade with some
changes in IGT series, where values for IGTA present large decrement. ηe obtained from
bulk-specific weight measurements seem to have no clear relationship with weathering grade (Figure 4.8b). The reason could be the percentage and size of interconnected pores and fractures contained in each samples. It means that porous system connectivity does not increase progressively with weathering degrees. The decrease in the computed porosity by imbibition could also be the result of a decrease in size of the pores with the increased alteration and then with a consequent difficulty in their saturation under low air vacuum conditions. Pore evolution is described in terms of structure of the groundmass, nature and degradation of the crystals and pumice fragments and post-depositional alteration processes: a) In general, the evolution of pores in SLA series is related to oxidation and argilization process and connected micro-cracks, as it could be observed by optical microscop. Reduction of porosity in SLA5 is related to hydrothermal processes, as large pores have been filled by new minerals, like amorphous silica and clay minerals.
b) The evolution of pore structure in SPRA series is related to grain-size content and grade of groundmass cementation. In this way, the dominant pore types are secondary; they could be developed during and after a selective dissolution of minerals.
c) The evolution of pore structure in IGT series is related to the high proportion of pumice fragments. Pumice fragments present very open structures, moreover fragmentation of the vesicles walls is very common, degradation of pumice enhances pores connection, consequently a very open structure. The most altered sample in IGT series presents a very dense structure and a drastic reduction in pores content. In this case, pores have been filled by
58 clays, amorphous minerals and small fragments of other materials transported by hydrothermal processes.
d) High porosity values in BoPRA series is related to the depositional processes. Reconstruction of pore structure and microscopy observations reveal also a high percentage of interconnected pores, which is also promoted by degradation and fragmentation of pumice content.
The combination of techniques described above, give a good presentation of grain size and pore size distribution of weathered/altered volcanic rocks. They provide qualitative and quantitative evaluations of total and effective porosity and allow the quantification of spatial pore structure and size distribution. The most relevant conclusions are as follow:
• Significant relationship exists between porosity and alteration/weathering grade for all the samples. Total porosity increases with grade (Figure 4.8a, b).
• Thin section image analysis allows describing pore shape characteristics on a plane (e.g. area, perimeter, circularity, and roughness). Thin section preparation, orientation and size could influence the final estimates.
• Porosity values obtained from mercury porosimeter are generally slightly higher than the values obtained by water immersion method. This could result by the forced of mercury intrusion and damaging or opening of small fractures.
• Connectivity and effective porosity estimates can be obtained from bulk specific weight and mercury porosimeter measurements.
•3D values of porosity can be computed starting from 2D data obtained by means of analysis of images of thin sections (e.g. see Farmer et al., 1991). The final value is controlled by the adopted transformation relationship.
•X-ray tomography is the fastest and more precise technique to obtain 3D textural information. This method allows measuring pore size and pore distribution.
Figure 4.8 a) Total porosity as a function of lithotype obtained from pycnometer test, thin section analysis (3D) and X-ray tomography (Micro CT). b) Effective porosity as a function of lithotype obtained from bulk specific
59 The fractal nature of weathered/altered volcanic rocks is a function of the lithology and the grade of alteration. Fractal dimension was determined from pore volume values collected in thin sections, x-ray tomographies and mercury porosimeter. A simpler approach proposed by Farmer et al. (1991) was followed to compute the 3D porosity by 2D shapes collected from thin section images. Once, 2D conversion was made, values collected from x-ray tomographies and thin sections, which represent connectivity of pores, were wide comparable. Fractal dimension is very useful in interpreting pore frequency distribution. Pore frequency distribution in lava series (SLA), suggests a fractal behaviour of porosity with values between 1.35 and 1.92. While, pore frequency distribution in pyroclastic sequence (SPRA), suggests a fractal behaviour of porosity with values between 1.36 and 1.45. Otherwise, D in IGT series ranges between 1.31 to 1.57 (see also, Table 4.1). In all series, D decreases progressively with weathering sequence, suggesting a relative increase in frequency of large pores (Figure 4.9b- f). The fractal dimension corresponding to the most altered lava rock (SL5) and most altered ignimbrite (IGTA), deviate from this trend, suggesting an increment of the relative frequency of smaller pores, as well as large pores filled by new minerals (amorphous silica and clay minerals). This hypothesis is supported by thin section observations and XRD analyses. Fractal dimension obtained from Mercury porosimeter, represents distribution of non interconnected pores. Variation of D values estimated from ηe data is lower than D values
estimated from ηt, in particular variation is small in SLA and IGT, where D vary from 2.03 to
2.13 and from 2.05 to 2.14, respectively. Larger variation is observed in SPRA series (1.9 to 2.15) where large anisotropy in the matrix is observed and is associated to hydrothermal processes (leaching and deposition). As well as, D values estimated from ηt, these values
decrease progressively with weathering sequence, even if some variations are observed in SLA2, SLA5 and IGT.
Figure 4.9a indicates that different values of the fractal dimension occur at different scale ranges. These results suggest that scaling relationships based on fractal geometry may be very useful in describing a more real distribution of pores. In addition, scaling relationship in pore fractal dimension is restricted principally by the mode of data acquisition. For example, acquisition of pore values in thin-section depends principally on the area of the thin-section, the resolution of the image and the quality of final results depends on a series of noise reduction, filtering, thresholding, and particle separation steps.
60 Figure 4.9 a) Fractal dimension calculated from pores size distributions obtained by different techniques (see legend); b) Pore volume distribution of SLA sequence from thin-section image reconstruction; c) Pore volume
distribution of SPRA and IGTF sequence from thin-section image reconstruction. The lack of information corresponds to the difficulty in preparing thin-sections in very soft samples. d) Pore volume distribution of SLA
sequence from x-ray tomography image reconstruction; e) Pore volume distribution of SPRA sequence from x- ray tomography image reconstruction;. f) Pore volume distribution of IGT sequence and BoBRA from x-ray tomography image reconstruction. g) Pore volume distribution of SLA sequence from mercury porosimetry; h)
Pore volume distribution of SPRA sequence from mercury porosimetry;. i) Pore volume distribution of IGT sequence and BoBRA from mercury porosimetry.