Understanding the Geotechnical Behaviour of Pumiceous Soil: A Review

Abstract

Pumiceous deposits, commonly found in volcanic regions such as the Ring of Fire and the Alpide Belt, pose significant engineering challenges due to the presence of highly crushable and compressible grains in their matrix. These deposits exhibit complex geotechnical characteristics and are frequently linked to natural events like landslides and earthquakes. Research in countries such as New Zealand, Japan, Indonesia, Italy, and Central and South America aims to better understand the mechanical behaviour of these materials. Key influencing factors include geological properties, microstructure, shearing characteristics, and the impact of particle breakage. Comparative studies have identified similarities in specific gravity, void ratio, particle size distribution, and shearing mechanisms across regions. However, notable differences appear when compared to hard-grained sands including higher void ratios, variations in relative density due to crushable grains, and increased angularity. Some responses of pumiceous deposits, such as strain softening, liquefaction resistance depending on gradation, and apparent cohesion from grain interlocking, mirror those of hard sands; however, particle crushing plays a crucial role in the behaviour. Accurate numerical modelling, which simulates crushing under different conditions, is essential for characterising pumiceous deposits in situ, providing engineers with a better understanding of these materials across diverse site conditions.

1. Introduction

Pumice is a unique geological material commonly produced during explosive volcanic eruptions. Buck et al. [1] defined its microscopical structures as being described by certain geological features, like magmatic flow (rhyolitic, andesitic, and trachytic) [2,3,4], emplacement (fall and pyroclastic density currents) [5,6], and type of volcanic material (tuff and ignimbrite) [7,8]. Moreover, reworking and redeposition via physical processes, such as rivers and wind, are crucial in mixing and spreading the pumice particles with and around other soils [9,10]. The movement of pumiceous materials across large areas results in the frequent encounter of pumiceous deposits in engineering activities, and their assessment is thus of substantial geotechnical interest [11].

Engineers and researchers in countries such as New Zealand [10,11], Japan [12], Italy [13], Indonesia [14], and South [15] and Central America [16] have observed the presence of pumice soils in various engineering projects and natural hazard events. They have used several geotechnical assessment techniques to characterise the in situ behaviour of pumice soils. Some of the distinctive features of pumice soil encountered so far are that it is lightweight, crushable, and angular [17] and has a vesicular nature with high water absorption capacity [18], a high friction angle with high surface roughness [19], and low crushing strength [20]. These properties of pumiceous materials differ significantly from those of hard-grained sand, and the presence of pumice grains impacts the response of the overall material under monotonic compression [21], strength and deformation reaction under drained condition [22], and liquefaction resistance [23]. Attempting to assess the behaviour of soil containing pumice using empirical correlations developed for hard-grained deposits is likely to lead to erroneous conclusions.

In the last two decades, various researchers [24,25,26,27,28,29] have improved the understanding of the engineering characteristics of pumice deposits. In addition, other research work has considered the influence of (non-plastic) fines on particle-crushing behaviour [30,31]. Most of this research examines the specific properties or characteristics of pumice sediments using either experimental or numerical analysis. However, a summary of the key outcomes from the experimental, numerical, and field studies on pumiceous deposits is still limited.

Alternative research on pumice soil has been conducted, such as determining its geological formation and modifications after emplacement [32], raft formation and dispersal, eruptions and water interaction mechanism [33], use in concrete production [34], as a road pavement material [35], and as a wastewater treatment material [36].

This paper presents an in-depth analysis of the distinctive geotechnical attributes of deposits containing pumice. It starts with a brief introduction of the geological origins of pumice and its role in different natural hazards. Additionally, it outlines the index and engineering characteristics of pumiceous deposits, along with challenges arising from particle breakage. Next, it describes various case studies involving engineering structures built on pumiceous sediments, investigated through experimental and numerical studies. A schematic representation of the literature review is illustrated in Figure 1. Finally, it concludes by summarising the primary conclusions drawn from the review.

2. Geological Origin of Pumice

The formation of pumice materials, or glassy silicic foam from volcanic eruptions, is linked to distinct geological factors such as the type of parent magma, deposition method, consolidation and welding processes, and reworking phenomena. In general, most magmatic eruptions from volcanoes are closely related to plate tectonics, with volcanic events common at plate tectonic boundaries. These events typically produce magma and ash clouds that spread over the vicinity of the source. Consolidation and welding of an existing deposit can occur due to repeated volcanic action in the area, while reworking by fluvial erosion, wind, and slope processes impacts the formation and spreading of volcanic rocks and soil across volcanically active regions.

Pumice deposits from different geographical locations (e.g., the Ring of Fire in Figure 2) typically originate from different geodynamic environments and parent magma (Table 1), where the magma is defined by its oxide and alkali composition (Figure 3). In general, the classification of volcanic rocks is described as a function of silica against the total alkali content [37]. Noor et al. [38] applied this approach to classify pumiceous deposits that had erupted from the Slamet volcano in Indonesia. They found that the pumice fragments had higher silica composition and lower alkali content. This graphical method helps to classify volcanic deposits based on the type of magma. Magma types that commonly produce pumice are rhyolitic [39,40], andesitic [41,42], and trachytic [43,44], as shown in Figure 3. The shaded portion in the figure highlights the common magma types that produce pumiceous deposits. In New Zealand, pumice deposited from the Taupō and Kaharoa eruptions is mainly formed from rhyolitic magma, and it is enriched with silica at around 70–78% by weight [45]. In 1991, significant eruptions from Mount Pinatubo in the Philippines expelled large volumes of pumice generated from an andesitic parent magma [22]. Trachytic and rhyolitic magma generates substantial pumice on Ascension Island [46].

Deposits generated from moderate and powerful volcanic eruptions are commonly categorised into three main types: lava, fall, and pyroclastic density currents or flows. Lava flows are among the most common hazards associated with non-explosive volcanic activity. Their movement is influenced by factors such as viscosity, eruption rate, and ground slope. Mafic lava, with its low viscosity, flows more easily, whereas silicic lava, with its high viscosity, flows less readily. In certain cases, specific types of volcanic events have been linked to the collapse of lava domes, leading to significant variations in the composition of volcanic deposits [47]. In central Mexico, the eruption from Popocatépetl volcano around 23.5 ka generated pumice and ash flows, while at the same time, the lava flow emplaced pumice around the volcano [48]. Young pumice and fall-out deposits were found along the lava flow with samples containing approximately 62% SiO₂. Pumice fragments from lava domes in the Payún Matrú caldera (Argentina) exhibited porphyritic textures with alkali feldspar and were similar to those from pumice cones and domes [49].

Table 1. Examples of magma type and geodynamics in regions with pumiceous deposits.

Type of Magma	Location	Geodynamic Context	Cited from
Rhyolitic	New Zealand	Subduction zone volcanism with significant pyroclastic deposits, plus fluvial and alluvial reworking of these deposits.	[40]
Andesitic	Indonesia	Subduction zone volcanism at the intersection of the Ring of Fire and Alpide Belt with widespread fall and surge deposits.	[44]
Trachytic–Andesitic	South Korea	Intraplate volcanism, resulting in pyroclastic deposits within the sea and onto Japan.	[49]
Andesitic and Rhyolitic	Chile	Subduction zone volcanism with large tephra deposits from multiple eruptions. Andesitic to the north, rhyolitic to the south.	[41]
Dacitic	Philippines	Subduction zone volcanism on the Philippines Mobile Belt resulting in the eruption of Mt Pinatubo and production of pyroclastic flow and fall deposits.	[50]

Table 1. Examples of magma type and geodynamics in regions with pumiceous deposits.

Type of Magma	Location	Geodynamic Context	Cited from
Rhyolitic	New Zealand	Subduction zone volcanism with significant pyroclastic deposits, plus fluvial and alluvial reworking of these deposits.	[40]
Andesitic	Indonesia	Subduction zone volcanism at the intersection of the Ring of Fire and Alpide Belt with widespread fall and surge deposits.	[44]
Trachytic–Andesitic	South Korea	Intraplate volcanism, resulting in pyroclastic deposits within the sea and onto Japan.	[49]
Andesitic and Rhyolitic	Chile	Subduction zone volcanism with large tephra deposits from multiple eruptions. Andesitic to the north, rhyolitic to the south.	[41]
Dacitic	Philippines	Subduction zone volcanism on the Philippines Mobile Belt resulting in the eruption of Mt Pinatubo and production of pyroclastic flow and fall deposits.	[50]

Fall deposits are materials expelled from a volcanic vent during an eruption, which travel through the atmosphere as airborne projectiles. These materials, collectively termed tephra, are classified by Fisher and Schmincke [50] based on their grain size into ash (<2 mm), lapilli (2 to 64 mm), and bombs (>64 mm). Notable geological features of fall deposits include well-sorted particles, angular pumice fragments, and the absence of cross-stratification in deposits [51]. Additionally, the explosiveness and eruption height of the fall materials are utilised by Cas and Wright [52] to classify the eruption styles into Vulcanian (more energetic), Strombolian (long-term), Surtseyan (submarine), Plinian (most powerful and explosive), and Pelean (explosion with collapse) [53]. Of these types, Plinian eruptions are considered the most explosive and powerful of volcanic events dispersing deposits widely because of their great eruptive plume height [54]. Pumiceous airfall deposits have been identified in the eruptions from Maninjau caldera, Indonesia (0.08 ± 0.02 Ma) [55]; Iwato in the Aira caldera, Japan (60 ka) [56]; Phlegraean Fields caldera, Italy (36 ka) [57]; Taupō, New Zealand (1.82 ka) [42]; and Calbuco, Chile (0.917–10.561 ka) [40].

In comparison, pyroclastic flows are rapid streams of hot gas and rock that swiftly advance over the earth’s surface. These flows typically adhere closely to the ground, moving downhill or spreading laterally under the influence of gravitational forces [58] and once deposited typically form an ignimbrite. Pumiceous juvenile materials are a significant component of ignimbrite deposits [59]. In the Taupō region, over the past 20,000 years, five out of ten eruptions have primarily produced fall deposits, while the rest have generated minor volumes of ignimbrites. Aramaki [60] introduced a classification system (Figure 4) for pyroclastic flow deposits, primarily based on the viscosity of the material. Most pyroclastic flows concentrate in the cross-hatched area in the figure. Fragments with lower porosity and higher viscosity are classified as Nueé Ardente with lower volume, while flow material with lower viscosity and higher vesiculation is categorised as pumice flow and has a higher volume of material. Additionally, materials with intermediate viscosity and vesicularity, lying between Nueé Ardente and pumice flow, are termed intermediate flow deposits. The distinct geological features of flow deposits include the non-uniform thickness of deposits across the topography, poorly sorted material (more fine-grained material), and no distinct stratification. These features differ from fall deposits due to their consolidation and welding characteristics. Some notable flow deposits with large amounts of pumice include those from the Tandikat volcano, Indonesia (0.08 ± 0.02 Ma) [55]; Fukaminato, Japan (32–31 ka) [56]; Campa Flegrei, Italy (21.3 ka ±170 years) [61]; and Bezymianny volcano, Russia (1997 AD) [62].

In addition, tuff and ignimbrite are classified based on the degree of welding, influenced by the temperature and compaction of pyroclastic materials after deposition. Ignimbrites, primarily resulting from pyroclastic density currents or flows, are characterised by a dense composition that includes pumice fragments, lithic fragments, and volcanic ash. These deposits exhibit varying degrees of consolidation, from loosely compacted to welded states, depending on the temperature at which they were emplaced [63]. In contrast, tuff is formed from compacted volcanic ash that originates from either ash fallout or pyroclastic flows, generally showing less compaction and welding compared to ignimbrites [64].

Pumice is a significant component within these volcanic deposits, especially within the context of caldera regions and tectonic graben zones, as noted in the study of Westerveld [65]. The process involves the eruption of vesiculated lava from magma chambers and the subsequent foundering of the chamber roofs, leading to the formation of ignimbrite and pumice tuff deposits. These deposits are associated with extensive pyroclastic flows and eruptions, contributing to the widespread distribution of pumice. The classification of pumice-bearing deposits is informed by the nature of the host material, whether ignimbrite or tuff. In ignimbrites, pumice often appears mixed with other volcanic materials, while in tuffs, the pumice may be less compacted and welded.

The importance of understanding the geological setting and the nature of pyroclastic deposits is further supported by studies exploring the geotechnical properties of pumice, which highlight significant differences in their behaviour and characteristics compared to other soil types [12,15,57,66]. The classification between welded and non-welded features is crucial in categorising these deposits and understanding their formation processes and material properties.

The reworking of pumiceous deposits plays an important role in sediment dispersion. The reworking and deposition by alluvial processes results in pumiceous material being mixed with other sedimentary deposit layers in river valleys, flood plains, and hilly areas [10]. Nakayama [67] identified the sedimentology of reworked volcanoclastic deposits in the Ohta tephra (Japan). The grain size, density, and lithofacies analysis revealed the flow and fall deposits from the Ohta tephra included sediments from channel fill, mudflow, swamp, lacustrine, and ephemeral pond deposits [67]. Since pumice fragments can be mixed into other deposits via erosional, fluvial, and alluvial processes, research [14,15,68,69] has been carried out to determine the influence of pumice on the geotechnical behaviour of the resulting sedimentary deposits.

Pumice deposition is primarily influenced by geological processes, such as magma composition, sedimentation dynamics, and subsequent processes like consolidation and reworking. Studies on the geotechnical characterisation [70,71] of pumiceous sediments that consider geological aspects are limited [10,15,55,72,73]; hence, further assessment of the influence of geological characteristics on pumiceous sediments is required.

3. Natural Hazard Events Involving Pumiceous Deposits

The appearance of pumice and its associated volcanic products within sedimentary deposits is evident in regions that are characterised as volcanic and tectonically active. The influence of pumiceous deposits is noted by several researchers during natural hazard events, like earthquakes and landslides [5,12,74,75,76]. Understanding the geotechnical behaviour of pumiceous deposits during such events is required for mitigating the failure of natural and human-made structures.

Some of the major earthquake and landslide events that have occurred in areas where the presence of pumiceous deposits was identified are summarised in Table 2. Pumiceous deposits have been implicated in numerous geohazard events worldwide, spanning various geological ages and locations. These events have occurred in diverse volcanic regions, such as Italy’s Campanian Plain, around Japan’s Hakone and Towada volcanoes, New Zealand’s Taupō Volcanic Zone, El Salvador’s Boqueron Volcano, the Philippines’ Mt. Pinatubo, Ecuador’s Cotopaxi, and Indonesia’s Mt. Bawakaraeng Caldera. The deposits range from ancient formations dating back 400–500 ka to more recent eruptions, such as the 1991 Mt. Pinatubo event. Both flow and fall deposits have been identified in research on geohazards, with the behaviour during seismic and rainfall events being a key focus of research.

Table 2. Natural hazard events involving pumiceous deposits: evaluation of key geotechnical factors.

Country	Event Year	Volcanic Region	Deposit Type Affected	Year of Deposition	Event	Factors Considered	Cited from
Italy	1883	Campanian Plain Mt. Epomeo	Flow and fall	1301 AD	Casamicciola earthquake (North Ischia), X MCS intensity	Landslide susceptibility analysis—field assessment (SPT, shear wave velocity)	[77]
Japan	1923	Hakone volcano	Flow	40 ka	Kanto earthquake with Mw 8.1	Fault modelling from geodetic data	[78]
Japan	1968	Towada volcano	Fall and flow	8.6 ka	Tokachi-Oki earthquake, measuring Mw 7.9	Geomorphological features—LiDAR DEM	[79]
New Zealand	1979	Taupō Volcanic Zone	Fall	Rotoehu ash—60 ka, Hamilton ash 0.08 to 0.38 Ma, Pahoia tephra 0.35 to 2.18 Ma	Rainfall-induced landslides in Omokoroa (1979) and Tauranga City (2005)	Microstructure, CPT, and Atterberg limits	[80]
Italy	1982	Campanian Plain	NR	NR	Flowslide at Monteforte Irpino, during construction with pyroclastic soil in backfill	PLAXIS analysis to estimate compressive strain due to impact	[81]
El Salvador	1986	Boqueron Volcano	Flow and fall deposit	Flow deposits from 1659 to 1917 and fall deposits in 1880	Landslides, differential fill settlement due to earthquake in Santa Marta	Local geological and soil conditions in earthquake	[82]
New Zealand	1987	Taupō Volcanic Zone	Fall	1820 AD	Earthquake in Eastern Bay of Plenty (Edgecumbe) with Mw 6.3	Soil profiling using SPT and CPT, and detailing of damage induced by liquefaction	[74]
Philippines	1991	Mt. Pinatubo	Flow and fall deposit	1991	Volcanic eruption with flow and fall deposits (no significant consolidation and cementation).	Grain size distribution, monotonic (drained torsional shear) and dynamic tests (resonant column and cyclic torsion)	[22]
Ecuador	1996	Cotopaxi volcanic eruption	Flow	1877	Earthquake-induced landslides in Pujili (Inter-Andean valley) (Mw 5.7)	Site response studies using horizontal-to-vertical spectral ratio technique	[75]
Italy	1997	Campi Flegrei volcano	Fall	79 A.D and 130 ka	Rainfall-induced landslide in Sorrento peninsula with pumice and ash	Physical model (combination of 3D mechanical and water flow model) from geological and geotechnical data	[83]
Italy	1998	Mt. Vesuvius	Fall	NR	Rainfall-induced landslides in Pizzo d’ Alvano massif	Laboratory analysis (direct shear, triaxial, and soil water characteristic curves) and finite element modelling using SEEP/W to analyse the triggering mechanism	[84]
Italy	1999	Mt. Vesuvius	Fall	79 AD	Rainfall-induced landslides in Cervinara that consist of two layers of air-fall pumice	Soil water characteristic curve estimation from small-scale physical and mathematical modelling	[85]
Mexico	1999	Los Humeros caldera	Flow and fall	25 ka	Rainfall-induced landslide and mass movements in welded and unwelded pumice	Sedimentology, grain size, permeability and stability analysis using Monte Carlo method	[86]
El Salvador	2001	Coatepeque and Ilopango volcano	Flow	400–500 ka	Earthquake-induced landslides and liquefaction of fine-grained pumice and pumice blocks	Liquefaction characteristics from grain size and the Swedish penetration test [76]; slope stability using the Diana Swandyne II model [87]	[76,87]
Indonesia	2004	Mt. Bawakaraeng Caldera	NR	NR	Precipitation-induced landslides and debris flow in South Sulawesi of volcanic deposits	Rainfall landslide analysis and prediction using empirical method by Crozier [88]	[89]
Italy	2005	Phlegraean field volcanoes	Flow and fall deposit	39 ka	Rainfall-induced landslides of welded and unwelded pyroclastic materials formed in Camaldoli hills, Naples	Geological and hydrological conditions assessed through matric suction and numerical analysis (VS2DTI code)	[90]
Japan	2008	Iwate prefecture	NR	90–100 ka	Earthquake-triggered landslide (Mw 7.2) in pumiceous deposits in Honshu	Field investigation using CPT and laboratory tests (cyclic triaxial) to assess liquefaction potential	[12]
Indonesia	2009	Tandikat volcano	Fall	0.08 ± 0.02 Ma	Earthquake (Mw 7.6) induced landslide of pyroclastic fall deposit in Pariaman City	Geological history, mineralogical studies, DCP, and shear strength characteristics	[55]
Chile	2010	Maipo volcanic complex	Fall	450 ± 60 ka	Widespread earthquakes and damages in Pudahuel ignimbrite	Monotonic and dynamic analysis using direct shear and dynamic back-pressured shear box	[66]
Chile	2010	Villarrica volcano	Flow and fall	11 ka	Landslides induced by an earthquake in south-central Chile	A range of geotechnical tests, including index property analysis, one-dimensional consolidation, drained direct shear, cyclic triaxial, and in situ tests (such as CPT)	[91]
Japan	2011	Fukushima prefecture	Fall	NR	Liquefaction triggered in the Kanto region (Mw 9.0) along with landslides	Soil profiling and shear strength characteristics are examined alongside cyclic triaxial tests	[92]
Japan	2016	Kumamoto plain (Mt. Aso)	Flow and fall deposit	90 ka to 0.3 ka	A sequence of earthquakes (Mw 6.2, 7 and 6) on the Kumamoto Plains triggered liquefaction and landslides	Liquefaction characteristics of pyroclastic deposits determined using cyclic triaxial tests, supplemented by SPT and CPT soil characterisation	[29]
Japan	2018	Tarumae, Shikotsu, and Eniwa volcano	Fall	Shikotsu—46 ka, Eniwa—19 to 21 ka Tarumae—9 ka	Earthquakes (Mw 6.7) and landslides in Hokkaido	Features of slope failure on pyroclastic deposits using geological data and DCP	[93]

Note: NR—not reported.

Table 2. Natural hazard events involving pumiceous deposits: evaluation of key geotechnical factors.

Country	Event Year	Volcanic Region	Deposit Type Affected	Year of Deposition	Event	Factors Considered	Cited from
Italy	1883	Campanian Plain Mt. Epomeo	Flow and fall	1301 AD	Casamicciola earthquake (North Ischia), X MCS intensity	Landslide susceptibility analysis—field assessment (SPT, shear wave velocity)	[77]
Japan	1923	Hakone volcano	Flow	40 ka	Kanto earthquake with Mw 8.1	Fault modelling from geodetic data	[78]
Japan	1968	Towada volcano	Fall and flow	8.6 ka	Tokachi-Oki earthquake, measuring Mw 7.9	Geomorphological features—LiDAR DEM	[79]
New Zealand	1979	Taupō Volcanic Zone	Fall	Rotoehu ash—60 ka, Hamilton ash 0.08 to 0.38 Ma, Pahoia tephra 0.35 to 2.18 Ma	Rainfall-induced landslides in Omokoroa (1979) and Tauranga City (2005)	Microstructure, CPT, and Atterberg limits	[80]
Italy	1982	Campanian Plain	NR	NR	Flowslide at Monteforte Irpino, during construction with pyroclastic soil in backfill	PLAXIS analysis to estimate compressive strain due to impact	[81]
El Salvador	1986	Boqueron Volcano	Flow and fall deposit	Flow deposits from 1659 to 1917 and fall deposits in 1880	Landslides, differential fill settlement due to earthquake in Santa Marta	Local geological and soil conditions in earthquake	[82]
New Zealand	1987	Taupō Volcanic Zone	Fall	1820 AD	Earthquake in Eastern Bay of Plenty (Edgecumbe) with Mw 6.3	Soil profiling using SPT and CPT, and detailing of damage induced by liquefaction	[74]
Philippines	1991	Mt. Pinatubo	Flow and fall deposit	1991	Volcanic eruption with flow and fall deposits (no significant consolidation and cementation).	Grain size distribution, monotonic (drained torsional shear) and dynamic tests (resonant column and cyclic torsion)	[22]
Ecuador	1996	Cotopaxi volcanic eruption	Flow	1877	Earthquake-induced landslides in Pujili (Inter-Andean valley) (Mw 5.7)	Site response studies using horizontal-to-vertical spectral ratio technique	[75]
Italy	1997	Campi Flegrei volcano	Fall	79 A.D and 130 ka	Rainfall-induced landslide in Sorrento peninsula with pumice and ash	Physical model (combination of 3D mechanical and water flow model) from geological and geotechnical data	[83]
Italy	1998	Mt. Vesuvius	Fall	NR	Rainfall-induced landslides in Pizzo d’ Alvano massif	Laboratory analysis (direct shear, triaxial, and soil water characteristic curves) and finite element modelling using SEEP/W to analyse the triggering mechanism	[84]
Italy	1999	Mt. Vesuvius	Fall	79 AD	Rainfall-induced landslides in Cervinara that consist of two layers of air-fall pumice	Soil water characteristic curve estimation from small-scale physical and mathematical modelling	[85]
Mexico	1999	Los Humeros caldera	Flow and fall	25 ka	Rainfall-induced landslide and mass movements in welded and unwelded pumice	Sedimentology, grain size, permeability and stability analysis using Monte Carlo method	[86]
El Salvador	2001	Coatepeque and Ilopango volcano	Flow	400–500 ka	Earthquake-induced landslides and liquefaction of fine-grained pumice and pumice blocks	Liquefaction characteristics from grain size and the Swedish penetration test [76]; slope stability using the Diana Swandyne II model [87]	[76,87]
Indonesia	2004	Mt. Bawakaraeng Caldera	NR	NR	Precipitation-induced landslides and debris flow in South Sulawesi of volcanic deposits	Rainfall landslide analysis and prediction using empirical method by Crozier [88]	[89]
Italy	2005	Phlegraean field volcanoes	Flow and fall deposit	39 ka	Rainfall-induced landslides of welded and unwelded pyroclastic materials formed in Camaldoli hills, Naples	Geological and hydrological conditions assessed through matric suction and numerical analysis (VS2DTI code)	[90]
Japan	2008	Iwate prefecture	NR	90–100 ka	Earthquake-triggered landslide (Mw 7.2) in pumiceous deposits in Honshu	Field investigation using CPT and laboratory tests (cyclic triaxial) to assess liquefaction potential	[12]
Indonesia	2009	Tandikat volcano	Fall	0.08 ± 0.02 Ma	Earthquake (Mw 7.6) induced landslide of pyroclastic fall deposit in Pariaman City	Geological history, mineralogical studies, DCP, and shear strength characteristics	[55]
Chile	2010	Maipo volcanic complex	Fall	450 ± 60 ka	Widespread earthquakes and damages in Pudahuel ignimbrite	Monotonic and dynamic analysis using direct shear and dynamic back-pressured shear box	[66]
Chile	2010	Villarrica volcano	Flow and fall	11 ka	Landslides induced by an earthquake in south-central Chile	A range of geotechnical tests, including index property analysis, one-dimensional consolidation, drained direct shear, cyclic triaxial, and in situ tests (such as CPT)	[91]
Japan	2011	Fukushima prefecture	Fall	NR	Liquefaction triggered in the Kanto region (Mw 9.0) along with landslides	Soil profiling and shear strength characteristics are examined alongside cyclic triaxial tests	[92]
Japan	2016	Kumamoto plain (Mt. Aso)	Flow and fall deposit	90 ka to 0.3 ka	A sequence of earthquakes (Mw 6.2, 7 and 6) on the Kumamoto Plains triggered liquefaction and landslides	Liquefaction characteristics of pyroclastic deposits determined using cyclic triaxial tests, supplemented by SPT and CPT soil characterisation	[29]
Japan	2018	Tarumae, Shikotsu, and Eniwa volcano	Fall	Shikotsu—46 ka, Eniwa—19 to 21 ka Tarumae—9 ka	Earthquakes (Mw 6.7) and landslides in Hokkaido	Features of slope failure on pyroclastic deposits using geological data and DCP	[93]

Note: NR—not reported.

Earthquake-induced events involving pumiceous deposits have been particularly notable in Japan and Chile, e.g., [78,91,92]. The 1923 Kanto earthquake in Japan (M_w 8.1) and the 2011 Tohoku earthquake (M_w 9.0) caused significant damage, including landslides in areas with pyroclastic deposits. Similarly, the 2016 Kumamoto earthquake sequence (M_w 6.2, 7, and 6) in Japan and the 2010 Chile megathrust earthquake (M_w 8.8) triggered landslides and liquefaction in volcanic deposits. Researchers studying these events (as outlined in Table 2) have focused on shear strength characteristics, liquefaction potential evaluation through cyclic triaxial tests, and seismic slope stability analysis. Additionally, horizontal-to-vertical spectral ratios have been used to identify soil column resonant frequencies.

Rainfall-induced landslides in pumiceous deposits have also been significant, with notable events occurring, including the 1997 and 1998 landslides in Italy’s Sorrento peninsula and Pizzo d’Alvano massif, respectively, and the 2005 landslides in New Zealand’s Tauranga City. For these events, researchers have examined the grain size distributions, hydraulic characteristics (e.g., saturated hydraulic conductivity), soil–water characteristic curves, and in situ soil suction measurements. Unsaturated shear strengths have been evaluated through direct shear and triaxial tests, and rainwater infiltration modelling has been employed to understand the triggering mechanisms.

Pumiceous deposits have been associated with various natural hazards in recent history. Events include the 1883 earthquake-induced landslides in Italy’s Campanian Plain and the 2018 Hokkaido Eastern Iburi earthquake in Japan, to rainfall-induced landslides, such as those along Italy’s Sorrento Peninsula in 1997 and the Pizzo d’Alvano massif in 1998. Incidents have occurred in diverse geological settings and involved pumiceous deposits of varying ages, such as the 0.08 Ma Tandikat volcano in Indonesia and the more recent products from the 1991 Mt. Pinatubo eruption in the Philippines. Testing methods, including in situ, laboratory, and numerical methods, were employed by researchers to characterise the behaviour of pumiceous deposits under various geohazard conditions [22,55].

Geotechnical testing of pumiceous deposits has been important in understanding their behaviour during earthquakes and landslides. Common tests include SPT, CPT, and dynamic cone penetration (DCP) tests for in situ characterisation. Laboratory tests, such as grain size analysis, Atterberg limits, monotonic and cyclic triaxial compression tests, and direct shear tests, are frequently employed to determine strength characteristics and liquefaction potential. Researchers also conduct soil–water characteristic curve analyses, saturated hydraulic conductivity tests, and resonant column tests to assess hydraulic properties and dynamic behaviour. Microstructural studies using scanning electron microscopy (SEM), X-ray computed tomography, and energy-dispersive X-ray spectrometry (EDS) have been executed to analyse the unique behaviour of pumiceous deposits.

Despite these comprehensive testing methods, there are limitations in fully characterising pumiceous deposits due to their complex nature. The high crushability and unique particle shape of pumice can lead to difficulties in interpreting standard test results. Additionally, the variability in deposit age, composition, and weathering across different locations poses challenges in developing universally applicable models. To address these limitations, there is an increasing need for advanced numerical modelling techniques to support a more detailed understanding. Finite element analyses, such as those using software like SEEP/W, VS2DTI, and Diana Swandyne II, have been employed to simulate rainwater infiltration, landslide early warning system design, and seismic response. The utilisation of various numerical modelling techniques on pumiceous deposits is discussed in Section 4.6, which reviews the research conducted on pumiceous deposits using various constitutive models and highlights the need for a specific model that considers the effects of particle crushing.

The collected articles on pumiceous deposits associated with various natural hazards, as outlined in Table 2, reveal the presence of air fall, flow, and mixed fall and flow deposits. Due to the wide emplacement of pumiceous deposits from eruptive and reworking processes, pumice has been identified in many different regions. As a result, research has been conducted focusing on both geological and geotechnical aspects, both of which are important to clarify the behaviour of pumiceous materials.

4. Geotechnical Characteristics of Pumiceous Soil

Understanding the influence of pumice in soil mechanics is critical because it significantly affects the soil response to natural hazards, like landslides [55,94] and liquefaction [95,96]. Pumice poses a unique challenge for geotechnical engineers due to its lightweight nature and tendency to break and compress easily [94,97,98]. Additionally, pumiceous soil behaviour deviates significantly from that of ordinary sands, particularly under monotonic and cyclic loading conditions [29,99]. The analysis of pumiceous deposits, like any unusual soil, includes examining the challenges and techniques associated with sampling, microstructural studies to assess the internal structure of the particles, specific gravity, grain size distribution, void ratio, density, and shearing characteristics using direct shear and triaxial testing (both monotonic and cyclic). Further, the influence of particle breakage mechanisms on index and engineering properties must be investigated. Understanding when and how pumice particles break under stress is crucial, as this significantly impacts the overall behaviour of the soil under different loading conditions.

4.1. Field Sampling

Ensuring accuracy in soil sampling operations is critical for analysing the in situ engineering properties of soils to avoid misinterpretation [100]. This is particularly relevant in crushable volcanic sediments, where unique characteristics, such as weak cementation [101,102], thin strata [2], and challenges in evaluating effective porosity [103], result in an inadequate representation of the engineering characteristics. To minimise disturbance during sampling, various techniques have been employed, including block sampling [101,104], thin-walled tube sampling [105], piston sampling [24], the use of specialised samplers such as the Shelby sampler [13], and advanced techniques like hydraulically activated fixed piston samplers [106], the Dames and Moore sampler [70], and the gel push sampler [107]. The gel push triple tube sampler (GP-TR) can recover good-quality samples in coarser material, including from sites with high pumice contents [17,108]. Stringer [17] highlighted the significance of sample disturbance in estimating cyclic resistance based on other similar research studies. Higher disturbance in the sample showed a reduction in cyclic resistance. The studies on sampling techniques illustrate the need for careful sampling using advanced methods to accurately represent the index properties, microstructural features, and engineering characteristics of pumiceous soil.

4.2. Index Properties

Before defining the engineering and liquefaction characteristics, initial assessments such as evaluating geological conditions and grain-size composition are critical [2]. Pumiceous deposits are commonly classified as silty sand and sand in soil classification schemes [2,13,24,105], and these deposits are often designated as highly susceptible to liquefaction [96,109,110]. In some regions, pumiceous deposits show the presence of clay due to weathering [111] and landslide events [112,113].

Index properties of pumiceous deposits indicate a lower specific gravity, lower density (total and dry), and higher void ratio than other natural deposits due to internal porosity. These results are consistently observed across all locations, with variations seemingly dependent on the type of deposit from the volcano (fall vs. flow), as summarised in Table 3. The data reveal distinct characteristics between fall and flow deposits, with fall deposits generally displaying higher void ratios (ranging from 2.2 to 5.65 in Hokkaido, Japan) compared to flow deposits (around 1.4 in the same region).

Compared to non-volcanic soils, the unique properties of pumiceous deposits could be associated with the numerous voids inside the pumice particles [114], which enhance the particles’ crushability under compression. These intraparticle voids create challenges in estimating index properties and push researchers to explore different techniques to define these characteristics accurately (e.g., specific gravity [115], void ratio [116], pumice content [20,117]). To illustrate, Esposito and Guadagno [118] adopted mercury intrusion porosimetry to characterise the porosity of pumice, which underestimated the presence of voids in the pumiceous clasts. It signifies that even adopting unique techniques can provide misleading information when characterising pumice. To overcome such issues, the micro-level characteristics of pumiceous particles have been observed to consider their influence on macro-level behaviour, which Orense and Pender [11] found to be significant. Comparing the properties of pumiceous deposits with natural sand highlights its unique characteristics, as demonstrated in Figure 5. Here, the unique nature of the pumiceous sediments was expressed through a high void ratio (up to 8.029), very low dry density (0.511 g/cm³), non-uniform specific gravity (1.3 to 2.7), and variable particle size distribution.

The influence of age on the geotechnical properties of volcanic deposits is complex and non-linear. While older deposits tend to exhibit lower void ratios and higher densities, likely due to natural compaction and cementation processes over time, this trend is not uniform across all samples. For instance, deposits from Santiago, Chile, dating from 450 ka ± 60 ka, demonstrate a relatively high bulk density of 1 g/cm³ [91]. However, the variability observed across different ages underscores the significant impact of local geological conditions, depositional environments, and material types on the mechanical properties of these deposits.

Pumiceous deposits emerge as a unique category within volcanic soils, characterised by extremely high void ratios, low specific gravities, and low densities. These distinctive properties are consistent across different regions, including Japan, New Zealand, and Italy, suggesting that the fundamental characteristics of pumice are largely determined by their formation process rather than by geographical location. Void ratios in pumiceous deposits can reach up to 8.029, as observed in samples from Hokkaido, Japan [5]. This exceptionally high porosity is attributed to inter- and intraparticle voids, with the internal vesicles within each pumice particle contributing significantly to the overall void ratio.

The dataset analysed in this study covers a broad temporal range, from recent eruptions like Mt. Pinatubo in the Philippines (1991 AD) to ancient deposits such as those from the Michoacán–Guanajuato Volcanic Field in Mexico (1.42 ± 0.12 to 0.33 ± 0.04 Ma). While countries like Japan, Italy, and regions in South and Central America are well represented in the dataset, there are notable geographical gaps. Regions such as Southeast Asia (beyond the Philippines and Indonesia), New Zealand (it is focused on the Taupō Volcanic Zone in the North Island), and the Middle East are underrepresented, highlighting the need for more comprehensive global studies of volcanic deposits.

Future research directions should focus on expanding the geographical coverage of volcanic soil studies, particularly in underrepresented regions. There is a pressing need for the development of standardised testing methods specifically tailored for pumiceous materials, given their unique properties and behaviour. Additionally, more emphasis should be placed on undisturbed sampling and in situ characterisation of pumiceous deposits to better understand their natural state properties, which are often influenced by factors such as fabric, structure, ageing, and stress history.

This brief analysis underscores the complex and variable nature of pumiceous soils. The consistent patterns observed in pumiceous materials across different geographical locations emphasise the importance of understanding their fundamental properties for accurate geotechnical analysis and design. As engineering projects increasingly encounter these challenging materials, continued research and development of specialised characterisation techniques will be crucial in improving geotechnical practices in volcanic regions worldwide. Microstructural studies are often adopted to enhance the accuracy of findings obtained from conventional testing techniques, facilitating a more precise characterisation of pumiceous sediments.

4.3. Microstructure and Mineralogy

The unique behaviour of pumiceous deposits is often confirmed through microstructural studies. Various techniques, such as scanning electron microscopy (SEM) [119], micro-computed tomography (CT) [10], optical microscopy [32], and environmental scanning electron microscopy (ESEM) [117], have been employed to analyse the microstructure of pumiceous sediments. Distinct features, including highly vesicular pumice particles [32], interconnected voids [118], rough texture [104], higher angular coefficient [11], higher interlocking effect with the variability of grain shapes [120], and weathering characteristics [4] are generally observed. The complex surface shapes of pumice grains often result in a higher void ratio for deposits than those of natural sands [11]. Microstructural studies have been shown to provide relatively accurate results in quantifying features like pumice content [117], bulk volume, volume of solids, and internal voids [11].

The chemical composition and oxide components of pumiceous deposits are influenced by the parental magma (rhyolitic [121], andesitic [22], trachytic [43], etc.). In addition to the magmatic characteristics, the type of deposition (fall or flow), weathering, and redistribution features shape the mineralogical characteristics of pumiceous sediments. Moreover, the presence of major oxide compounds and minerals in pumice deposits is used to map the behaviour under different loading conditions. In general, pumice particles consist primarily of SiO₂ [8,122] with a solid density of around 2.3 to 2.35 [20,116], although specific gravity testing suggests this has a wide range, as shown in Table 3. In most cases, the specific gravity of pumiceous sand is much lower than natural sand [123,124,125]. Apart from the particle size, the presence of other minerals, such as Al₂O₃, FeO, Na₂O, and CaO, has some influence on the specific gravity of soil with the presence of heavier nano-silica impurities, like iron oxides, able to increase the density beyond that of quartz particles (2.65) [122]. Pumiceous deposits with specific gravity close to or greater than silica sand are also documented in the literature [14,22,126].

Table 3. Summary of pumiceous deposits’ physical parameters in different locations.

1 Japan
Region	Erupted from	Year of Origin	Type of Material	Soil Classification	Void Ratio	Specific Gravity	Total Density	Dry Density	Cited from
Hokkaido	Shikotsu volcanic zone	46,000 BP	Fall deposit	Silty sand	5.65	NR	1.13	0.41	[28]
Hokkaido	Shikotsu volcanic zone	8700–9200 BP	Fall deposits	Poorly graded gravel	2.9	1.29	NR	0.33	[114]
	Shikotsu volcanic zone			Poorly graded gravel	3.3	1.3		0.3
	Shikotsu volcanic zone			Well-graded sand	2.2	1.33		0.41
	Shikotsu volcanic zone			Poorly graded sand	2.4	1.19		0.35
Sakotsu	Shikotsu volcanic zone	40,000 BP	Flow deposit	Well-graded with non-plastic fines	1.74	2.39	NR	0.98	[123]
Kyushu	Mt Aso Caldera	31,000 BP	Flow deposit	Silty sand	NR	2.3	1.13	0.6	[127]
Kumamoto	Mt Aso Caldera	30,000 BP	Fall deposit	Well-graded sand	3.84	2.47	NR	0.51	[110]
Kagoshima	Sakurajima volcano	11,000 BP	Fall deposit	Well-graded sand with fines	1.551	2.307	1.45	1.05	[58,128]
Ebino	Kakuto Formation	NR	Flow and fall deposits	Silty loam	1.664	2.38	1.26	0.89	[113]
Hokkaido, Japan	Shikotsu volcanic zone	15,000 to 25,000 BP	Flow and fall deposit	Silty sand	1.4	2.4	1.2	0.9	[129]
Hokkaido	Tarumae volcano	Ta-da 8.7–10 ka	Fall deposit	Medium sand, fine-grained to coarse sand	MG—8.029	2.542, 2.624, 2.627	8.74, 8.91, 1.023	2.81, 5.30, 5.11	[5]
					FG—3.950
					CS—4.141
2 Italy
Campania	Phlegraean fields	32,000 BP	Flow deposit	Sand and silt with traces of clay	1.38	2.49	1.65	1.07	[13,109]
	Campanian region Mt. Vesuvius	3760 BP	Fall deposit	Well-graded granular material	2.23	2.66	1.48	0.88	[130]
	Roccamonfina, Phlegraean Fields, and Somma–Vesuvius	17,000 BP	Fall deposit	Silty sand	2.125	2.61	1.55	0.83	[109]
	Vesuvius and Phlegraean	NR	Fall deposit	An equal proportion of sand and gravel	2.21	2.5	NR	0.78	[103]
	Campanian volcanic zone	15,000 BP	Flow deposit	Well-graded sand to silt	2.9	2.61	NR	1.13	[131]
	Somma–Vesuvius volcano	8000 BP	Fall deposit	Well-graded sand with a small clay fraction	2.7	2.64	NR	0.71	[132]
	Mount Lattari ridge	3800 BP	Fall deposit	Well-graded sand/gravel	2.75	2.55	NR	0.63	[118]
3 South East Asia
Luzon, Philippines	Mt. Pinatubo	1991	Fall and flow deposit	Well-graded silty sand	0.92	2.71	1.1	NR	[22,42]
Tagaytay, Philippines	Taal volcano	710–1120 BP	Fall deposit over the flow deposits	Silty clay	NR	NR	0.95	NR	[3]
Indonesia	Mt. Merapi	11,792 ± 90 BP	Undifferentiated fall and flow deposit	Poorly graded sand with silt and gravel	3.22	2.28	1.16	0.54	[124,133,134]
Indonesia	Mt. Tandikat	0.08 ± 0.02 Ma	Fall deposit	Well-graded sand	2	2.66	1.5	0.88	[14]
4 South and Central America
Chile	Cordón–Yelcho volcanic complex	16,500 BP	Fall deposit	Sand with traces of clay	5.324	2.65	1.24	0.42	[111,112]
Quito, Ecuador	Pichincha Volcanic Complex	1800 AD	Fall deposit	Silty sand	1.92	2.7	1.71	1.29	[126]
Ecuador	Cotopaxi volcano	4500 BP	Fall deposit	Poorly graded sand with silt and gravel	0.61	2.41	NR	1.18	[125]
El Salvador	Central America Volcanic Arc	1500 BP	Flow deposit	Sandy silt or silty sand	1.14	2.35	1.4	0.91	[135]
San Salvador	Volcán San Salvador	430 ± 20 AD	Flow deposit with low permeable fine ash	Sandy silt or silty sand	0.98	2.4	1.33	NR	[121]
Eastern Quito, Ecuador	Cangahua	12,000 to 50,000 BP	Flow deposit	Fine sand with silt	1.03	2.58	1.5	1.1	[136]
Santiago	Maipo Volcanic complex	450,000 60,000 BP	Flow deposit	Silty sand	NR	2.33	1.3	1.1	[15]
Arequipa, Peru	El Misti volcano	2030 BP	Flow deposit	Silty sand	NR	2.6	NA	1.25	[6]
Mexico	Michoacán–Guanajuato Volcanic Field	1.42 ± 0.12 to 0.33 ± 0.04 Ma	Fall deposit	Well-graded sand to gravel with silt	0.48	NR	1.38	NA	[137]
5 Others (New Zealand and Turkiye)
New Zealand	Taupo Ignimbrite	1820 ± 150 BP	Fall deposit	Well-graded medium to coarse sand	2.43	2.35	1.56	0.94	[18]
New Zealand	Tuhua tephra (Mayor Island)	7600 BP	Fall deposit	Non-plastic Sandy silts	2.18	2.41	NR	NR	[104]
Isparta, Turkiye	Gölcük volcano	24,000 ± 2000 BP	Fall deposit	Silty sand	0.85	2.6	1.6	1.24	[101]

Note: NR—not reported.

Table 3. Summary of pumiceous deposits’ physical parameters in different locations.

1 Japan
Region	Erupted from	Year of Origin	Type of Material	Soil Classification	Void Ratio	Specific Gravity	Total Density	Dry Density	Cited from
Hokkaido	Shikotsu volcanic zone	46,000 BP	Fall deposit	Silty sand	5.65	NR	1.13	0.41	[28]
Hokkaido	Shikotsu volcanic zone	8700–9200 BP	Fall deposits	Poorly graded gravel	2.9	1.29	NR	0.33	[114]
	Shikotsu volcanic zone			Poorly graded gravel	3.3	1.3		0.3
	Shikotsu volcanic zone			Well-graded sand	2.2	1.33		0.41
	Shikotsu volcanic zone			Poorly graded sand	2.4	1.19		0.35
Sakotsu	Shikotsu volcanic zone	40,000 BP	Flow deposit	Well-graded with non-plastic fines	1.74	2.39	NR	0.98	[123]
Kyushu	Mt Aso Caldera	31,000 BP	Flow deposit	Silty sand	NR	2.3	1.13	0.6	[127]
Kumamoto	Mt Aso Caldera	30,000 BP	Fall deposit	Well-graded sand	3.84	2.47	NR	0.51	[110]
Kagoshima	Sakurajima volcano	11,000 BP	Fall deposit	Well-graded sand with fines	1.551	2.307	1.45	1.05	[58,128]
Ebino	Kakuto Formation	NR	Flow and fall deposits	Silty loam	1.664	2.38	1.26	0.89	[113]
Hokkaido, Japan	Shikotsu volcanic zone	15,000 to 25,000 BP	Flow and fall deposit	Silty sand	1.4	2.4	1.2	0.9	[129]
Hokkaido	Tarumae volcano	Ta-da 8.7–10 ka	Fall deposit	Medium sand, fine-grained to coarse sand	MG—8.029	2.542, 2.624, 2.627	8.74, 8.91, 1.023	2.81, 5.30, 5.11	[5]
					FG—3.950
					CS—4.141
2 Italy
Campania	Phlegraean fields	32,000 BP	Flow deposit	Sand and silt with traces of clay	1.38	2.49	1.65	1.07	[13,109]
	Campanian region Mt. Vesuvius	3760 BP	Fall deposit	Well-graded granular material	2.23	2.66	1.48	0.88	[130]
	Roccamonfina, Phlegraean Fields, and Somma–Vesuvius	17,000 BP	Fall deposit	Silty sand	2.125	2.61	1.55	0.83	[109]
	Vesuvius and Phlegraean	NR	Fall deposit	An equal proportion of sand and gravel	2.21	2.5	NR	0.78	[103]
	Campanian volcanic zone	15,000 BP	Flow deposit	Well-graded sand to silt	2.9	2.61	NR	1.13	[131]
	Somma–Vesuvius volcano	8000 BP	Fall deposit	Well-graded sand with a small clay fraction	2.7	2.64	NR	0.71	[132]
	Mount Lattari ridge	3800 BP	Fall deposit	Well-graded sand/gravel	2.75	2.55	NR	0.63	[118]
3 South East Asia
Luzon, Philippines	Mt. Pinatubo	1991	Fall and flow deposit	Well-graded silty sand	0.92	2.71	1.1	NR	[22,42]
Tagaytay, Philippines	Taal volcano	710–1120 BP	Fall deposit over the flow deposits	Silty clay	NR	NR	0.95	NR	[3]
Indonesia	Mt. Merapi	11,792 ± 90 BP	Undifferentiated fall and flow deposit	Poorly graded sand with silt and gravel	3.22	2.28	1.16	0.54	[124,133,134]
Indonesia	Mt. Tandikat	0.08 ± 0.02 Ma	Fall deposit	Well-graded sand	2	2.66	1.5	0.88	[14]
4 South and Central America
Chile	Cordón–Yelcho volcanic complex	16,500 BP	Fall deposit	Sand with traces of clay	5.324	2.65	1.24	0.42	[111,112]
Quito, Ecuador	Pichincha Volcanic Complex	1800 AD	Fall deposit	Silty sand	1.92	2.7	1.71	1.29	[126]
Ecuador	Cotopaxi volcano	4500 BP	Fall deposit	Poorly graded sand with silt and gravel	0.61	2.41	NR	1.18	[125]
El Salvador	Central America Volcanic Arc	1500 BP	Flow deposit	Sandy silt or silty sand	1.14	2.35	1.4	0.91	[135]
San Salvador	Volcán San Salvador	430 ± 20 AD	Flow deposit with low permeable fine ash	Sandy silt or silty sand	0.98	2.4	1.33	NR	[121]
Eastern Quito, Ecuador	Cangahua	12,000 to 50,000 BP	Flow deposit	Fine sand with silt	1.03	2.58	1.5	1.1	[136]
Santiago	Maipo Volcanic complex	450,000 60,000 BP	Flow deposit	Silty sand	NR	2.33	1.3	1.1	[15]
Arequipa, Peru	El Misti volcano	2030 BP	Flow deposit	Silty sand	NR	2.6	NA	1.25	[6]
Mexico	Michoacán–Guanajuato Volcanic Field	1.42 ± 0.12 to 0.33 ± 0.04 Ma	Fall deposit	Well-graded sand to gravel with silt	0.48	NR	1.38	NA	[137]
5 Others (New Zealand and Turkiye)
New Zealand	Taupo Ignimbrite	1820 ± 150 BP	Fall deposit	Well-graded medium to coarse sand	2.43	2.35	1.56	0.94	[18]
New Zealand	Tuhua tephra (Mayor Island)	7600 BP	Fall deposit	Non-plastic Sandy silts	2.18	2.41	NR	NR	[104]
Isparta, Turkiye	Gölcük volcano	24,000 ± 2000 BP	Fall deposit	Silty sand	0.85	2.6	1.6	1.24	[101]

Note: NR—not reported.

Pumiceous deposits commonly contain quartz, feldspar, plagioclase, and minor traces of magnetite, biotite, hornblende, hypersthene, and hematite. Natural factors influence the mineral compositions in pumice. A comprehensive study conducted by Colombo et al. [4] on volcanic soils of south-central Italy (Mount Gauro and Lake Vico) focused on the mineralogical changes within pumiceous deposits under Mediterranean climates. Their analysis revealed significant findings: the redistribution of fine materials (silt) onto pumice grains, indicating low amounts of short-range aluminosilicates within non-allophanic soils (vitreous); rapid weathering of biotite and sanidine, suggesting the presence of more permeable soil; the presence of zeolites, signifying an early stage of weathering; and negligible halloysite content, characterising the deposits as young. Considering the presence of key minerals in pumiceous deposits will be important for analysing the weathering features and their influence on the geotechnical properties of the deposit.

4.4. Shearing Characteristics

For engineering projects on sites dominated by pumiceous deposits, it is essential to evaluate the behaviour of pumiceous deposits when subjected to shearing. The distinctive features of pumiceous deposits are that they are highly crushable and compressible due to their vesicular nature [97,116,122,138,139], which is considered problematic from an engineering and construction viewpoint. Researchers [12,15,28,104,140] in the past few decades have investigated the shearing behaviour of pumiceous soil. Tests such as single-particle crushing, direct shear, triaxial (K₀ compression, monotonic, and cyclic), and torsional shear have been employed to analyse pumiceous soil behaviour.

Shearing of pumiceous deposits is often controlled by the contact between pumice particles. The interface contact properties of pumice particles are analysed using single-particle crushing strength tests. The behaviour of pumice particles of varying sizes was studied by various authors, such as He and Senetakis [141] (gravel-sized pumice), Orense and Pender [11] (pumice of size range 0.6–2.5 mm), and Basoka et al. [142] (retained on 2 mm, 0.85 mm, and 0.425 mm) and the results were compared with those of natural sand. The surface damage, reflected in a typical progressive fracturing load–displacement relation, is similar to the behaviour of silica particles, as noted by Nakata et al. [143]. Even though plots of Young’s Modulus have similar characteristics, the fitted Young’s moduli of the pumice grains were much lower than those of sand and other geomaterials [141]. Additionally, a trend of decreasing crushing strength with increasing particle size is observed in pumice particles. This trend is comparable to that of silica sand; however, the particle-crushing strength of pumice is at least an order of magnitude less than that of silica sand due to the higher crushability of pumice [11]. The study of the micromechanical behaviour of pumice particles provides an understanding of the normal contact and the shearing response of these crushable volcanic sediments. The interaction mechanism should be compared with other shearing tests, such as direct shear, triaxial, or torsion tests under monotonic and cyclic conditions, to accurately characterise pumice.

4.4.1. Direct Shear

The direct shear test is used to assess the shear strength of pumiceous soil [28,90,128,144,145,146,147]. The effective and total friction angle and cohesion values need to be established through consolidated drained shear tests conducted on soils with different particle sizes and relative densities. It is reported that the degree of particle crushing from shearing tests in reconstituted coarse-grained crushable volcanic soil is more significant than that in fine-grained soil, indicating differences in their behaviour during the shearing process [144]. To describe further, the large-grained particles are more irregular in shape and obstruct each other during shearing; this produces higher cohesion. Meanwhile, the fine-grained material has more contact points during shearing, and the grain-to-grain obstruction is less than the friction that develops between the grains. This creates a lower effective cohesion and higher friction angle than coarse-grained particles and results in lower proportions of particle crushing in the fine-grained particles. The irregular pumice grains also provide an interlocking effect that partially contributes to the dilation behaviour [141] and demonstrates a higher angle of friction (

Table 4. Direct shear strength characteristics from different regions.

Country	Sample Type	Effective Cohesion (kPa)	Effective Angle of Internal Friction (°)	Cited from
Japan	Undisturbed	54.6	34.8	[110]
Italy	Undisturbed	3	36–41	[147]
Korea	Disturbed	NR	38.3 to 44.2	[43]
Japan	Undisturbed	NR	35.5	[148]
	Disturbed	NR	34.08
China	Disturbed (coarse grain)	100–130	25–28	[144]
	Disturbed (fine grain)	<15	35–38
Italy	Fine	5.39–9.61	31.02–61.23	[146]
	Coarse	1.47–8.14	38.49–53.61
	Well graded	32.75	70.12
New Zealand	Undisturbed and reconstituted	12–19	25–27	[149]

Note: NR—not reported.

) and strain-softening behaviour.

Table 4 presents a summary of the shear characteristics of pumiceous sediments from various countries, comparing undisturbed and disturbed samples while highlighting the differences in cohesion and the angle of internal friction. One undisturbed sample from Japan exhibited an effective cohesion of 54.6 kPa and an effective angle of internal friction of 34.8° [110]. Another study reports a friction angle of 35.5° without providing a specific cohesion value [148]. In contrast, Italian undisturbed samples show much lower cohesion, around 3 kPa, but maintain a friction angle between 36° and 41° [147]. New Zealand samples, which include both undisturbed and reconstituted states, demonstrate cohesion values ranging from 12 to 19 kPa and friction angles of 25° to 27° [149].

Shifting the focus to disturbed samples, those from South Korea reveal a notable range in the effective angle of internal friction, from 38.3° to 44.2°, although cohesion was not considered in the analysis [43]. In China, disturbed samples differentiate between coarse-grained and fine-grained sediments. Coarse-grained samples exhibit a high cohesion range of 100 to 130 kPa but have a lower friction angle of 25° to 28°. In contrast, fine-grained samples show less than 15 kPa in cohesion and a friction angle between 35° and 38° [144].

The data indicate a consistent trend across different regions and sample types: pumiceous sediments, regardless of disturbance state, generally exhibit high internal friction angles. The test results support prior research findings that the angularity and rough surface texture of pumice grains can enhance interparticle interlocking and, thus, resistance to shearing. Notably, coarse-grained samples demonstrate higher cohesion values due to the increased interlocking effects among larger particles, as observed in the Chinese samples, with cohesion values reaching up to 130 kPa. This enhanced interlocking contributes to a more stable and cohesive structure, ultimately increasing the sediment’s overall resistance to deformation.

However, there are some anomalies worth noting. The friction angle values for Italian fine and coarse samples display a broad range, with fine particles reaching up to 61.23° and coarse particles reaching up to 53.61° [146]. This wide range suggests significant variability in the mechanical behaviour of these samples, potentially influenced by factors such as particle size distribution and mineral composition.

Bilotta et al. [147] compared the shear strength values of undisturbed and remoulded pumiceous samples and noticed that the shear strength of remoulded samples is higher than that of the undisturbed samples, which is likely related to the reduction in porosity due to remoulding. In situ direct shear testing was conducted by Hashimoto et al. [27], the results showing a marked difference in the behaviour of undisturbed and reconstituted pumiceous samples during consolidation and shearing due to the poor reproducibility of reconstituted samples. A similar in situ apparatus with different dimensions was used by Yamamoto et al. [128], who observed higher friction angles for low density soils and shear stress increases with the increase in shear displacement and normal stress [128]. To determine the critical state shear strength of unsaturated pyroclastic deposits, pumiceous sediments with varying moisture contents (up to full saturation) have been tested and analysed [66,145,150,151]. The test results demonstrate an abrupt decrease in shear strength near saturation [150]. In addition, loading of the sediments causes crushing of pumice particles and induces variation in drained shear strength [151].

Estimating the shear behaviour of crushable pumiceous sediments requires employing suitable testing methods beyond direct shear tests. Francesco et al. [90] disregarded the results of direct shear tests due to unreliable outcomes arising from the crushing phenomena. However, numerous researchers [28,90,128,144,145,146,147] still utilise direct shear tests as one of the tools to assess shear behaviour. Consequently, additional shear testing methods should be adopted to accurately analyse the shear behaviour of pumiceous deposits under different loading conditions.

4.4.2. Triaxial

Triaxial tests are widely regarded as one of the most efficient and effective techniques for simulating the in-situ stress conditions of soil. Various loading conditions, like K₀ compression, monotonic, and cyclic, have been applied to define the behaviour of pumice sand and pumiceous soil [17,152,153,154]. Kikkawa et al. [153] investigated the K₀ compression test to distinguish the compression behaviour of loose and dense pumice sand and found little difference between relative densities (Figure 6a). For a similar stress level, pumice sand produces a higher magnitude of strain than Toyoura sand, highlighting the crushable nature of pumice (Figure 6b). Additionally, the stress relaxation of crushable pumice sand can be influenced by its initial density. In contrast, hard-grained sand exhibits the same behaviour regardless of its density. Pender [155] observed that at the end of K₀ compression, initially free-flowing dry pumice sand exhibits apparent cohesion (Figure 6c); this feature is also observed in beach sand [156] due to capillary forces and interlocking effects. As the findings from K₀ compression tests reveal inconsistent behaviours between pumiceous and hard-grained sands, evaluation through additional triaxial tests is essential to further understand the effects of crushing on shearing mechanisms.

Since pumiceous deposits are also associated with areas susceptible to liquefaction [74,157] and earthquake-induced landslides [27,158], researchers have widely analysed the liquefaction potential of soil containing pumice over the past two decades [22,104,105,152]. The liquefaction characteristics of pumice sand and pumiceous soil have been analysed through monotonic [104,159], cyclic [18,139], and post-liquefaction monotonic [160] tests.

Under monotonic undrained loading conditions, the stress–strain relationship is influenced by factors such as confining pressure, gradation, and relative density. For pure pumice with relative densities ranging from low to high (26–85%), the behaviour remains consistent at both low and high confining pressures, which indicates that relative density has no significant effect on behaviour under monotonic undrained conditions. At low confining pressures, pure pumice sand exhibits a stiff response at small strain levels and greater dilatancy after phase transformation. At high confining pressures, pure pumice sand shows strain softening at large strain levels [159]. In addition, pure pumice sands do not reach steady-state deformation due to particle breakage (Figure 7a,b).

Pumiceous soils exhibit varied ultimate strengths for loose and dense samples corresponding to specific confining pressures in drained conditions. These strengths are not significantly different from those observed in quartz sand. Despite the comparable strength, the volume change behaviour of pumiceous soils differs markedly from that of quartz sand. Specifically, loose and dense pumiceous soils show contractive behaviour (a decrease in volume), whereas dense quartz sand exhibits dilation and loose quartz sand contraction (Figure 8). Another contrasting aspect is the comparable failure strengths of loose and dense pumice under increasing confining pressure, a behaviour not observed in quartz sand [161].

To better understand the behaviour of pumiceous deposits under seismic conditions, cyclic triaxial and post-cyclic monotonic tests are adopted. Under cyclic shear conditions, the plots of double-amplitude axial strain and residual pore pressure ratio for pumice sand under different cyclic shear stress ratios exhibit a consistent trend, as shown in Figure 9 [11]. Particle breakage within pumice sand influences the rapid development of axial strain during the early stages of cyclic loading; however, relative density does not significantly affect the liquefaction resistance curves (Figure 10), unlike Toyoura sand, where relative density has a pronounced effect [11,138]. Particle crushing gradually stabilises the soil structure in both loose and dense pumice sand samples, resulting in higher cyclic resistance [11]. Despite the lower density of the pumice sand samples in Figure 8, its liquefaction resistance is higher than or like that of Toyoura sand. Moreover, confining pressure and gradation significantly influence the liquefaction resistance of pumice sand, where higher confining pressure and greater fines content result in lower liquefaction resistance, a trend also seen in natural sand [11] (Figure 10). When comparing the liquefaction resistance of pure pumice with other geomaterials, pumice exhibits 1.5 to 2.5 times higher resistance than natural sands despite its lower density (e.g., Figure 9).

Comparing the behaviour of pumiceous sediments with varying grain sizes under monotonic loading conditions reveals common trends and unique characteristics for pumiceous sand [29,105], pumiceous silt with non-plastic fines [104,162], pumiceous gravel-sand [154], and pumiceous silty sand with plastic fines [120]. For pure pumice, the relative density has a less significant effect on the stress–strain curves. However, pumiceous deposits exhibit important effects when subjected to different confining pressures and relative densities, as shown in Figure 11. The most obvious effect is the lack of development of peak response under high confining pressures.

Pumiceous materials tend to approach a critical state with state parameters higher than those of hard-grained sand. For pumiceous sand, both intact and reconstituted specimens exhibit similar stress–strain paths; however, their dilatancy behaviour and excess pore pressure generation differ [27]. In pumiceous sediments, an increase in confining pressure helps to prevent the crushing of material, with a cushioning effect noted in natural pumiceous silt under 100 kPa confining pressures (see Figure 12) [104]. Hyodo et al. [58] investigated the impact of fines content on highly crushable Shirasu volcanic soil under a higher confining pressure of 1000 kPa. They observed a significant difference in the particle distribution after shearing, both in samples with fines and those without fines. The influence of fines content on the undrained behaviour of pumiceous deposits is also reported by Kokusho et al. [163]. Sapporo pumice sand with fines showed contractive behaviour, whereas the sand without fines showed a strain-hardening response. In summary, while various pumiceous sediments show similarities in their response to effective consolidation pressure and critical-state behaviour, differences arise in their dilatancy, pore pressure generation, and strain-hardening/softening behaviours based on the pumice content, presence of fines, and the applied confining pressure.

Pumiceous sand has been shown to exhibit a higher resistance to liquefaction, which is largely attributed to particle breakage and the development of a stable soil skeleton characterised by a linear trend in axial deformation [139]. The cyclic resistance of pumiceous sand increases with the degree of particle breakage as the crushed particles enhance the soil’s stability. Relative density also plays an important role in the liquefaction resistance of pumiceous sand, with results from Asadi et al. [139] highlighting that the effect of relative density is more pronounced in pumiceous sand compared to Toyoura sand (Figure 13). However, Orense et al. [164] noted that while relative density does affect the pure pumice sand’s behaviour, its impact is not as significant as it is for Toyoura sand, possibly due to variations in pumice content. Additionally, the pumice content significantly affects the sample cyclic response, with higher pumice proportions leading to increased cyclic resistance in undisturbed samples [29]. It is also important to note that sample disturbance during retrieval has been shown to impact the cyclic resistance of crushable sediments, as disturbance may cause additional particle crushing and alter the sand’s structural integrity [12,165]. To address these issues, reconstitution methods, such as the under-compaction method proposed by Ladd [166], are often employed to maintain a homogeneous state in a reconstituted specimen, minimising unnecessary particle breakage [29].

In contrast to pumiceous sand, pumiceous silt demonstrates lower resistance to liquefaction under cyclic loading. While pumiceous sand benefits from particle breakage in enhancing its stability, the same does not hold for pumiceous silt. At similar relative densities, pumiceous silt shows lower liquefaction resistance when compared to pure pumice, Shirasu soil, hard-grained sand, and pumice mixed with 30% fines [58,104]. One explanation for this is the fines fraction, which is reported to provide a cushioning effect that prevents significant particle crushing, ultimately resulting in lower liquefaction resistance [104]. Furthermore, relative density has a limited effect on the behaviour of pumiceous silt, as Figure 14 shows that it does not appear to significantly influence the excess pore water pressure (EPWP) or axial strain development under cyclic conditions [30]. The lack of particle crushing in pumiceous silt limits its ability to densify and develop a stable soil structure under cyclic loading.

Additionally, Watabe [162] examined the liquefaction resistance of pumiceous silt with non-plastic fines and found that the degree of compaction, skeletal structure, and saturation conditions of the soil significantly impact its liquefaction resistance. Watabe’s [162] study highlights the need for further research on the effects of compaction in pumiceous soils to better understand its role in enhancing liquefaction resistance. This is particularly important as the skeletal structure formed under different compaction levels may significantly alter the soil’s response to cyclic loading and its overall stability.

Asadi et al. [167] examined the post-liquefaction monotonic response of reconstituted natural pumiceous sand and compared it with hard-grained Toyoura sand. They found that while the effect of relative density on recovery strain (the axial strain required to mobilise 1 kPa of axial stress) is observed in pumice, it is less pronounced than in Toyoura sand. The crushability and angularity of natural pumice, both of which are influenced by its higher initial Young’s modulus and lower recovery modulus compared to hard-grained sand (Figure 15), play a significant role in its behaviour [167]. Stringer [17] conducted similar tests on undisturbed pumiceous soil samples to evaluate the effect of pumice content on the stress–strain behaviour. Unlike hard-grained sands, pumiceous soils do not exhibit issues like necking or localised loosening during cyclic loading. However, significant changes in cyclic resistance are observed in soils with low pumice content (Figure 16), while pumice mixed with cohesive soils shows reduced strength and stiffness [168]. Studies on the post-cyclic deformation of pumiceous deposits [12,17,167,168] emphasise the need to compare pumice content, geological origin, and deposit age to consider the effects of particle crushing under cyclic loading.

4.4.3. Other Laboratory Tests

Other laboratory tests, such as bender element [68], undrained cyclic torsion [11,71], and monotonic torsion [71], have also been performed on pumiceous sediments. Kiyota et al. [71] studied the liquefaction behaviour of pumiceous deposits under cyclic torsional simple shear and identified that flow failure occurred under nearly constant residual strength. This was attributed to the crushability and structural alteration of pumiceous deposits due to weathering, indicating that the liquefaction resistance of their samples was related to the crushability of pumice.

The crushability of pumiceous soils, influenced by particle size and fines content, plays a critical role in the determination of shear modulus. Orense and Pender [11] found that the strain-dependent normalised shear modulus plot of pumiceous soils varied significantly with fines content—pumiceous soils without fines exhibited more elastic behaviour compared to pumiceous soils with higher fines content. Furthermore, the reduction rate of the shear modulus with increasing confining pressure in pumiceous soils was found to be similar to that of Toyoura sand. Bender element tests also illustrated a decrease in the rate of reduction of shear modulus with an increase in confining pressure, which is more significant than that of Toyoura sand. Lower shear modulus could be related to particle crushing, higher angularity, and a larger range of the difference between maximum and minimum void ratios (e_max − e_min). Test results from monotonic and cyclic shear tests illustrate the stable structure of pumiceous specimens due to crushing, resulting in higher liquefaction resistance [165]. Similar behaviour is also observed in other crushable materials like calcareous sand [169].

In summary, pumiceous soils show significant deviations from typical critical-state behaviour. For example, in K₀ triaxial compression tests, loose and dense pumice sands show varying levels of stress relaxation depending on their initial density. In contrast, hard-grained sands exhibit more uniform stress relaxation that remains consistent, regardless of initial density. Monotonic undrained tests reveal that particle breakage in pumice sands prevents steady-state deformation, a behaviour not commonly seen in hard-grained sands. Additionally, cyclic tests demonstrate that liquefaction resistance increases with pumice content due to the stabilising effect of particle crushing, even though fine-grained pumiceous soils may exhibit lower resistance due to fines content and confining pressure.

Further differences are observed in drained triaxial tests, where pumiceous soils, although displaying strengths comparable to quartz sand, show markedly different volume change behaviours [161]. This contrast is particularly notable when considering the relative density independence of pumice under undrained conditions, as its dilatancy and pore pressure generation are significantly affected by confining pressure and fines content. Other tests, such as torsion and bender element, underscore the importance of particle crushing in shaping the mechanical behaviour of pumiceous sediments, indicating the need for further research to better understand these effects, particularly in relation to seismic performance and cyclic loading.

4.5. Particle Crushing

Pumiceous soils exhibit unique shear strength and deformation behaviour due to their crushable and highly porous nature. As explained above, one of the key factors influencing their mechanical response is particle crushing, which plays a significant role in both coarse- and fine-grained volcanic soils and is a critical factor affecting the geotechnical properties of pumiceous soils. The analysis of the compiled literature presented in Table 5 provides an in-depth exploration of geotechnical characterisation studies across various regions, with a focus on understanding the influence of particle crushing in pumice-rich soils. These studies reveal a complex interaction between soil properties, mechanical behaviour, and the influence of particle breakage phenomena. By examining parameters such as the particle size distribution, specific gravity, void ratio, and maximum dry density, researchers have identified the need for precise approaches to soil characterisation and classification in the presence of crushable particles. Additionally, the impacts of particle crushing reach far beyond simple classification, influencing the core principles of soil mechanics and geotechnical engineering. Observations from drained triaxial tests [115] reveal notable changes in the mobilised friction angles and shear strength characteristics as a result of particle breakage under varying stress conditions. The significance of particle breakage is particularly pronounced in assessments related to liquefaction resistance [23], soil’s seismic response using surface wave analysis [99], and slope stability analyses [28], where the effects coming from initial relative density and confining pressure have been identified.

The table indicates that shearing in pumiceous deposits primarily alters the particle size distribution, as seen in the reduction of particle size [27,154,170,171] and relative breakage [167,172]. Particle breakage can lead to the underestimation of relative density in both field and laboratory soil penetration tests, such as the CPT and SPT [23,173,174]. Studies have also reported a higher slope parameter (λ) associated with the critical-state line can occur [29]. Results for reconstituted samples [12,120,164] vary with higher apparent cohesion intercepts [146] and subtle differences in post-cyclic liquefaction behaviour [160]. Overall, the examination of particle crushing presented in the literature highlights its complex impact on soil properties and engineering behaviour.

Table 5. Summary of particle crushing characteristics in pumice and related sediments.

Country	Test Conducted	Sample Considered	Comments on Particle Breakage Influence	Cited from
New Zealand	Specific gravity and void ratio estimation	Processed pumice sand	Vesicular nature causes overestimated void ratios due to particle breakage.	[116]
New Zealand	Drained triaxial test	Mercer sand (processed)	Crushing alters particle size distribution, increasing volumetric strain.	[115]
Mexico	Soil characterisation (Refraction Micrometer)	Pumiceous silty sand	Particle breakage alters dynamic properties, requiring non-intrusive testing methods.	[99]
Japan	Undrained triaxial test	Undisturbed volcanic sand	Particle breakage in undisturbed samples impacts cyclic loading conditions.	[12]
New Zealand	Liquefaction analysis using field and lab-based methods	Natural pumiceous soil and processed pumice sand	Crushing causes an underestimation of liquefaction resistance during CPT but increases resistance in lab tests.	[23]
New Zealand and Japan	Drained triaxial test	Pumice from Japan (0.15–0.30 mm) and New Zealand (0.05–1.18 mm)	Mobilised friction angle decreases with axial strain. Particle breakage increases with confining pressure but does not reach the critical state.	[175]
Italy	Drained triaxial (isotropic)	Pumice deposits from Pizzo D’Alvano ridge	Crushing begins at low confining pressures (~30 kPa), reducing friction angles.	[13]
Turkey	Uniaxial compressive strength	Volcanic soil from Isparta	Particle crushing caused sample fragmentation during extraction.	[101]
New Zealand	Paleo-liquefaction analysis (radiocarbon dating, trench analysis, shear wave velocity test)	Pumiceous sand (Waikato)	Particle crushing observed correlates with liquefaction potential based on shear wave velocity.	[176]
Germany	Drained and undrained monotonic and cyclic triaxial	Soft-grained pumice (Eifel) and hard-grained ash (Chile)	Particle crushing increases relative density and enhances peak shear strength due to interlocking.	[177]
Italy	Cyclic triaxial test	Reconstituted pumice and volcanic soil	Varied results in reconstituted samples; higher crushing rates than in undisturbed soils.	[120]
New Zealand	Maximum dry density (MDD)	Natural deposits and processed pumice	NZS4402 MDD test leads to significant crushing, altering density estimates.	[117]
New Zealand	Multi-stage triaxial test	Waikato pumice and Toyoura sand	Higher breakage during cyclic loading impacts strength recovery and pore pressure.	[178]
Australia/New Zealand	Particle size distribution (PSD)	Hemipelagic calcareous sand and graded pumice	Particle breakage causes reclassification from ML to SM due to finer grains.	[179]
New Zealand	CPT and cyclic triaxial tests	Edgecumbe pumice-rich soil	Particle breakage reduces CPT liquefaction estimates; lab tests show higher resistance due to structure stabilisation.	[108]
Italy	Drained triaxial tests (isotropic)	Pumice deposits	Higher apparent cohesion was observed due to particle breakage and rearrangement of the soil skeleton.	[146]
New Zealand	Cyclic triaxial tests	Natural pumiceous soil from Waikato and Toyoura sand	Particle crushing enhances interlocking and cyclic resistance.	[180]
Italy/New Zealand	Cyclic triaxial tests	Cervinara (Italy), Rangiriri (NZ), and Toyoura sand	Crushing forms collapsible microstructure, impacting liquefaction resistance.	[105]
New Zealand	Undrained monotonic triaxial tests	Tuhua pumiceous silt and hard-grained sand	Minimal crushing in pumiceous silt compared to coarser pumice.	[104]
Japan	Direct shear tests	Pumiceous sand from Dozou-sawa River	Particle crushing under low confining pressures affects shear strength; reconstituted samples exhibit more crushing.	[27]
Japan	Drained and undrained triaxial tests	Tarumae pumice and normal sand	Particle breakage shifts the critical-state line, impacting the void ratio.	[172]
Japan	Geotechnical analysis of slope failure (multimodal assessment)	Shikotsu pyroclastic deposits	Particle crushing causes volume reduction during shear.	[28]

Table 5. Summary of particle crushing characteristics in pumice and related sediments.

Country	Test Conducted	Sample Considered	Comments on Particle Breakage Influence	Cited from
New Zealand	Specific gravity and void ratio estimation	Processed pumice sand	Vesicular nature causes overestimated void ratios due to particle breakage.	[116]
New Zealand	Drained triaxial test	Mercer sand (processed)	Crushing alters particle size distribution, increasing volumetric strain.	[115]
Mexico	Soil characterisation (Refraction Micrometer)	Pumiceous silty sand	Particle breakage alters dynamic properties, requiring non-intrusive testing methods.	[99]
Japan	Undrained triaxial test	Undisturbed volcanic sand	Particle breakage in undisturbed samples impacts cyclic loading conditions.	[12]
New Zealand	Liquefaction analysis using field and lab-based methods	Natural pumiceous soil and processed pumice sand	Crushing causes an underestimation of liquefaction resistance during CPT but increases resistance in lab tests.	[23]
New Zealand and Japan	Drained triaxial test	Pumice from Japan (0.15–0.30 mm) and New Zealand (0.05–1.18 mm)	Mobilised friction angle decreases with axial strain. Particle breakage increases with confining pressure but does not reach the critical state.	[175]
Italy	Drained triaxial (isotropic)	Pumice deposits from Pizzo D’Alvano ridge	Crushing begins at low confining pressures (~30 kPa), reducing friction angles.	[13]
Turkey	Uniaxial compressive strength	Volcanic soil from Isparta	Particle crushing caused sample fragmentation during extraction.	[101]
New Zealand	Paleo-liquefaction analysis (radiocarbon dating, trench analysis, shear wave velocity test)	Pumiceous sand (Waikato)	Particle crushing observed correlates with liquefaction potential based on shear wave velocity.	[176]
Germany	Drained and undrained monotonic and cyclic triaxial	Soft-grained pumice (Eifel) and hard-grained ash (Chile)	Particle crushing increases relative density and enhances peak shear strength due to interlocking.	[177]
Italy	Cyclic triaxial test	Reconstituted pumice and volcanic soil	Varied results in reconstituted samples; higher crushing rates than in undisturbed soils.	[120]
New Zealand	Maximum dry density (MDD)	Natural deposits and processed pumice	NZS4402 MDD test leads to significant crushing, altering density estimates.	[117]
New Zealand	Multi-stage triaxial test	Waikato pumice and Toyoura sand	Higher breakage during cyclic loading impacts strength recovery and pore pressure.	[178]
Australia/New Zealand	Particle size distribution (PSD)	Hemipelagic calcareous sand and graded pumice	Particle breakage causes reclassification from ML to SM due to finer grains.	[179]
New Zealand	CPT and cyclic triaxial tests	Edgecumbe pumice-rich soil	Particle breakage reduces CPT liquefaction estimates; lab tests show higher resistance due to structure stabilisation.	[108]
Italy	Drained triaxial tests (isotropic)	Pumice deposits	Higher apparent cohesion was observed due to particle breakage and rearrangement of the soil skeleton.	[146]
New Zealand	Cyclic triaxial tests	Natural pumiceous soil from Waikato and Toyoura sand	Particle crushing enhances interlocking and cyclic resistance.	[180]
Italy/New Zealand	Cyclic triaxial tests	Cervinara (Italy), Rangiriri (NZ), and Toyoura sand	Crushing forms collapsible microstructure, impacting liquefaction resistance.	[105]
New Zealand	Undrained monotonic triaxial tests	Tuhua pumiceous silt and hard-grained sand	Minimal crushing in pumiceous silt compared to coarser pumice.	[104]
Japan	Direct shear tests	Pumiceous sand from Dozou-sawa River	Particle crushing under low confining pressures affects shear strength; reconstituted samples exhibit more crushing.	[27]
Japan	Drained and undrained triaxial tests	Tarumae pumice and normal sand	Particle breakage shifts the critical-state line, impacting the void ratio.	[172]
Japan	Geotechnical analysis of slope failure (multimodal assessment)	Shikotsu pyroclastic deposits	Particle crushing causes volume reduction during shear.	[28]

4.6. Engineering Challenges in Pumiceous Deposits

Pumiceous materials present unique challenges in geotechnical design due to their unusual properties, such as high compressibility, crushability, and low specific gravity. Traditional soil models, such as the Mohr–Coulomb and critical-state soil mechanics frameworks, rely on simplified parameters like cohesion, internal friction angle, and void ratio to predict soil behaviour. However, these models often fall short of accurately representing pumiceous soils’ complex, non-linear behaviour, particularly under varying stress conditions. For example, the Mohr–Coulomb model assumes linear shear strength characteristics, which may not capture the strain-softening and crushing effects inherent in pumiceous materials. Similarly, critical-state models can struggle to account for the significant changes in grading and void ratio that occur due to particle breakage. As a result, simplistic design parameters, such as shear strength and stiffness, may not fully reflect the behaviour of pumiceous soils, making it difficult to predict settlement and liquefaction potential.

This complexity necessitates a more advanced approach, such as numerical modelling, which allows for the inclusion of these non-linear and crushable behaviours in design simulations, leading to more reliable predictions. Various researchers [25,181,182] have considered the use of numerical modelling to simulate pumiceous soil behaviour under different loading conditions. Some examples of numerical modelling with varying soil-constitutive models to represent the behaviour of pumiceous deposits are shown in Table 6.

Murashev et al. [182] proposed using raft foundations as a cost-effective ground improvement solution for crushable pumiceous deposits. A time history analysis of soil–foundation interaction was completed using a hardening soil model with small-strain stiffness (HSsmall) and the Mohr-Coulomb model in a finite element analysis. This approach accurately simulated the significant reduction in stiffness and strength of the pumiceous deposits during liquefaction events. Additionally, the effectiveness of deep soil mixing columns in reducing liquefaction susceptibility was validated through 2D finite element modelling, aligning well with outcomes from limit equilibrium analysis [183]. However, there are some limitations in these analyses as a result of not including the influence of pumice content.

In more advanced numerical modelling, Gens et al. [184] demonstrated the effectiveness of the Particle Finite Element Model (PFEM) and Discrete Element Model (DEM) in accurately capturing the contact non-linearities of pumiceous and calcareous sand. Similarly, the failure mechanism of the Sakuraoka (Japan) slope, which is formed from weathered pumice, can be accurately modelled using a strain-softening constitutive model [185]. This approach could represent the strain-softening behaviour of pumice deposits in a 2D analysis [181].

Other numerical analysis techniques, such as the 2D elastoplastic finite element method using the Drucker–Prager failure criterion with the associated flow rule [186], have been instrumental in analysing saturated–unsaturated seepage behaviour on sites with pumice deposits. Numerical modelling to consider the behaviour of pumiceous deposits has also been considered for sinkhole analysis [187], vibration analysis from underground trains [188], and slope stability assessments including the impact of climate change [187,188,189].

Table 6. Examples of soil constitutive models adopted for pumiceous soils in numerical modelling.

Type of Study	Modelling Method	Cited from
Soil–structure interaction in pile analysis	Hardening soil model	[25]
Saturated–unsaturated seepage analysis	UW softening model	[186]
Cone penetration simulation	Tresca constitutive model (PFEM)	[184]
	Apollonian package with upscale rules (DEM)
Dynamic soil–foundation interaction	HS Small and M-C	[182]
Hydromechanical modelling of landslides	Multiphase model of elastoplastic porous media in conjunction with second-order work (Hill’s criteria)	[190]
Stability analysis of underground human-made caves	M-C yield criterion with non-associative flow rule	[187]
Ground propagation of vibrations	Winkler beam model	[188]
Influence of human-made cuts on slope stability	Elastoplastic M-C model	[191]
Effect of climate change on slope stability	Rigid soil skeleton neglecting gas flux	[189]
Flow-like landslides	Soft soil model	[192]
Seismic slope stability	Linear elastic model	[193]
One-dimensional compression test	Linear and Hertz–Mindlin contact model	[26]

Table 6. Examples of soil constitutive models adopted for pumiceous soils in numerical modelling.

Type of Study	Modelling Method	Cited from
Soil–structure interaction in pile analysis	Hardening soil model	[25]
Saturated–unsaturated seepage analysis	UW softening model	[186]
Cone penetration simulation	Tresca constitutive model (PFEM)	[184]
	Apollonian package with upscale rules (DEM)
Dynamic soil–foundation interaction	HS Small and M-C	[182]
Hydromechanical modelling of landslides	Multiphase model of elastoplastic porous media in conjunction with second-order work (Hill’s criteria)	[190]
Stability analysis of underground human-made caves	M-C yield criterion with non-associative flow rule	[187]
Ground propagation of vibrations	Winkler beam model	[188]
Influence of human-made cuts on slope stability	Elastoplastic M-C model	[191]
Effect of climate change on slope stability	Rigid soil skeleton neglecting gas flux	[189]
Flow-like landslides	Soft soil model	[192]
Seismic slope stability	Linear elastic model	[193]
One-dimensional compression test	Linear and Hertz–Mindlin contact model	[26]

The challenges posed by pumiceous deposits in engineering construction projects are highlighted through various case studies. When the Tauranga (NZ) Harbour Bridge crossing of the Stella Passage was under construction, lower-than-expected pile shaft capacities were attributed to lower friction angles and the interlocking nature of pumiceous sand, necessitating the use of driven plugs to enhance friction pile capacity [98]. Pile installation effects such as flexing of piles and the external cutting shoe created an annular gap between the pile and the surrounding sand, contributed in this case to reducing the contact surface area and thus diminishing the pile’s load transfer efficiency. Direct shear tests on pumiceous deposits demonstrate higher friction due to interlocking behaviour (Table 5); however, stiffness parameters can be lower [194].

In Rotorua (NZ), careful consideration was required in relation to the liquefaction susceptibility of pumiceous soils for the construction of a police station building, highlighting the lack of sufficient detail in determining the liquefaction potential of sites where soils have high pumice content [182]. Studies in Hokkaido, Japan, revealed challenges in developing sufficient shaft friction and end-bearing capacity of piles in pumiceous soils. It was suggested from these studies that design methods for skin friction piles in pumiceous and other crushable soils be established [195]. Other infrastructure projects include the seismic design of a highway embankment, completed using finite element modelling [196] and soil–structure interactions at Carmine Bell Tower using the linear dynamic analysis method [197]. Both underscore the need for a thorough site analysis and innovative problem-solving to ensure the resilience of infrastructure projects located on pumiceous deposits.

5. Conclusions

This review summarises the origin, chemical composition, and characteristic geotechnical and soil mechanics properties of pumiceous deposits from different areas of the world. It highlights shortcomings in the current understanding of pumiceous deposits, specifically in relation to particle crushing. The main conclusions of this study are outlined as follows:

• The formation and mineral composition of pumiceous deposits are primarily influenced by geological processes, such as magma composition and eruption dynamics. While the primary mineral components of pumice typically include silica, alumina, and feldspar, the effects of physical and chemical weathering over time can lead to changes, such as the breakdown of pumice into clay minerals. Therefore, analysing the impact of eruption dynamics and weathering on the index and engineering properties of pumiceous sediments is necessary.
• Research on pumiceous deposits is primarily concentrated in Italy, Japan, and New Zealand. However, there is a need for research on the soil mechanics of pumiceous deposits from other parts of the world, such as Chile, Ecuador, Indonesia, and Mexico. This research would enable comparisons of the complex behaviour of crushable sediments across multiple tectonically and volcanically active regions.
• Characterising the index properties of pumiceous deposits involves challenges, especially those associated with sampling and measurement of specific gravity and maximum dry density due to the high void ratio and rough vesicular nature.
• Particle crushing affects the results of penetration tests (such as SPT and CPT), leading to an underestimation of relative density and liquefaction resistance. However, an empirical method based on shear wave velocity produced a good correlation. Limited studies on the shear wave-based method were applied, and more data are required to validate the outcomes.
• The shearing characteristics of pumiceous deposits show some similarities to hard-grained sand, such as the formation of apparent cohesion in direct shear tests and the jigsaw plot observed in single-particle crushing tests. However, key differences are observed in their monotonic behaviour. For example, pumice exhibits stress–strain curves in K₀ compression that are similar for both loose and dense samples, indicating relative density independence under undrained monotonic conditions. Additionally, pumiceous sand’s triaxial test results do not correlate well with critical-state soil mechanics, as they may not reach a steady state due to particle breakage. Further research is also required to understand how plastic fines influence the monotonic shearing behaviour of pumiceous deposits.
• In cyclic loading, pumiceous sediments demonstrate distinct liquefaction characteristics. Liquefaction resistance appears to be affected by particle size and fines content, with higher confining pressure and greater fines content apparently reducing the liquefaction resistance. Pumiceous sands tend to exhibit higher liquefaction resistance compared to Toyoura sand; however, it is not clear if this is a trend in relation to all hard-grained sands. The presence of non-plastic fines significantly impacts cyclic behaviour.
• Due to the complex interaction between particle crushing and geotechnical properties, further research is needed to explore how different stress conditions (e.g., high confining pressures, variable cyclic loads) influence the degree of crushing and the subsequent changes in shear strength, stiffness, and volume change behaviour.
• The challenges for construction on pumiceous deposits highlight the need for further research on soil–structure interaction in areas affected. Moreover, developing a comprehensive soil constitutive model that can accurately simulate particle crushing will improve the understanding of complex soil–structure interaction in areas where pumiceous deposits are found.