A study on the thermal conductivity of compacted bentonites

Thermal conductivity of compacted bentonite is one of the most important properties in the design of high-level radioactive waste repositories where this material is proposed for use as a buffer. In the work described here, a thermal probe based on the hot wire method was used to measure the thermal conductivity of compacted bentonite specimens. The experimental results were analyzed to observe the effects of various factors (i.e. dry density, water content, hysteresis, degree of saturation and volumetric fraction of soil constituents) on the thermal conductivity. A linear correlation was proposed to predict the thermal conductivity of compacted bentonite based on experimentally observed relationship between the volumetric fraction of air and the thermal conductivity. The relevance of this correlation was finally analyzed together with others existing methods using experimental data on several compacted bentonites.

💡 Research Summary

The paper investigates the thermal conductivity (K) of compacted bentonites, a critical parameter for the design of high‑level radioactive waste (HLW) repositories where bentonite acts as a buffer material. Using a commercial hot‑wire thermal properties analyzer (KD2, Decagon Devices), the authors measured K on a series of compacted specimens of three bentonite types: MX80 (Wyoming, USA), Febex (Spain) and Kunigel (Japan). The specimens were prepared by sieving, equilibrating at controlled relative humidity, and then statically compacted in a 50 mm × 70 mm mould to dry densities ranging from 1.4 to 1.8 Mg m⁻³ and water contents from 9 % to 18 %. After compaction, a 1.3 mm‑diameter hole was drilled in the centre of each specimen, the probe was greased, and K was recorded following the ASTM D5334‑00 protocol.

The experimental campaign revealed several systematic trends. First, K increased monotonically with both dry density (ρd) and water content (w) at constant ρd, confirming the well‑known influence of compaction and moisture on heat transfer. Second, a hysteresis effect was observed: specimens that reached a given w by drying exhibited higher K than those that reached the same w by wetting, which the authors attribute to microstructural changes (reduction of air pores and improved solid‑solid contacts) during drying. Third, when K was plotted against various volumetric fractions—degree of saturation (Sr), solid fraction (Vs/V), water fraction (Vw/V), and air fraction (Va/V)—only Va/V displayed a clear linear relationship: K decreased linearly as Va/V increased. This finding aligns with the physical fact that air’s thermal conductivity (≈ 0.025 W m⁻¹ K⁻¹) is orders of magnitude lower than that of water (0.57 W m⁻¹ K⁻¹) and the mineral matrix (≈ 2 W m⁻¹ K⁻¹).

To assess the predictive capability of existing models, the authors applied three widely used correlations:

1. Johansen (1975) – a semi‑empirical method that requires the thermal conductivity of the solid phase (Ks) calculated from quartz content. Using the quartz fractions reported for each bentonite, the model systematically over‑predicted K for MX80 and Febex (by > 20 %) while performing better for Febex (a = 1.11, b = 0.19).

2. De Vries (1963) – a weighted‑average model that incorporates shape factors for solids and air. By calibrating Ks to match the most saturated sample for each bentonite (Ks ≈ 1.5–1.9 W m⁻¹ K⁻¹), the model achieved much lower errors (a ≈ 1.10 for MX80 and Febex, a = 1.02, b = 0.08 for Kunigel).

3. Sakashita‑Kumada (1998) – a semi‑theoretical model that accounts for micro‑ and macro‑pores. Using parameters derived by Ould‑Lahoucine et al. (2002), the model over‑estimated K for MX80 (present work) and Febex but performed well for the MX80 data of Kahr & Müller‑Vonmoos (1982) and for Kunigel.

Building on the clear linear dependence of K on Va/V, the authors proposed a new, highly parsimonious correlation:

\