Water Content Measurement in Expansive Soils Using the Neutron Probe

Capacitance-type methods for measuring soil water content are known to be unreliable in expansive soils, as cracking disrupts the intimate contact between the soil and the measuring device. The neutron probe, which infers water content from the thermalisation of a cloud of neutrons, is potentially less affected by cracking. The effect of cracking on neutron probe measurements was investigated by a series of numerical simulations using an axisymmetric finite element model based on seven-group neutron-diffusion theory. The simulations employed a consistent soil cracking model based on Maryland clay, in which crack volumes are determined from the changes in void ratio in the shrinking bulk soil. The results show that the presence of cracks in a clay soil affects the inferred water content and that measurements affected by air-filled cracking under-predict not only the water content in the uncracked soil peds but also the average water content in the larger cracked soil mass. The reason for this under-prediction is understood by considering the spatial distribution of the thermalised neutrons in the cracked and uncracked soils. The fast neutrons emitted from the source are seen to diffuse preferentially along air-filled cracks, traveling a large distance from the detector before they become thermalised, thus reducing their likelihood of being back-scattered to the detector where they can be counted. The proximity of the first crack to the probe in the ground also affects the measurement. Water-filled cracks are seen to have the opposite (but lesser) effect to air-filled cracks. A comparison of a simple uniform width crack model to a more realistic model in which crack width varies with changing water content shows that the model is sensitive to crack distribution and that the linear calibration expressions that are typically employed for neutron probes are likely to be unreliable in cracked clay soils