The American Petroleum Institute (API) filter press test has been used for decades in the construction industry as part of the quality control regime for bentonite-based excavation support fluids. The industry has carried over the use of this test to polymer fluids despite the lack of published evidence of its suitability for these fluids and the very different mechanisms by which polymer fluids and bentonite slurries achieve excavation support. This paper presents the first systematic investigation of this issue through a combination of laboratory testing and theoretical analysis. The investigation demonstrates the very different behaviours of bentonite slurries and polymer fluids. In contrast to the results for bentonite slurries, API filter press results for polymers are shown to be highly sensitive to the filter paper used. In particular, repeatability testing revealed a substantial variation in the polymer fluid loss rates attributable to three primary factors: (a) the filter paper pore size, (b) filter paper damage resulting from the applied test pressure, (c) apparent 'clogging' of the filter paper pore space. Furthermore, the study demonstrates the poor repeatability of the API filter press test for polymer fluids even when filter papers of the same type are used. Interestingly, analysis of polymer flow with respect to filter paper pore size and the applied pressure showed that the filter papers were behaving as porous media rather than a simple bundle of capillaries; their behaviour could not be modelled using a simple capillary bundle model. Importantly, this finding shows that the filter press may provide a rapid method of assessing the apparent viscosity of polymer fluids in porous media at high shear rates; data which cannot be obtained by rotational viscometry, and which would otherwise require resort to permeameter testing of coarse soils. The investigation demonstrates that the filter press test is not useful for the on-site quality control of polymer fluids but, given the theory presented in the paper, it can be a useful laboratory tool that provides valuable insight into polymer fluid flow behaviour in soils of high hydraulic conductivity, the most challenging soils for polymer fluid support.

Marine clay may experience stiffness degradation and catastrophic failure when subjected to complex ocean dynamic loadings. This can result in instability, destruction, or capsizing of offshore structures. In this study, marine clay was regarded as a non-Newtonian fluid with shear-thinning behaviour, and the mechanism of progressive stiffness degradation during cyclic loading was discussed from the perspective of fluid dynamics. A series of cyclic direct simple shear tests were conducted on undisturbed marine clay obtained from three offshore sites. Further, the stiffness degradation and flow characteristics under different plasticity index (I-P) and cyclic stress ratio (CSR) conditions were investigated and quantified using the stiffness degradation index (delta) and average flow coefficient (kappa), respectively. The results revealed that the decrease in delta with the increasing number of cycles (N) in a semi-log scale can be categorised into three modes: (1) linear (nonfailure), (2) fast-linear-fast (failure), and (3) linear-stable (failure). Consequently, a two-parameter model was proposed to predict the delta of failure marine clay from different sea areas with varying I-P and CSR values. Moreover, with the increase in N, kappa of the nonfailure marine clay increased gradually in a very limited range, thus exhibiting illiquidity characteristics; by contrast, kappa of the failure marine clay exhibited a slow linear-exponential-rapid linear growth pattern, thus indicating a change in liquidity from weak to strong. Finally, a unified model linking the stiffness degradation and flow characteristics of marine clay under different types and conditions was proposed, where kappa at the cyclic failure state (the failure criterion is a double-amplitude shear strain of 15%) was denoted as kappa(f). Evidently, all data points of kappa/kappa(f) similar to delta were distributed in a narrow range, and a virtually negative exponential relationship was observed between kappa/kappa(f) and delta.