Existing layer-averaged numerical models for turbidity currents have mostly adopted the global minimum time step (GMiTS) for solution updating, which confines their computational efficiency and limits their attractiveness for field applications. This paper presents a highly efficient layer-averaged numerical model for turbidity currents by implementing the combined approach of the local graded-time-step (LGTS) and the global maximum-time-step (GMaTS). The governing equations are solved for unstructured triangular meshes by the shock-capturing finite volume method along with a set of well-balanced evaluations of the numerical flux and geometrical slope source terms. The quantitative accuracy of the model, given reasonably estimated empirical and model parameters (e.g., bed friction, water entrainment, sediment deposition and erosion coefficients), is demonstrated by comparing the numerical solutions against laboratory data of the current front positions and deposition profiles, as well as field data of the current front positions. The improved computational efficiency is demonstrated by comparing the computational cost of the present model against that of a traditional model that uses a GMiTS. For the present simulated cases, the maximum reduction of the computational cost is approximately 80% (e.g., a simulation that cost 1 h before will only require 12 min with the new model).

现有的浑浊流平均层数模型大多采用全局最小时间步长（GMiTS）进行溶液更新，这限制了它们的计算效率并限制了其在现场应用中的吸引力。本文通过实现局部梯度时间步长（LGTS）和全局最大时间步长（GMaTS）的组合方法，提出了一种高效的浊流平均层平均数值模型。对于非结构化三角形网格，控制方程通过捕获有限体积方法以及一组对数值通量和几何斜率源项的均衡评估来求解。给定合理估计的经验和模型参数（例如，床层摩擦力，水分夹带，沉积物沉积和侵蚀系数）的模型的定量精度，通过将数值解与当前前沿位置和沉积剖面的实验室数据以及当前前沿位置的现场数据进行比较来证明这一点。通过将本模型的计算成本与使用GMiTS的传统模型的计算成本进行比较，可以证明提高了计算效率。对于当前的模拟情况，计算成本的最大减少量约为80％（例如，使用新模型在1 h之前进行的模拟仅需要12分钟）。通过将本模型的计算成本与使用GMiTS的传统模型的计算成本进行比较，可以证明提高了计算效率。对于当前的模拟情况，计算成本的最大减少量约为80％（例如，使用新模型在1 h之前进行的模拟仅需要12分钟）。通过将本模型的计算成本与使用GMiTS的传统模型的计算成本进行比较，可以证明提高了计算效率。对于当前的模拟情况，计算成本的最大减少量约为80％（例如，在使用新模型的情况下，之前1小时的模拟仅需要12分钟）。