Abstract Computational fluid dynamics (CFD) simulations of large-scale furnaces or reactors for thermal conversion of solid fuels remains challenging partially due to the high computational cost related to the particle sub-models. Owing to the thermally thick nature, it is particularly expensive to simulate the conversion of large fuel particles such as biomass particles. To address this issue, a fast-solving particle model was developed in this work with special attention to the computational efficiency. The model spatially discretizes a fuel particle in one homogenized dimension. The conversion process of the fuel particle is treated as a reactive variable-volume one-dimensional transient heat conduction problem. The model also utilizes several features that are typically found in sharp interphase-based models to reduce the computational cost. Validation of the model was carried out by comparing with experimental results under both pyrolysis and combustion conditions. The accuracy and computational efficiency of the model was thoroughly examined by varying the degrees of temporal and spatial discretization. It was found that the model well predicted pyrolysis and combustion of a single biomass particle within a broad range of temporal and spatial discretization. The time used to simulate the conversion of a biomass particle using the developed model can be more than one order of magnitude smaller than the conversion process itself. It was also revealed that a well-predicted conductive heat transfer inside the particle is essential for a precise simulation of the drying and devolatilization process. The char conversion process, however, is less sensitive to the external heat transfer as it is mainly controlled by the mass diffusion process. Further studies showed that a time step of 1 × 10 − 3 s and a spatial discretization of 20 cells were sufficient for simulating the conversion of typical fuel particles in grate-fired and fluidized-bed furnaces. We also demonstrated that when the particle model was implemented in a CFD solver, only 2.2% of computational overhead was introduced by the model. As the model can efficiently employ fixed time stepping, optimal load balancing during parallel computing of many simultaneous conversion processes becomes trivial. This performance opens up new possibilities for treating fuel polydispersity in Eulerian CFD simulations of biomass conversion.

中文翻译：

---

用于生物质热化学转化的快速求解粒子模型

摘要 用于固体燃料热转化的大型熔炉或反应器的计算流体动力学 (CFD) 模拟仍然具有挑战性，部分原因是与粒子子模型相关的高计算成本。由于热厚性质，模拟大燃料颗粒（例如生物质颗粒）的转化特别昂贵。为了解决这个问题，在这项工作中开发了一个快速求解的粒子模型，特别关注计算效率。该模型在一个均质化维度中空间离散化燃料颗粒。燃料颗粒的转化过程被视为反应性变体积一维瞬态热传导问题。该模型还利用了一些通常在基于相间的尖锐模型中发现的特征来降低计算成本。通过与热解和燃烧条件下的实验结果进行比较，对模型进行了验证。通过改变时间和空间离散化程度，彻底检查了模型的准确性和计算效率。发现该模型在广泛的时间和空间离散化范围内很好地预测了单个生物质颗粒的热解和燃烧。使用开发的模型模拟生物质颗粒转化所用的时间可能比转化过程本身小一个数量级以上。研究还表明，粒子内部良好预测的传导热传递对于精确模拟干燥和脱挥过程至关重要。然而，char 转换过程，对外部传热不太敏感，因为它主要由质量扩散过程控制。进一步的研究表明，1 × 10 − 3 s 的时间步长和 20 个单元的空间离散化足以模拟典型燃料颗粒在炉排燃烧和流化床炉中的转化。我们还证明，当粒子模型在 CFD 求解器中实现时，模型仅引入了 2.2% 的计算开销。由于该模型可以有效地采用固定时间步长，因此许多同时转换过程的并行计算期间的最佳负载平衡变得微不足道。这种性能为在生物质转化的欧拉 CFD 模拟中处理燃料多分散性开辟了新的可能性。进一步的研究表明，1 × 10 − 3 s 的时间步长和 20 个单元的空间离散化足以模拟典型燃料颗粒在炉排燃烧和流化床炉中的转化。我们还证明，当粒子模型在 CFD 求解器中实现时，模型仅引入了 2.2% 的计算开销。由于该模型可以有效地采用固定时间步长，因此许多同时转换过程的并行计算期间的最佳负载平衡变得微不足道。这种性能为在生物质转化的欧拉 CFD 模拟中处理燃料多分散性开辟了新的可能性。进一步的研究表明，1 × 10 − 3 s 的时间步长和 20 个单元的空间离散化足以模拟典型燃料颗粒在炉排燃烧和流化床炉中的转化。我们还证明，当粒子模型在 CFD 求解器中实现时，模型仅引入了 2.2% 的计算开销。由于该模型可以有效地采用固定时间步长，因此许多同时转换过程的并行计算期间的最佳负载平衡变得微不足道。这种性能为在生物质转化的欧拉 CFD 模拟中处理燃料多分散性开辟了新的可能性。我们还证明，当粒子模型在 CFD 求解器中实现时，模型仅引入了 2.2% 的计算开销。由于该模型可以有效地采用固定时间步长，因此许多同时转换过程的并行计算期间的最佳负载平衡变得微不足道。这种性能为在生物质转化的欧拉 CFD 模拟中处理燃料多分散性开辟了新的可能性。我们还证明，当粒子模型在 CFD 求解器中实现时，模型仅引入了 2.2% 的计算开销。由于该模型可以有效地采用固定时间步长，因此许多同时转换过程的并行计算期间的最佳负载平衡变得微不足道。这种性能为在生物质转化的欧拉 CFD 模拟中处理燃料多分散性开辟了新的可能性。