发表机构
KU Leuven; LPP-ERM/KMS; École Polytechnique Fédérale de Lausanne（荷语鲁汶大学; 等离子体物理实验室/欧洲核子研究组织比利时分支; 洛桑联邦理工学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究在SOLPS-ITER中校准多种交叉场输运模型，发现恒定扩散系数模型预测性能最佳，而k模型在提供物理描述的同时校准质量相当，并预测了密度扫描中的差异。
AI 中文摘要
交叉场湍流输运仍然是边缘等离子体模拟中最大的不确定性之一，通常通过经验输运系数来近似。在本工作中，我们使用TCV-X21参考案例的测量数据，对SOLPS-ITER中实现的几种交叉场输运模型进行了校准和评估。所考虑的模型范围从传统的恒定扩散系数描述到自洽的k模型，其中异常扩散系数随等离子体条件演化。模型参数通过基于梯度的优化进行估计，通过最小化前向场配置中模拟与实验的上游和偏滤器轮廓之间的差异。校准结果表明，增加自由参数的数量显著提高了与校准数据集的吻合度。然而，这些更灵活的模型在应用于反向场配置时表现出较差的预测能力，表明存在过拟合。相比之下，最简单的恒定扩散系数模型在仅需要少量校准参数的情况下提供了最佳的整体预测性能。k模型实现了与恒定扩散系数模型相当的校准质量，并以相似的精度再现了实验轮廓，同时提供了异常输运空间变化的基于物理的描述。对密度扫描的预测揭示了在校准点附近不明显的差异。k模型预测随着密度增加，分离面附近的输运水平增加，导致上游轮廓更宽，并且与恒定扩散系数模型相比，偏滤器翻转更早发生。所提出的框架为SOLPS-ITER中的模型校准提供了一条系统且高效的途径，并提供了一组校准的k模型参数，可用于未来的研究。
英文摘要
Cross-field turbulent transport remains one of the largest uncertainties in edge plasma simulations and is commonly approximated through empirical transport coefficients. In this work, we calibrate and assess several cross-field transport models implemented in SOLPS-ITER using measurements from the TCV-X21 reference case. The considered models range from conventional constant-diffusivity descriptions to the self-consistent k-model, in which anomalous diffusivities evolve along with plasma conditions. Model parameters are estimated through gradient-based optimization by minimizing discrepancies between simulated and experimental upstream and divertor profiles in forward field configuration. The calibration results show that increasing the number of free parameters substantially improves agreement with the calibration dataset. However, these more flexible models exhibit poor predictive capability when applied to the reversed field configuration, indicating overfitting. In contrast, the simplest constant-diffusivity model provides the best overall predictive performance while requiring only a small number of calibrated parameters. The k-model achieves a calibration quality comparable to the constant-diffusivity model and reproduces the experimental profiles with similar accuracy, while simultaneously providing a physics-based description of the spatial variation of anomalous transport. Predictions for a density scan reveal differences that are not apparent near the calibration point. The k-model predicts increasing transport levels around the separatrix with increasing density, leading to broader upstream profiles and an earlier onset of divertor rollover compared to the constant-diffusivity model. The presented framework provides a systematic and efficient route for model calibration in SOLPS-ITER, and a set of calibrated k-model parameters to be employed in future studies.
