A converged CFD solution is a starting point. Numerical checks and comparison with physical evidence answer different questions about a model.

DEEB-FLOW concept illustration of a heat-transfer experiment and numerical comparison
交互式概念示意图:沿受热管道的局部对流传热系数,取自 DEEB-FLOW 案例。横向移动指针可将网格从 4 个单元细化到 32 个单元:计算得到的阶梯状曲线逐步收敛到与网格无关的曲线(虚线),读数栏给出网格收敛指数。与实测点及其不确定度误差棒之间残留的差距,正是验证所要评判的内容。数值为示意性质。

本文暂无您所用语言的版本,以下为英文内容。

Verification examines the numerical calculation

Verification asks whether the numerical solution represents the mathematical model with sufficient accuracy for the intended use. Residual behaviour, conservation balances and the stability of engineering outputs are relevant checks. A small residual alone does not establish that pressure loss, heat transfer or another decision variable is insensitive to numerical choices.

Mesh sensitivity is about quantities that matter

Refine the mesh systematically and compare the same engineering quantities under consistent physical conditions. Watch where refinement changes the answer, particularly near walls, jets, interfaces and steep gradients. Where appropriate, a grid-convergence study can help estimate discretisation uncertainty. Agreement between two meshes is useful evidence, but is not experimental validation.

Validation compares the model with physical evidence

Validation considers how well the chosen physical model describes the real system for its intended use. Compare with relevant measurements under matching operating conditions, with measurement uncertainty in view. Matching a single measured quantity does not prove that every internal field is correct or that the model is reliable outside the conditions examined.

Report the limits alongside the conclusion

Record assumptions, boundary conditions, numerical checks and the conditions covered by the evidence. Separate numerical uncertainty from uncertain inputs and modelling limitations. If relevant measurements are unavailable, say so and define what a future test should measure. The useful outcome is a decision supported by stated evidence, not an unqualified claim that the model is validated.

来源与延伸阅读

相关服务

模型审核与定制研究

明确模型能够支撑哪些结论、不确定性存在于何处,以及还需核实哪些内容。

了解服务
全部文章