AgentFEM-Pipe-ROM / POWER_CASES.md
HaomingLuo's picture
Pipe Assembly Lab v1.3: four power-piping research sections, FE operators and validation
0e50fc0 verified
|
Raw History Blame Contribute Delete
4.89 kB

Power-piping research cases · v1.3

Four service families provide reusable solid-FE component operators, rather than attaching service names to the old small-pipe library. Each family has three straight lengths (3D, 5D, 8D) and three 90° bends (R/D = 1.5, 2, 3). Each bend also contains a 2D tangent at both ends. Layouts are researcher-designed.

Research family OD × wall / mm E / GPa α / µm/(m·K) Example ΔT / K
Main steam / 主蒸汽 450 × 60 170 13 546
Hot reheat / 再热热段 800 × 30 170 13 546
Cold reheat / 再热冷段 800 × 20 190 12.5 330
High-pressure feedwater / 高压给水 450 × 50 195 12.5 280

Provenance and assumptions

The main-steam cross-section only derives from Gao & Huo, Feasibility Research on Ultra-supercritical 350 MW Unit and Analysis of Main Equipment Parameter, Table 9: 330 mm bore and 60 mm calculated wall, excluding corrosion allowance. The derived OD is 450 mm. It is not a purchased/as-built pipe size. Source: https://doi.org/10.16516/j.gedi.issn2095-8676.2019.04.007

The other three cross-sections are author-selected benchmark geometries. All tabled effective E and α values, temperature rises, layouts, springs, damper settings and forces are benchmark assumptions, not values extracted from that paper, certified grade-specific material data, or measured plant data. Density is 7850 kg/m³ and Poisson ratio 0.3. ΔT is a uniform imposed wall temperature rise, not an absolute operating temperature or a transient history. Material constants remain fixed when ΔT is varied.

The service taxonomy follows main steam, hot reheat, cold reheat and main feedwater. Context: National Energy Administration, https://www.nea.gov.cn/2011-02/14/c_131071160.htm

Models and records

  • 24 new components, 35,280 displacement DOFs each; Q2 displacement on a Q1 hexahedral geometry, 24 axial × 24 circumferential × 2 radial cells.
  • 12 rigid mechanical port DOFs plus 12 fixed-interface modes per component.
  • 108 assembly records: four families × three layouts × three support choices × three temperature rises. Labels come from the assembled component ROM.
  • 12 harmonic sweeps: four families × three support choices, with a 1000 N peak Y force at the final joint, assumed 2% modal damping, and a 10,000 N·s/m grounded Y damper at that same joint. Their excitation frequency is f1.
  • UI initial layouts use springs at joints 3 and 5; they are editable examples, not identical to every dataset layout or an optimized support design.
  • Original 48 components and 512 assembly records remain intact.

Mass and unconstrained axial expansion are checked analytically. Python/SciPy and browser assembly are compared for all 108 records. Harmonic response is checked against an independent complex solve, including 12/24 system-mode retention. Four R/D=3 bends are compared against 30 × 32 × 3 meshes (81,984 DOFs). The refinement check covers isolated bends, not every route, stress, or higher mode. Reports state each metric explicitly.

Scope

These are dry-pipe, small-displacement linear elastic models. No pressure stiffening, fluid/insulation mass, spring preload, gaps/friction, creep, fatigue, thermal gradients or piping-code acceptance calculation is included. A viscous damper is not a hydraulic snubber. Displayed response is centreline motion, not wall stress. This release supports reproducible geometry/support studies and operator reuse; it is not a plant digital-twin calibration or design sign-off.

Reproduce

In the downloaded dataset repository, first run:

mkdir -p data reports
ln -s ../components data/components
ln -s ../assemblies data/assemblies
ln -s ../dynamics data/dynamics

These relative links recreate the development layout without copying matrices. Skip a link if it already exists. With AgentFEM/DOLFINx, run python source/validate_power.py and python source/check_power_mesh.py. To regenerate component operators, run python source/generate_power.py; existing components are reused, so use a separate workspace for a clean rebuild. With Node.js: node source/check_power.cjs. With NumPy/SciPy: python source/check_power_harmonic.py. For a compact model-only example, run node example_power.cjs from the downloaded model repository. No solid-FE environment is needed for that example. The data/model repositories include the required operator data and source; the Space contains only compact browser operators and needs no Python backend.

中文说明

本版增加的是可复用、可验证的四大管道研究组件。主蒸汽截面有公开计算尺寸 来源;其他截面和工况为研究设定。先研究管路走向、支撑位置和刚度对热位移、 接口载荷与振动的影响,不将结果用作现场管道的安全判定。