research: prepare B1 residual contract and physics branch findings
This commit is contained in:
1 parent
f9a5b4eafe
commit
0a60f5d6fd
7 files changed
+493
No files matched your search
@@ -0,0 +1,245 @@
|
||||
{
|
||||
"status": "READ_ONLY_METADATA_PREPARATION",
|
||||
"unix": 1794886604.6362166,
|
||||
"files": [
|
||||
{
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/contract.json",
|
||||
"bytes": 3377828,
|
||||
"sha256": "eb2d68b2f29135ecdc8ccf48165ae95cec9123183e8d03cb04ab9ad1174b3306"
|
||||
},
|
||||
{
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/models-terminal.json",
|
||||
"bytes": 115493,
|
||||
"sha256": "fe2892928610e7720b8b0698cf322cf30d92ed3ed6d53790980cd762aa0fca7c"
|
||||
},
|
||||
{
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/splits.json",
|
||||
"bytes": 246443,
|
||||
"sha256": "534b2a25f92dbac2335b6282e1dfe0bb8ecdbc8db55bdfc24c7cffa83f9e931b"
|
||||
},
|
||||
{
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/preparation-freeze.json",
|
||||
"bytes": 956849,
|
||||
"sha256": "ac7add88b54f0cd75b667cacbb813aac40a23b9a7f47557e635efd7dbbc2d3d0"
|
||||
},
|
||||
{
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/training-contract.json",
|
||||
"bytes": 3181,
|
||||
"sha256": "7efc6d5e8fa6079007f2e137e80f8354e1408839413c63d9b4af443d82086cb9"
|
||||
},
|
||||
{
|
||||
"path": "tools/r4bc_common.py",
|
||||
"bytes": 9645,
|
||||
"sha256": "27543fff987975ece767167f9771e6621c825035a053805604e4a696794dfcbd"
|
||||
},
|
||||
{
|
||||
"path": "tools/r4bc_fit.py",
|
||||
"bytes": 7787,
|
||||
"sha256": "7827c1dac077baf0659a93c8f992c3243285e4f6c0ab31a5f6b3711a148b5247"
|
||||
},
|
||||
{
|
||||
"path": "tools/r4bc_postfit.py",
|
||||
"bytes": 9752,
|
||||
"sha256": "02df42bc8cc9ce553db5dfbc047cf80955ceee5348faa306c37cab05933d9623"
|
||||
},
|
||||
{
|
||||
"path": "tools/gre24_models.py",
|
||||
"bytes": 1819,
|
||||
"sha256": "53b6fee8b700b9b982b06d29908ce2ce23e0c57512c46dbf74b2aade79e3ac3a"
|
||||
},
|
||||
{
|
||||
"path": "tools/gre24_fit.py",
|
||||
"bytes": 8912,
|
||||
"sha256": "63af0305a2c22b524d95ffee50e972848decdd042b1d24f0b3b946d9152f7688"
|
||||
},
|
||||
{
|
||||
"path": "tools/gre24_features_v2.py",
|
||||
"bytes": 11152,
|
||||
"sha256": "7dcd5ac218ffdc1ca149edcd87e8658fb3c59e18c0fb9ce47985c7f1357d2cdf"
|
||||
},
|
||||
{
|
||||
"path": "tools/naive_ordering_math.py",
|
||||
"bytes": 4327,
|
||||
"sha256": "8a71dfb139f58c523077e8e94d7cbe7ad34eb7680b6b2535bfcf7a94001ad86c"
|
||||
}
|
||||
],
|
||||
"reference_fit_costs": [
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 100,
|
||||
"fold": 0,
|
||||
"worker_cpu_seconds": 155.67591421400002,
|
||||
"learning_nodes": 8742060,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0065/output/result.json",
|
||||
"sha256": "c2db06f3f7d50b9aed9f8fce3473c638910e22a8a0db4346988b64fbadcb55d6"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 100,
|
||||
"fold": 1,
|
||||
"worker_cpu_seconds": 183.466841757,
|
||||
"learning_nodes": 9731293,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0132/output/result.json",
|
||||
"sha256": "3cc7e364a2efb161e4054772c408738c5e831045c7a2f6a578be711b4e668d88"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 100,
|
||||
"fold": 2,
|
||||
"worker_cpu_seconds": 186.616576194,
|
||||
"learning_nodes": 9875407,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0199/output/result.json",
|
||||
"sha256": "f1835aa7ee39ee6590486639db5ded7e52faab8e8a9fede0b98b7fcd58fe6b82"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 100,
|
||||
"fold": 3,
|
||||
"worker_cpu_seconds": 150.393601184,
|
||||
"learning_nodes": 8118914,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0266/output/result.json",
|
||||
"sha256": "44b1cf072813edb0962e8a9cbbc305ebb6279e8a476a9853254c2660b9bd1d43"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 100,
|
||||
"fold": 4,
|
||||
"worker_cpu_seconds": 180.15494359800002,
|
||||
"learning_nodes": 9749420,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0333/output/result.json",
|
||||
"sha256": "df29238712181831d8198b8d7c295b7db51a4ed0b73d430b9f571275300d4306"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 50,
|
||||
"fold": 0,
|
||||
"worker_cpu_seconds": 86.940281822,
|
||||
"learning_nodes": 4881636,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0400/output/result.json",
|
||||
"sha256": "2d708e0955800bc11c6411d30a6302ac57409e088b7425cea7bbf2a6d3b823ad"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 50,
|
||||
"fold": 1,
|
||||
"worker_cpu_seconds": 88.841513789,
|
||||
"learning_nodes": 4406392,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0467/output/result.json",
|
||||
"sha256": "5bbb466ac810f7e9ec96f4fd01a0a16291c4499b5b86b2b7a55116c5d6eac087"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 50,
|
||||
"fold": 2,
|
||||
"worker_cpu_seconds": 87.62662574500001,
|
||||
"learning_nodes": 4746521,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0534/output/result.json",
|
||||
"sha256": "1c313ba584494a0cf94abb59bbaf16d1e18e3092f53812488aba374ce309c2d5"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 50,
|
||||
"fold": 3,
|
||||
"worker_cpu_seconds": 83.545909702,
|
||||
"learning_nodes": 4757238,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0601/output/result.json",
|
||||
"sha256": "3dfe0eca09efc38506912907091189e88284de2a72b004a0e2a5c448d93b96ff"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 50,
|
||||
"fold": 4,
|
||||
"worker_cpu_seconds": 61.084871126,
|
||||
"learning_nodes": 3392006,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0668/output/result.json",
|
||||
"sha256": "ab20a73a194a366c0de51c7892b4ba062fd3e812d4ba51ef99709a963924c03b"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 25,
|
||||
"fold": 0,
|
||||
"worker_cpu_seconds": 53.414578368,
|
||||
"learning_nodes": 2858048,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0735/output/result.json",
|
||||
"sha256": "455a95020c1bf529f11d75a297bf8a3c487f575d1aa14df38c0e2f64b2850dc7"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 25,
|
||||
"fold": 1,
|
||||
"worker_cpu_seconds": 45.063439855,
|
||||
"learning_nodes": 2258884,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0802/output/result.json",
|
||||
"sha256": "917f8b1d68612b4e61b7d3761a381cee7cc524e787a72b005ab68dc760445829"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 25,
|
||||
"fold": 2,
|
||||
"worker_cpu_seconds": 29.572757034,
|
||||
"learning_nodes": 1537599,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0869/output/result.json",
|
||||
"sha256": "a5fb0fcaa2b5db970be800a835f7e2be25aab8c5169784fada2db544355146fc"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 25,
|
||||
"fold": 3,
|
||||
"worker_cpu_seconds": 50.053610123999995,
|
||||
"learning_nodes": 2699713,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-0936/output/result.json",
|
||||
"sha256": "b9213e48f55b9bd9035d47e8c89727a614dc6fd43bacd2e743f0b1ce47dfe9a6"
|
||||
},
|
||||
{
|
||||
"protocol": "grouped",
|
||||
"pool": 25,
|
||||
"fold": 4,
|
||||
"worker_cpu_seconds": 31.792714728000004,
|
||||
"learning_nodes": 1748016,
|
||||
"path": "artifacts/r4bc-execution-2026-11-16/jobs/job-1003/output/result.json",
|
||||
"sha256": "56844a0d2fac971e35cd8555e8de7c015478ed55226ee3b9f342289313160af3"
|
||||
},
|
||||
{
|
||||
"protocol": "v4_random",
|
||||
"split": 0,
|
||||
"worker_cpu_seconds": 442.71303165,
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/fits/split-0/GBRT/result.json",
|
||||
"sha256": "1e0c54a791d1a4e93aebf7bf12aba31b91deaeb17df653dd0f334dd27ed1e00b"
|
||||
},
|
||||
{
|
||||
"protocol": "v4_random",
|
||||
"split": 1,
|
||||
"worker_cpu_seconds": 418.26682598,
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/fits/split-1/GBRT/result.json",
|
||||
"sha256": "6aa4c5d461298568297f2f62ac873fcaaf7ba20ae6cfd40c43693297317dcad4"
|
||||
},
|
||||
{
|
||||
"protocol": "v4_random",
|
||||
"split": 2,
|
||||
"worker_cpu_seconds": 465.94808978699996,
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/fits/split-2/GBRT/result.json",
|
||||
"sha256": "5f3b5360e2ab65682ce20651e78d6682df478143663a8a97330f755c73c45dea"
|
||||
},
|
||||
{
|
||||
"protocol": "v4_random",
|
||||
"split": 3,
|
||||
"worker_cpu_seconds": 408.30420073700003,
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/fits/split-3/GBRT/result.json",
|
||||
"sha256": "395f4b92b8d0fb601843fa5831cb95ccc590d50a5219f881f025caaac77522bc"
|
||||
},
|
||||
{
|
||||
"protocol": "v4_random",
|
||||
"split": 4,
|
||||
"worker_cpu_seconds": 392.260725441,
|
||||
"path": "artifacts/gre24-v4-reproduction-2026-11-16/protocol/fits/split-4/GBRT/result.json",
|
||||
"sha256": "b004ac886db580f186cca84b0c1855c7747baf065b657bfbc9e32012abd2bd26"
|
||||
}
|
||||
],
|
||||
"historical_grouped_A_cpu_seconds": 1474.2441792400002,
|
||||
"historical_v4_GBRT_cpu_seconds": 2127.4928735949998,
|
||||
"source_note": "Historical receipts only; not a new fit cost measurement",
|
||||
"labels_read": false,
|
||||
"feature_arrays_read": false,
|
||||
"new_processes_launched": false,
|
||||
"host_precheck": "No active MeltMind research Python process observed; no existing H1/H2 output root found"
|
||||
}
|
||||
@@ -0,0 +1,14 @@
|
||||
{
|
||||
"status": "PASS",
|
||||
"tests": 5,
|
||||
"cpu_seconds": 0.170161,
|
||||
"wall_seconds": 0.1361013029818423,
|
||||
"peak_rss_bytes": 35389440,
|
||||
"project_arrays_read": false,
|
||||
"models_fitted": 0,
|
||||
"scope": "Synthetic residual interface only; clock estimator and topology specification not selected",
|
||||
"code_sha256": {
|
||||
"tools/phase2_residual_math.py": "06609f1351c1e13b82f11a300e39d92a01966f17cbb2b7fd822020140ab570ab",
|
||||
"tools/test_phase2_residual_math.py": "1027274d55fb2ec447b73c9760517b657f8306bb43481f340a399b3e67c34f93"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
# H1-RES:固定EIK跳连残差学习(合同草稿)
|
||||
|
||||
用户已通过TRON明确同意H1/H2尝试;Claude `1794886306.593149`分工:Mold数学规格,Melt合同/实现/唯一执行,Assist事后审计。**当前开始合同和实现准备,无训练启动;等待Mold短规格并完成Claude一次合同主审,不是再次请求用户go。** 本任务使用H1-RES命名,与R4旧H1消融区别。
|
||||
|
||||
后续已实际读到Claude唯一[研究树bf0a6c0f](https://gitea.frank14f.online/Frank14f/MeltMind/src/commit/bf0a6c0f/docs/claude/research-tree.md):B1/B2为当前两个PROMOTED分支,其他OPEN分支先规格/cheap gate,至多两枝同时正式推进。执行顺序沿TRON转达B1合同→B3/B4/B7门→B6/B5;新枝不自动派队列,研究树只由Claude维护。每个终态记录认知、保留反例、剪枝理由和再开条件。旧O1覆盖不足、O2固定配方无增益与图solver资源/数值失败不泛化为数学方向无效。
|
||||
|
||||
## 研究问题与固定对照
|
||||
|
||||
在不增加A的19输入列、不改变同协议训练数据、权重、XGB容量/超参数或测试评分的前提下,把训练目标改为 `r=t_star-c*F`,解码 `t_hat_star=c*F+r_hat`,是否同时改善全场误差、最后位置与顺序?`F`为既存全tet域strict volume CDF,`c`只能来自对应学习池。无因果惩罚、无后处理修正或事后剪裁。不把收益必然归因于新增物理信息:F已在A首列,改变的是固定跳连与学习目标。
|
||||
|
||||
Mold须在同一合同最终版明确:c的具体估计量及全VTU/tet端点口径、各池使用哪些训练角色、F浮点精度与t_star一致性、拓扑指标的定义。未决定部分不静默用默认值;其余接口和冻结规则可先准备。
|
||||
|
||||
## 数据与拟合单元
|
||||
|
||||
- 分组主协议:复用 `artifacts/r4bc-execution-2026-11-16/contract.json` 的552件/204组件、五折、25/50/100%三个固定嵌套学习池和原角色。每格一个新H1-RES XGB,合计15。训练支持域全tet节点、单位权、float32、100树/深度8/学习率0.1/平方损失/seed0,沿A。验证/测试不用于新选参;原A检查点与原A/EIK评分保留。
|
||||
- v4论文随机协议:复用 `artifacts/gre24-v4-reproduction-2026-11-16/protocol/splits.json` 五个315test/314train原划分及615成功/14原失败准备清单;这是五个独立随机split,非五折CV。每split新H1-RES一个,合计5;原GBRT使用train角色(包含其FNN验证子集),此处沿原GBRT而非擅用FNN学习角色。seed沿原split,原全部失败不补抽。
|
||||
- 总计20次新固定配置拟合,无HPO、无旧模型重训、无新v2下载或留验筛查。两协议数字分开,grouped域是探索后复用,不称全新盲测。
|
||||
- F、mass、raw_ids、features、角色与t_set均来自对应版本固定缓存,启动前核metadata与hash。严禁因文件名相同就把v4/v5字段视为相同。原始失败/旧源码完全保留;新实现独立文件和输出根。
|
||||
|
||||
## 数据访问与输出冻结
|
||||
|
||||
训练/时钟只读当前格允许的训练侧缓存目标,逐件记录访问角色及来源hash;不重新解码原始标签。参数固定,无验证驱动择优。负残差合法,不套用“目标非负”检查到残差;还原到达场后保留负值等物理异常并单列,不悄悄裁剪。
|
||||
|
||||
20次拟合全部终态、模型hash和NA先Git冻结,之后预测;预测完整清单及评分代码再Git冻结,最后统一评分。测试预测函数不得读目标字段,旧A/EIK输出能复用时按hash复用。若新增拓扑指标需要全tet场而旧A只有查询预测,须在主审前列明既有模型的额外预测成本与身份,不偷用查询子图冒充真实tet连通。
|
||||
|
||||
## 评分与待核拓扑项
|
||||
|
||||
既有分组主指标:组件等权relative L2;同时保留既定最后位置距离/Rg、Kendall tau-b及其失败/未定义分母。固定query IDs和multiplicity只计一次,原同域比较与组件配对描述性区间沿用。v4先报告原全tet三分母RRMSE、<0.1比例和逐split统计;新增顺序/末端指标不得与grouped混汇。
|
||||
|
||||
Claude提出的拓扑项尚待Mold核正:单浇口分流再会合不一定是湿区H0合并;不能称焊线真值。规格需明确原tet图/复形、子或超水平集、并列值、持久性阈值、空间匹配距离、无事件的分母和计算费用。该项最终定义及primary/secondary定位须在运行前一次主审冻结;不因测试效果改变指标。
|
||||
|
||||
## 工程预算建议(交同一次主审)
|
||||
|
||||
总≤20 worker CPU-h=72000秒,初拟准备/合成/接口3600秒、拟合43200秒、预测/拓扑/评分25200秒,共享累计账,失败CPU计一次;不得自动扩预算。最多4单线程worker,每fit CPU/墙钟各1800秒、RSS8GiB,总RSS36GiB,新增磁盘20GiB,0重试。先grouped fold0/pool100和v4 split0作实际成本检查,成功且余量可容纳全部计划及后处理预留后自动继续,不额外用户go。
|
||||
|
||||
只读历史费用依据:15个旧A worker合计1474.244179 CPU秒;五个v4 GBRT合计2127.492874 CPU秒(旧v4线程数8,worker CPU不是墙钟)。这是参考,不保证残差拟合成本相同。真实第一格成本与端点/拓扑准备新增费必须另报。CPU布局拟worker8/10/12/14、协调30/31,留0–7给后续GPU feeders;启动前复核主机占用。
|
||||
|
||||
H2与H1同一目标的接口在本合同冻结后供Assist架构稿引用。H2先≤2 GPU-h pilot成本门、全部≤16 GPU-h另列唯一合同/一次Claude主审;本合同不启动H2或任何第三实验。
|
||||
|
||||
## 当前实际准备回执
|
||||
|
||||
`artifacts/phase2-h1-planning-2026-11-17/input-and-cost-inventory.json`保存原合同/角色元数据/代码哈希及旧费用来源;没有读取项目目标或特征数组。独立`phase2_residual_math.py`只接受调用方已经固定的c与F,不私设时钟估计器。五项合成检查(负残差/还原、零残差先验、不裁剪、非法输入、置换一致)通过,CPU0.170161秒,纳入本轮准备账;无模型拟合或项目数据访问。Mold规格、拓扑门定义和最终主审尚未收到,因此这里不冒称已冻结可运行科学合同。
|
||||
@@ -0,0 +1,11 @@
|
||||
# 更值得追问的物理结构:到达顺序之外还缺什么
|
||||
|
||||
MeltAstra|2026-11-17|一页讨论稿,供Claude整合H1/H2;没有第三队列。
|
||||
|
||||
**1.“从浇口能否走到已湿点”可能比湿区H0合并更有用。** 【原始依据】Autodesk说明单浇口绕孔分流也能产生焊/熔接线,且结果涉及前沿会合角,强度还受温压过程影响([官方文档](https://help.autodesk.com/view/MFC/2025/ENU/?guid=GUID-BEEA8CE2-D25E-4A87-851E-E2068AB9D94B))。【推断】单浇口、无内部熔体源且已湿材料不断开的假设下,已湿点应有一条不经过更晚到达点的通路连回浇口。绕孔两支碰面前已通过上游连通,故会合不必是湿区H0合并。【可检验】H1/H2的预测是否产生远离浇口的持久“提前湿岛”?它可成为因果可达性诊断,非焊线真值。实际有限网格/并列时间及误差容差由Mold定义;干区持久分支可能更适合描述末端。当前不添加损失或新运行。
|
||||
|
||||
**2.体积分位统一的是进度,不是几何或液压。** 【证据】既有true-order+训练侧时钟oracle相对L2为0.014781899;R2重排0.073059692与原0.072992654接近([冻结报告](https://gitea.frank14f.online/Frank14f/MeltMind/src/commit/9b7121a/docs/naive-ordering-results-2026-11-15.md))。【推断】把各件标到0→1是有用公共进度,但同一分位可对应完全不同的前沿形状、分支占据与流阻;标量不是整个物理状态。oracle差值不分解排序/时钟/模型误差,也非精度下限。【可检验】在H1/H2既定结果上同时看场、顺序、末端和经核正的拓扑诊断:残差是否改善场而破坏连通可达性?不因单个指标较好事后增添变体。
|
||||
|
||||
**3.可以压缩传输关系,而不必先压缩整个几何。** 【文献已读】[Boo23](https://arxiv.org/abs/2311.14554v1)构造固定离散域上的守恒通量;[Pro25](https://doi.org/10.48550/arXiv.2506.02337)以固定图边关系协调内部通量,均未证明变几何注塑泛化。【推断】更有意义的潜变量可能是“片层/原3D区域+接口响应”,让局部细节通过少量端口影响全局。湿域、温度与非牛顿状态改变响应,故静态几何latent与当前物理latent需分开。【可检验】H2的粗层通信是否真正沿局部连通传递分支影响,是架构规格可审查的问题;硬守恒或新压力求解器不是本轮默认组件。
|
||||
|
||||
**取舍:** H1先隔离固定物理跳连的学习效果;H2再检验实际几何通信的增量。上述发现提出能失败的问题,不把薄隙失效、7%平台或L2平滑导致拓扑错误当已证因果,也不据D2判定更多训练必定无效。
|
||||
@@ -0,0 +1,97 @@
|
||||
{
|
||||
"content": [
|
||||
{
|
||||
"type": "text",
|
||||
"text": "{\n \"search_id\": \"search_7b97bbf64f06ea1fcf2febf3704b5987\",\n \"results\": [\n {\n \"url\": \"https://help.autodesk.com/view/MFC/2025/ENU/?guid=GUID-BEEA8CE2-D25E-4A87-851E-E2068AB9D94B\",\n \"title\": \"Moldflow Communicator Help | Weld lines result | Autodesk\",\n \"publish_date\": null,\n \"excerpts\": [\n \"Weld lines result\\n\\nThe Weld lines result displays the angle of convergence as two flow fronts meet. The presence of weld lines may indicate a structural weakness and/or a surface blemish.\\nTip: By overlaying the Fill time result and stepping through the animation, you can see how the flow fronts converge.\\nThe term “weld line” is often used to mean both weld and meld lines. The only difference between them is the angle at which they are formed; weld lines form at lower angles than meld lines. Weld lines can cause structural problems and make the part visually unacceptable, but they are unavoidable when the flow front splits and comes together around a hole, or if the part has multiple gates.\\nConsider the processing conditions and position of the weld lines to determine whether the weld lines will be high quality. Weld and meld lines should be avoided, particularly weld lines in areas that require strength or a smooth appearance.\\nNote: For Midplane or Dual Domain studies, this result is the same as a custom weld line plot with a meeting angle of 135 degrees. For a different meeting angle, create a custom weld line plot and specify the meeting angle.\\nProcessing conditions help to determine the quality of weld or meld lines. Weld line strength is influenced by the temperature at which the weld line is formed and the pressure exerted on the weld until the part freezes; pressure is 0 at the time the weld line forms. Typically a “good” weld will occur if the temperature of the melt at the weld line as it forms is no more than 20°C below the injection temperature.\\n\\nUsing this result\\n\\nWeld lines can be moved by changing the fill pattern to make the flow fronts meet at a different place. To move weld lines:\\n* Alter the gate locations.\\n* Change the thickness of the part.\\nTo improve the quality of weld lines:\\n* Increase the melt temperature, injection speed, or packing pressure. This will enable the flow fronts to weld to each other more effectively.\\n* Increase the diameters of gates and runners to make it easier to pack the part.\\n* Move injection locations to make weld lines form closer to the gates. The weld line is then created with a higher flow front temperature and is packed with more pressure.\\n* Move injection locations to make flow fronts meet more obliquely, turning the weld line into a meld line.\\n* Place a vent in the area of the weld line. This will remove air traps, which could further weaken the weld line.\\n* Optimize the design of the runner system.\\n* Reduce runner dimensions and maintain the same flow rate. Shear heating can then be utilized to increase the melt temperature at the flow front.\\nSolving one problem can introduce other problems to the injection molding process. Carefully consider all the relevant aspects of the mold design specification before choosing an option.\\n\\nThings to look for\\n\\nThe Weld Lines result helps you identify the following problems:\\n* Structural problems: The part may be more likely to fracture or deform at a weld line, especially if the weld line is of a low quality. This weakness is a more serious problem in areas of the part that are subject to stress.\\n* Visual defects: A weld line can cause a line, notch, or color change on the surface of the part. If the weld line is positioned on a non-critical part surface, such as the bottom, this may not be a problem.\\nPages in this section\\n* Customizing weld line display properties\\n* Creating a custom weld line angle result\\nParent page: Fill or Flow results\\n\\nWas this information helpful?\\n\\n* Yes\\n* No\\nExcept where otherwise noted, this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License . Please see the Autodesk Creative Commons FAQ for more information.\\n* Privacy Statement\\n* Legal Notices & Trademarks\\n* Report Noncompliance\\n* © 2025 Autodesk Inc. All rights reserved\"\n ]\n },\n {\n \"url\": \"https://help.autodesk.com/view/MOLDFLOW/2013/ENU/caas.html?url=caas/vhelp/help-dev-autodesk-com/v/Simulation-Moldflow/enu/2013/Help/3Insight-360/4443-Results4443/4813-Flow-res4813/4885-Weld-and4885.html\",\n \"title\": \"Weld and meld lines result\",\n \"publish_date\": null,\n \"excerpts\": [\n \"Weld and meld lines result\\n\\nThis result is generated at the end of a Fill analysis.\\nWeld and meld lines are a weakness or visible flaw created when two or more flows meet and converge while filling a part. The presence of weld or meld lines may indicate a structural weakness and/or a surface blemish. The only diffeLine truncated
|
||||
}
|
||||
],
|
||||
"structuredContent": {
|
||||
"search_id": "search_7b97bbf64f06ea1fcf2febf3704b5987",
|
||||
"results": [
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MFC/2025/ENU/?guid=GUID-BEEA8CE2-D25E-4A87-851E-E2068AB9D94B",
|
||||
"title": "Moldflow Communicator Help | Weld lines result | Autodesk",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Weld lines result\n\nThe Weld lines result displays the angle of convergence as two flow fronts meet. The presence of weld lines may indicate a structural weakness and/or a surface blemish.\nTip: By overlaying the Fill time result and stepping through the animation, you can see how the flow fronts converge.\nThe term “weld line” is often used to mean both weld and meld lines. The only difference between them is the angle at which they are formed; weld lines form at lower angles than meld lines. Weld lines can cause structural problems and make the part visually unacceptable, but they are unavoidable when the flow front splits and comes together around a hole, or if the part has multiple gates.\nConsider the processing conditions and position of the weld lines to determine whether the weld lines will be high quality. Weld and meld lines should be avoided, particularly weld lines in areas that require strength or a smooth appearance.\nNote: For Midplane or Dual Domain studies, this result is the same as a custom weld line plot with a meeting angle of 135 degrees. For a different meeting angle, create a custom weld line plot and specify the meeting angle.\nProcessing conditions help to determine the quality of weld or meld lines. Weld line strength is influenced by the temperature at which the weld line is formed and the pressure exerted on the weld until the part freezes; pressure is 0 at the time the weld line forms. Typically a “good” weld will occur if the temperature of the melt at the weld line as it forms is no more than 20°C below the injection temperature.\n\nUsing this result\n\nWeld lines can be moved by changing the fill pattern to make the flow fronts meet at a different place. To move weld lines:\n* Alter the gate locations.\n* Change the thickness of the part.\nTo improve the quality of weld lines:\n* Increase the melt temperature, injection speed, or packing pressure. This will enable the flow fronts to weld to each other more effectively.\n* Increase the diameters of gates and runners to make it easier to pack the part.\n* Move injection locations to make weld lines form closer to the gates. The weld line is then created with a higher flow front temperature and is packed with more pressure.\n* Move injection locations to make flow fronts meet more obliquely, turning the weld line into a meld line.\n* Place a vent in the area of the weld line. This will remove air traps, which could further weaken the weld line.\n* Optimize the design of the runner system.\n* Reduce runner dimensions and maintain the same flow rate. Shear heating can then be utilized to increase the melt temperature at the flow front.\nSolving one problem can introduce other problems to the injection molding process. Carefully consider all the relevant aspects of the mold design specification before choosing an option.\n\nThings to look for\n\nThe Weld Lines result helps you identify the following problems:\n* Structural problems: The part may be more likely to fracture or deform at a weld line, especially if the weld line is of a low quality. This weakness is a more serious problem in areas of the part that are subject to stress.\n* Visual defects: A weld line can cause a line, notch, or color change on the surface of the part. If the weld line is positioned on a non-critical part surface, such as the bottom, this may not be a problem.\nPages in this section\n* Customizing weld line display properties\n* Creating a custom weld line angle result\nParent page: Fill or Flow results\n\nWas this information helpful?\n\n* Yes\n* No\nExcept where otherwise noted, this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License . Please see the Autodesk Creative Commons FAQ for more information.\n* Privacy Statement\n* Legal Notices & Trademarks\n* Report Noncompliance\n* © 2025 Autodesk Inc. All rights reserved"
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MOLDFLOW/2013/ENU/caas.html?url=caas/vhelp/help-dev-autodesk-com/v/Simulation-Moldflow/enu/2013/Help/3Insight-360/4443-Results4443/4813-Flow-res4813/4885-Weld-and4885.html",
|
||||
"title": "Weld and meld lines result",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Weld and meld lines result\n\nThis result is generated at the end of a Fill analysis.\nWeld and meld lines are a weakness or visible flaw created when two or more flows meet and converge while filling a part. The presence of weld or meld lines may indicate a structural weakness and/or a surface blemish. The only difference between a meld line and weld line definition is the angle at which they are formed.\nA meld line is typically formed by parallel flows. Meld lines tend to be stronger and generally less visible than weld lines. The quality of the meld line is dependent on the material type, the type and amount of fillers, and the pressure and temperature at the meld line.\nWeld lines are formed at lower angles. Weld lines can cause structural problems, and they can also make the part visually unacceptable. Therefore weld and meld lines should be avoided if possible. However, weld lines are unavoidable when the flow front splits and comes together, around a hole, or has multiple gates. Look at the processing conditions and the weld line position to decide if the weld lines will be of a high quality. Avoid weld lines in areas which need strength, or need to appear smooth.\nNote This result is the same as a custom weld line plot with a meeting angle of 179 degrees. For a different meeting angle, create a custom weld line plot and specify the meeting angle.\nProcessing conditions help to determine the quality of weld or meld lines. Weld line strength is influenced by the temperature at which the weld line is formed and the pressure exerted on the weld until the part freezes, (pressure is 0 at the weld line). A good weld typically occurs when the melt temperature where the weld line is formed is no more than 20°C below the injection temperature.\n\nUsing this result\n\nWeld and meld lines can cause structural problems, and they can also make the part visually unacceptable. However, some weld and lines are unavoidable, so you need to look at the processing conditions and the weld/meld line position to decide if they will be of a high quality.\nTo examine the processing conditions under which the weld or meld lines occurred, you can change the weld line properties.\nTo move weld lines, change the fill pattern to make the flow fronts meet at a different place:\n* Alter gate positions.\n* Change the part thickness.\nTo improve the quality of weld lines:\n* Increase the melt temperature, injection speed, or packing pressure. This will allow the flow fronts to weld to each other better.\n* Increase the diameters of gates and runners, to make it easier to pack the part.\n* Move injection locations to make weld lines form closer to the gates, so the weld line is created with a higher flow front temperature and is packed with more pressure.\n* Move injection locations to make flow front meet more obliquely, turning the weld line into a meld line.\n* Place a vent in the area of the weld line. This will remove air traps, which could further weaken the weld line.\n* Optimize runner system design.\n* Reduce runner dimensions and maintain the same flow rate to use shear heating to increase the melt temperature at the flow front.\nSolving one problem can often introduce other problems to the injection molding process. Each option requires consideration of all relevant aspects of the mold design specification.\nNote Accurate weld or meld line prediction depends on the quality of the mesh. Refining the mesh can improve weld line prediction, especially around holes.\n\nThings to look for\n\nThis result helps you identify the following problems:\nStructural problems The part may be more likely to fracture or deform at a weld line, especially if the weld line is of a low quality. This weakness will be more of a problem in areas of the part which will be subject to stress. Visual defects A weld line can cause a line, notch or color change on the surface of the part. If the weld line is positioned on a non-critical part surface (for example, the bottom of the part), this may not be a problem."
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/cloudhelp/2023/ENU/MoldflowInsight-CLC-Troubleshoot/files/Troubleshooting-molding-problems/LM_WELD_AND_MELD_LINES.html",
|
||||
"title": "Troubleshooting weld lines and meld lines",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Troubleshooting weld lines and meld lines\n\nA weld or meld line on plastic parts can cause structural problems and/or be visibly unacceptable.\nA weld or meld line is a weakness or visible flaw created when two or more flow paths meet during the filling process. Weld lines can be caused by material flowing around holes or inserts in the part, multiple injection gates or variable wall thickness where hesitation or \"race tracking\" can occur. If the different flow fronts have cooled before meeting, they don't interfuse well and can cause a weakness in the molded part. A line, notch and/or color change can appear.\nNote: The Weld lines result in the Study Tasks pane may not show all weld lines if the model mesh is too coarse.\n\nDifference between weld and meld lines\n\nThe difference between a weld and meld line is determined by the angle at which the converging flow fronts meet.\nweldline angle\nIn the above diagram, the converging flow fronts (indicated by black arrows) meet. If the angle, theta , is greater than 135° a meld line will form. If theta is less than 135° a weld line will form.\n\nWeld lines\n\nWhen a weld line forms, the thin frozen layers at the front of each flow path meet, melt, and then freeze again with the rest of the plastic. The orientation of the plastic at the weld is therefore perpendicular to the flow path. The following animation shows plastic filling a cavity. The weld line occurs where two flow fronts meet, and the polymer molecules are misaligned.\nIt is the sharp difference in molecular orientation at the weld which causes the significant decrease in strength at this point.\nweld line\n\nMeld lines\n\nA meld line occurs when two flow fronts blend together at an oblique angle. The orientation of the plastic molecules is therefore more uniform than the orientation after a weld line has formed. The following diagram shows the length of a part where a meld line forms.\nflow front angles\nThe red arrows show the direction of plastic flow. The white lines represent the orientation of the polymer molecules after the meld line has formed.\nMeld lines are normally stronger than weld lines and are often much less visible.\nmeld line\nNote: The term weld line is often used to mean both weld and meld lines.\n\nRemedies\n\nWeld and meld lines on a plastic part can cause structural problems and be visually unacceptable. (A line, notch and/or color change can appear.) Therefore weld and meld lines should be avoided if possible (when the cavity has unbalanced flow paths unnecessary weld and meld lines can occur).\nIf it is not possible to remove a weld/meld line, it should be positioned in the least sensitive area possible. Avoid weld lines in areas which need strength, or which need to appear smooth. This can be done by changing the polymer injection location or altering wall thicknesses to set up a different fill time. With a different fill time, flow fronts may meet at a different location and therefore the weld/meld line will move.\n* Moving:\n* Change the gate positions.\n* Change the part thickness.\n* Improving the quality:\n* Increase melt and mold temperature. This will allow the flow fronts to interfuse more.\n* Increase ram speed.\n* Optimize runner system design. Reduce runner dimensions and maintain the same flow rate to use shear heating to increase the melt temperature at the flow front.\nNote: The processing conditions help to determine the quality of the weld or meld line that has formed. A good weld occurs when the melt temperature is no lower than 20°C below the injection temperature.\nSolving one problem can often introduce other problems to the injection molding process. Each option requires consideration of all relevant aspects of the mold design specification.\nParent page: Troubleshooting molding problems"
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MOLDFLOW/2013/ENU/caas.html?url=caas%2Fvhelp%2Fhelp-dev-autodesk-com%2Fv%2FSimulation-Moldflow%2Fenu%2F2013%2FHelp%2F4Adviser%2F1747-Results1747%2F1806-Fill-ana1806.html",
|
||||
"title": "Fill analysis results",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Fill analysis results\n\nThis topic specifies the results generated for a Fill analysis on a thermoplastic material.\n\nText based results\n\nThe following table lists the text results that are created for each Fill analysis.\n* Results\n* Results Summary\n\nGraphical results\n\nThe following table lists the graphical results that are created for a Fill analysis and indicates whether each result is supported by the following analysis technologies:\n* Dual Domain\n* 3D\nFor more information about a result, including how to interpret the display, click the result name.\n* Result | Available for analysis type\n* Air traps result | \n* Average temperature result | \n* Confidence of fill result | \n* Fill time result (Plastic Flow is an alternate method of displaying the Fill time result.) | \n* Frozen layer fraction at end of fill result | \n* Grow from result | \n* Orientation at skin result | \n* Injection pressure | \n* Pressure drop result | \n* Quality prediction result | \n* Temperature at flow front result | \n* Time to reach ejection temperature result | \n* Weld lines result | \n\nTopics in this section\n\n* Fill analysis results (Quick Reference)\nView a summary of the Fill analysis result.\n* Air traps result\nAn air trap occurs where the melt traps and compresses a bubble of air or gas between two or more converging flow fronts, or between the flow front and the cavity wall. Typically, the result is a small hole or a blemish on the surface of the part. In extreme cases, the compression increases the temperature to a level that causes the plastic to degrade or burn.\n* Average temperature result\nThe Average temperature result shows the average bulk temperature through the thickness of the part at the end of fill.\n* Confidence of fill result\nThe Confidence of fill result displays the probability of plastic filling a region within the cavity under conventional injection molding conditions.\n* Fill time result\nThe Fill time result shows the position of the flow front at regular intervals as the cavity fills.\n* Frozen layer fraction at end of fill result\nThe Frozen layer fraction at end of fill result represents the thickness fraction of the frozen layer at the end of filling.\n* Grow from result\nThe Grow from result, which is produced by a Fill analysis, shows the areas that are filled from each gate in multi-gated parts so that you can determine whether the flow within the part is balanced.\n* Orientation at skin result\nThe Orientation at skin result provides a good indication of how molecules will be oriented on the outside of the part, showing the average principal alignment direction for the whole local area at the end of filling.\n* Injection pressure result\nThe Injection pressure result, which is produced by a Fill analysis, shows the maximum injection pressure value obtained before the velocity/pressure switch-over occurs during the filling phase.\n* Pressure drop result\nThe Pressure drop result uses a range of colors to indicate the region of highest pressure drop through to the region of lowest pressure drop. This result indicates how much pressure is necessary to fill the different areas of the part.\n* Quality prediction result\nThe Quality prediction result is used to estimate the quality of the mechanical properties and appearance of the part. This result is derived from the pressure, temperature, and other results.\n* Temperature at flow front result\nThe Temperature at flow front result, which is produced by a Fill analysis, shows the temperature of the polymer when the flow front reaches a specified point in the center of the plastic cross-section.\n* Time to reach ejection temperature result\nThe Time to reach ejection temperature result shows the amount of time required to reach the ejection temperature, which is measured from the start of fill.\n* Unfilled cavity result\nThe Unfilled cavity result, which is produced by a Fill analysis using a 3D analysis technology, shows areas in a model where the part has not filled.\n* Weld lines result\nThe Weld lines result displays the angle of convergence as two flow fronts meet. The presence of weld lines may indicate a structural weakness and/or a surface blemish."
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/cloudhelp/2016/ENU/MFIA-Learn/files/GUID-B72D523E-0396-4F5A-9F10-CAA944ED6C83.htm",
|
||||
"title": "Basic model analysis (tutorial)",
|
||||
"publish_date": "2015-04-01",
|
||||
"excerpts": [
|
||||
"Basic model analysis\n\nIn the previous tasks, you prepared the tutorial model for analysis. You selected a material, designed a multi-cavity mold, and modeled the runner system and cooling system.\nIn this task, you will:\n* Examine the information generated by the analysis.\n* Examine the graphical representation of the analysis results.\n* Determine the result value at a selected point on the mold.\n* Look at a brief description of some of the major results.\nOnce an analysis is complete, a selection of results are made available in the Study Tasks pane. For a Fill (or Fill+Pack) analysis, the results appear in a Flow folder.\n\n...\n\n* The Filling tab presents the analysis inputs and the summary results, including bulk temperature, shear stress, and frozen layer fraction at the end of the filling stage. The low mesh match percentage is also reported in this tab.\n* The Machine Setup tab outlines the processing parameters used for the analysis. This includes the material used, the machine specifications, and temperature and injection settings that should be used on the injection molding machine.\n* The Filling-Check tab presents the results of the checks performed at the beginning of the analysis.\n\n...\n\n9. Select Logs from the Study Tasks pane to show the Logs pane.\nAutodesk Moldflow Insight gives detailed information about the part design, material, and processing conditions used in the simulation.\nSome of the Fill analysis results available are listed below.\nFill Time The Fill time result shows the flow path of the plastic through the part. This result can be animated against the time required to fill the part. Pressure at V/P switchover The Pressure result shows the pressure distribution through the flow path inside the mold at the switchover point from velocity to pressure control. Average velocity The Average velocity result plot displays the flow direction, and the average magnitude of velocity of the polymer inside the mold cavity. Orientation at core The Orientation at core result is created at the end of a filling analysis and provides a good indication of how molecules or fibers would be oriented at the part core when using a fiber-filled material. Time to reach ejection temperature The Time to reach ejection temperature result shows the time taken for the part to freeze to ejection temperature, measured from the start of the cycle. Bulk temperature The Bulk temperature result displays the weighted average temperature across the thickness of the part at the time the result was written. Bulk temperature represents the energy that is transported through a particular location. Weld lines The Weld lines result indicates the presence and location of weld in the filled part model. Weld lines appear as black lines on the model, and can indicate weakness. To see all of the weld lines, you may need to rotate the model. Air traps The Air traps result shows the regions where the melt stops at a convergence of at least two flow fronts, or at the last point of fill. The regions highlighted in the result are positions of possible air traps. To see all of the air traps, you may need to rotate the model. Air traps appear as large red outlines on the model.\nThe Fill Time and Pressure results are good starting points when checking the results of an analysis. They will show you where there may be filling or pressure distribution problems associated with the part design. If problems are indicated, you may need either to change the part design or use a different combination of material and processing conditions.\nThe results in this study indicate a good fill pattern and part design.\nClick the Next topic link below to move on to the next task of the tutorial.\nParent topic: Analysis overview (tutorial)\nPrevious topic: Designing a cooling circuit\nNext topic: Tutorial review"
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MFIA/2023/ENU?contextId=conc_analysis_result_fill_weld_lines",
|
||||
"title": "Moldflow Insight Help | Weld lines result | Autodesk",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Moldflow Insight Help | Weld lines result | Autodesk\nFor a different meeting angle, create a custom weld line plot and specify the meeting angle. Processing conditions help to determine the quality of weld or meld lines."
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/cloudhelp/2024/ENU/MoldflowComm-CLC-Results/files/GUID-3D06ED04-FC59-4B68-A103-AAA5EB073EB8.html",
|
||||
"title": "Fill or Flow results",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Fill or Flow results\n\nThis help topic specifies the results generated for analyses that include Fill, Fill+Pack, or Standalone Pack analysis sequences.\nThese results are generated for material with or without fiber or filler content.\n\nText based results\n\nThe following table lists the text results that are created for each these analyses.\n* Results\n* Analysis Log\n\n...\n\n* Extension rate (3D) | \n* Fill time | \n* Flow front speed | \n* Flow front velocity | \n* Flow front temperature, mold-cavity | \n* Flow rate, beams 2 | \n* Time to reach ejection temperature | \n* Frozen layer fraction |\n\n...\n\n* Temperature | \n* Temperature (3D) | \n* Temperature at flow front | \n* Thermal conductivity tensor 3 | \n* Throughput | \n* Time to reach ejection temperature | \n* Unfilled cavity | \n* Velocity | \n* Velocity (3D) | \n* Viscosity | \n* Vent region pressure 1 | \n* Volumetric shrinkage | \n* Average volumetric shrinkage | \n* Volumetric shrinkage at ejection | \n* Weld lines | \n* Weld and meld lines | \n* Weld surface formation (3D) result | \n* Weld surface movement (3D) result | \n* 1 | This result is available only when the option to Perform venting analysis is selected in the solver parameters.\n\n...\n\n* Volumetric shrinkage result\n* Volumetric shrinkage at ejection result\n* Weld lines result\n* Weld and meld lines result\n* Weld surface derivation\n* Weld surface formation (3D)\n* Weld surface movement (3D)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MOLDFLOW/2013/ENU/caas.html?url=caas%2Fvhelp%2Fhelp-dev-autodesk-com%2Fv%2FSimulation-Moldflow%2Fenu%2F2013%2FHelp%2F2-0Insight%2F0024-Tutorial24%2F0050-Analysis50%2F0051-Selectin51.html",
|
||||
"title": "Selecting the molding process",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"10. Select Fill .\nNotice in the Project View pane that there is a symbol next to the study name. It shows you that you have selected a Fill analysis for that study. When the analysis has been run and results are available, the inside of the symbol is colored in.\nWhen analyzing a part from an imported CAD model, start with a basic filling (\"Fill\") analysis to check the filling pattern, injection pressure, and the location of air traps or weld lines. Once you are satisfied with the filling parameters for the part, you can run a Fill+Pack analysis (filling and packing phases). Then run further analyses such as cooling or warpage.\nClick the Next topic link below to move on to the next task of the tutorial."
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/cloudhelp/2015/ENU/MoldflowComm/files/GUID-E81A1CF3-0A91-43DE-8954-005E76897E51.htm",
|
||||
"title": "Moldflow Communicator 2015 Help: Air traps result",
|
||||
"publish_date": "2013-04-03",
|
||||
"excerpts": [
|
||||
"Air traps result\n\nAn air trap occurs where the melt traps and compresses a bubble of air or gas between two or more converging flow fronts, or between the flow front and the cavity wall. Typically, the result is a small hole or a blemish on the surface of the part. In extreme cases, the compression increases the temperature to a level that causes the plastic to degrade or burn.\nAir traps are often due to converging flow fronts caused by racetrack or hesitation effects, or by non-uniform or non-linear fill patterns. Even when the part has balanced flow paths, inadequate venting can cause air traps to occur at the ends of flow paths.\nFor a Midplane or Dual Domain model, the Air traps result shows a thin, continuous line wherever an air trap is likely to occur.\n\n...\n\nThe following methods can be used to prevent air traps:\n* Use flow leaders or deflectors.\n* Increase the injection speed to eliminate air traps caused by converging flow fronts and hesitation.\n* Decrease the injection speed to reduce air traps caused by poor venting, and to prevent burn marks.\n* Decrease the part wall thickness ratio to reduce racetracking.\n* Move the injection locations so that the air traps form in areas that are easy to vent, such as the parting plane.\nNote: Changes such as these can only be made using a licensed Autodesk Simulation Moldflow Adviser or Autodesk Simulation Moldflow Insight Product.\nParent topic: Flow results"
|
||||
]
|
||||
},
|
||||
{
|
||||
"url": "https://help.autodesk.com/view/MFAA/2024/ENU/?guid=GUID-AA3B3176-BE1B-4E55-AF16-46FE7E803C75",
|
||||
"title": "Moldflow Adviser Help | Find an injection location | Autodesk",
|
||||
"publish_date": null,
|
||||
"excerpts": [
|
||||
"Find an injection location\n\nRun a Gate Location analysis to find suitable injection locations and automatically place an injection location on a copy of the model.\nThe location of the injection point on the model affects whether the part will suffer from defects such as short shots, weld lines or over packing.\nNote: Gate Location analysis results do not indicate whether an injection location is physically practical or cosmetically acceptable.\n1. If the Analysis Wizard is not displayed, click ( Home tab > Molding Process Setup panel > Analysis Wizard ).\n2. Select the Sequence tab and select the Gate Location checkbox.\n\n...\n\n7. Select Finish to close the Analysis Wizard .\nYou can select regions of the model where you do not want to allow injection locations.\nThe default selection method is using a tolerance angle. Adjacent areas are selected when they meet at an angle less than the tolerance angle. By increasing the tolerance angle, or selecting a number of regions, you can prohibit injection locations from being placed on the front face of the cellphone cover.\n8. Click ( Boundary Conditions tab > Injection Locations panel > Prohibited Gate Regions ).\nThe Tools tab displays the Prohibited Gate Regions pane.\nThe Selection Method pane allows you to define how a region will be chosen. Ensure Tolerance Angle is selected form the drop down box. Scroll down, if necessary, to see this section of the dialog."
|
||||
]
|
||||
}
|
||||
],
|
||||
"warnings": null,
|
||||
"metadata": null,
|
||||
"session_id": "session_5786437d5aeb6028b9eeee2f2a0e37a7"
|
||||
},
|
||||
"isError": false
|
||||
}
|
||||
@@ -0,0 +1,46 @@
|
||||
"""Array-only residual interface; no I/O, model, labels, or clock fitting policy.
|
||||
|
||||
The caller supplies a frozen, training-only clock. It also fixes the F precision
|
||||
in the reviewed contract. Passing F explicitly avoids a hidden extra feature.
|
||||
"""
|
||||
import numpy as np
|
||||
|
||||
|
||||
def prior(cdf, clock):
|
||||
f = np.asarray(cdf)
|
||||
if f.ndim != 1 or f.dtype.kind != 'f' or not len(f):
|
||||
raise ValueError('Expected nonempty floating-point CDF vector')
|
||||
if not np.isfinite(f).all() or np.any(f < 0) or np.any(f > 1):
|
||||
raise ValueError('CDF outside [0,1] or nonfinite')
|
||||
if not np.isscalar(clock) or not np.isfinite(clock) or clock <= 0:
|
||||
raise ValueError('Clock must be a finite positive frozen scalar')
|
||||
return np.asarray(f, dtype=np.float64)*float(clock)
|
||||
|
||||
|
||||
def residual_target(target_star, cdf, clock, *, fit_dtype):
|
||||
y = np.asarray(target_star, dtype=np.float64)
|
||||
base = prior(cdf, clock)
|
||||
if y.shape != base.shape or not np.isfinite(y).all() or np.any(y < 0):
|
||||
raise ValueError('Invalid original normalized arrival target')
|
||||
dtype = np.dtype(fit_dtype)
|
||||
if dtype not in (np.dtype('float32'), np.dtype('float64')):
|
||||
raise ValueError('Fit precision must be explicitly float32 or float64')
|
||||
out = np.asarray(y-base, dtype=dtype)
|
||||
if not np.isfinite(out).all():
|
||||
raise ValueError('Residual overflow at fit precision')
|
||||
# Negative residuals are valid. Never clip or impose a positivity loss here.
|
||||
return out
|
||||
|
||||
|
||||
def decode_seconds(predicted_residual, cdf, clock, t_set):
|
||||
residual = np.asarray(predicted_residual, dtype=np.float64)
|
||||
base = prior(cdf, clock)
|
||||
if residual.shape != base.shape or not np.isfinite(residual).all():
|
||||
raise ValueError('Invalid predicted residual')
|
||||
if not np.isscalar(t_set) or not np.isfinite(t_set) or t_set <= 0:
|
||||
raise ValueError('Invalid pre-solve seconds scale')
|
||||
out = (base+residual)*float(t_set)
|
||||
if not np.isfinite(out).all():
|
||||
raise ValueError('Decoded prediction overflow')
|
||||
# Keep negative predictions visible to scoring/diagnostics; no silent clamp.
|
||||
return out
|
||||
@@ -0,0 +1,38 @@
|
||||
"""Synthetic interface checks only; no model fit or project data access."""
|
||||
import unittest
|
||||
import numpy as np
|
||||
from phase2_residual_math import prior, residual_target, decode_seconds
|
||||
|
||||
|
||||
class ResidualInterfaceTests(unittest.TestCase):
|
||||
def test_negative_residual_and_exact_reconstruction(self):
|
||||
f = np.array([0., .25, .75]); y = np.array([0., .125, .5])
|
||||
r = residual_target(y, f, 2., fit_dtype='float64')
|
||||
self.assertTrue(np.any(r < 0))
|
||||
np.testing.assert_array_equal(decode_seconds(r, f, 2., 4.), y*4.)
|
||||
|
||||
def test_zero_residual_is_prior(self):
|
||||
f = np.array([0., .25, .5], dtype=np.float32)
|
||||
np.testing.assert_array_equal(decode_seconds(np.zeros(3), f, 1.5, 2.), np.asarray(f, dtype=np.float64)*3.)
|
||||
|
||||
def test_no_posthoc_clamp(self):
|
||||
y = decode_seconds(np.array([-1., 2.]), np.array([0., 1.]), 1., 1.)
|
||||
np.testing.assert_array_equal(y, [-1., 3.])
|
||||
|
||||
def test_bad_shapes_and_inputs(self):
|
||||
for f in [np.array([]), np.array([np.nan]), np.array([-0.1]), np.array([1.1]), np.zeros((2, 1))]:
|
||||
with self.assertRaises(ValueError): prior(f, 1.)
|
||||
for clock in [0., -1., float('nan'), float('inf')]:
|
||||
with self.assertRaises(ValueError): prior(np.array([.5]), clock)
|
||||
with self.assertRaises(ValueError): residual_target(np.array([1., 2.]), np.array([.5]), 1., fit_dtype='float32')
|
||||
with self.assertRaises(ValueError): decode_seconds(np.array([0.]), np.array([.5]), 1., 0.)
|
||||
|
||||
def test_permutation_equivariance(self):
|
||||
f = np.array([0., .1, .1, .7]); y = np.array([0., .05, .2, .8]); order = np.array([2, 0, 3, 1])
|
||||
a = residual_target(y, f, .9, fit_dtype='float32')
|
||||
b = residual_target(y[order], f[order], .9, fit_dtype='float32')
|
||||
np.testing.assert_array_equal(a[order], b)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
unittest.main()
|
||||
Reference in new issue
Block a user