# Logistics Process Modeling > Check logistics-simulation input data, summarize station-level process flows, decide station granularity and naming, and draft simulation model abstractions in Markdown, HTML, Mermaid, or SVG. Use when Codex needs to turn multilingual process evidence, equipment lists, Gantt cycle-time sheets, safety-zone/layout data, AGV/APC/EMS information, or vehicle/model-variant rules into a clear manufacturing/logistics simulation flow and modeling plan. - Skill: `lang7910/logistics-process-modeling` (Agent Skill) - Install (CLI): `npx skillmds@latest add lang7910/logistics-process-modeling` - Raw SKILL.md: https://api.skillmd.com/api/skills/lang7910/logistics-process-modeling/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Web & Frontend - Author: Lang7910 (https://skillmd.com/u/lang7910) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/lang7910/logistics-process-modeling --- # Logistics Process Modeling Use this skill to audit available input data, understand process flow, choose station granularity and naming, and express a logistics/manufacturing simulation model before or while building it in simulation software. Treat body-in-white, welding, general manufacturing, logistics, warehouse, AGV/AMR, conveyor, EMS, APC, machining, assembly, and packaging systems as related variations. Do not hard-code body-shop assumptions unless the sources or user instructions support them. Source language may be Chinese, English, German, or another major engineering language. Preserve original names for traceability, and add normalized aliases only for matching and modeling. This skill may consist only of `SKILL.md`. Scripts, references, or assets are optional and should be added only when they solve a repeated, general problem. If the user later adds or corrects guidance, apply the newest intent and keep generally useful rules. ## Workflow 1. Audit input data and gaps. 2. Confirm station granularity, station identifiers, model variants, and naming rules. 3. Extract process flow per line and variant from source evidence. 4. Convert the process into simulation objects and simplified logic. 5. Output a clear Markdown/HTML/Mermaid/SVG representation with assumptions and open questions. Prefer read-only inspection of source materials. Write derived conclusions separately unless the user explicitly asks to edit original files. ## Input Data Audit If a customer/project input-data checklist workbook is supplied, use it as the audit baseline. Otherwise create a checklist from the available evidence. Common data categories: - Latest layout drawing or layout image: production line, conveyor, AGV/AMR, EMS, FDS, APC, storage, transfer points. - Design takt or cycle plan: planned takt, target CT, model mix, batch rules. - Process description: PPT/PDF/Word describing stations, operations, routes, joins, manual loading, tooling, robots, and variant differences. - Safety-zone or lockout-zone division: safety layout, electrical/safety proposal, PLC/safety-zone mapping, screenshots. - Equipment list: station-level equipment, quantities, maintainability/availability inputs, robot/tooling lists. - Gantt-style cycle-time sheets: station/action timing, begin/duration/end, model variants, concurrent operations. - Buffers and WIP: APC quantity, buffer capacity, loading tables, sequence tables, pass tables, trolleys. - Interline transfer: AGV/AMR/EMS/conveyor routes, SNP, capacity, dispatch rules, speed, loading/unloading time. - Model switching: variant routes, batch size, setup loss, route switching, bypass/skip rules, connection-control logic. AGV, APC, EMS, and model-switching data may be project-specific. Do not treat missing optional categories as fatal; record whether they are not applicable, missing, or still needed. ## Station Granularity Confirm station granularity before drawing the model. It controls object count, route logic, CT assignment, availability assignment, and naming. Use multiple sources to determine granularity: - Gantt sheets are usually the clearest source for station/action boundaries. - Availability/TA summaries show which station identifiers are expected for failure modeling. - Equipment lists show which equipment is grouped under which station. - Process plans explain joins, transfers, robot handoffs, manual loading, and variant routes. Watch for inconsistent station numbering: - Some lines use `010, 015, 020...`. - Some use `010, 020, 030...`. - Some mix `OP`, `ST`, line prefixes, side suffixes, or tool-specific names. Do not silently merge or split stations because numbering “looks similar.” If evidence conflicts, list the alternatives and ask for confirmation or mark `待确认`. ## Naming Rules Use the project’s naming standard if one exists. If none exists, propose a compact and consistent rule. Common patterns: - Line abbreviation + station number: `LINE010`. - Line abbreviation + side + station number: `LINEL010`, `LINER010`. - Line + `_` + station number: `LINE_010`. - Robot/transfer object: `LINE010_RB01` or `LINE010_R01`. - Special loss/helper objects: `LINE010_PL`, `setup_loss`, `LINE_LZ1_PL`. For Plant Simulation-style models, avoid characters the software cannot use in object names, such as `-`; prefer `_`. Keep display labels and source aliases separately if the original name contains punctuation or multilingual text. ## Process Flow Extraction Extract process flow line by line and, when needed, variant by variant. For each line/station, capture: - Inputs: upstream line, trolley, APC, AGV, EMS, manual parts, loading table, buffer, source, rework/quality path. - Main operation: GEO, welding, respot, glue, loading, unloading, inspection, fastening, handling, takeout, transfer. - Equipment/actions: robots, fixtures, guns, grippers, turntables, lifters, conveyors, clamps, sensors, manual operators. - Outputs: downstream station, buffer, APC, AGV/AMR, EMS, drain, quality takeout, merge point. - Variant differences: route, station usage, CT, manual inputs, equipment usage, setup/batch logic. - Evidence: file/sheet/page/row and unresolved assumptions. Keep one line per file when that makes the result easier to review. Use one structured large file only when the user asks for a combined report or the line interactions are central. ## Modeling Abstraction Draw logical sequence and material flow, not physical floor position, unless the user asks for a layout-driven view. Use these objects consistently: - `Source`: material input, manual part input, trolley, APC/AGV/EMS arrival, external upstream line. - `Station`: one station-level process object carrying equivalent station CT and availability. - `AssemblyStation`: merge point where a main part and side input/another line/manual-loaded part must synchronize. - `Robot`: transfer/handling robot only when it needs separate route, capacity, timing, or blocking behavior. - `Buffer`: zero- or low-logic storage, APC slot, loading table, pass table, cache, turntable, or WIP point when no station-level failure/CT is assigned. - `Drain`: line output, transfer handoff, APC/AGV/EMS departure, quality takeout, scrap/rework exit. - `SetupLoss`: optional helper object or logic for batch/model switching loss. Use arrows for flow: ```text Source → Station → Buffer → AssemblyStation → Station → Drain ``` ## Simplification Rules Use the simplest object that preserves the behavior the simulation must test. - If a station-level CT and availability already represent all robot/manual/tooling actions inside that station, create one `Station` for that station and avoid assigning the same CT/availability again to internal helper robots. - If a transfer robot is only needed to represent space, capacity, blocking, or buffer behavior and its timing/failure is already folded into a station, model it as a zero/low-CT helper or buffer-like object. - If a robot or transfer process is the station-level operation and appears as an independent station in CT/availability/equipment sources, it can be modeled as a `Station` or station-equivalent object. - Direct manual loading into a station does not always require an `AssemblyStation`; use one when synchronization/merge behavior matters. - If a station receives material from another line or side branch, use `AssemblyStation` or an explicit merge representation. - If a loading table, turntable, cache, or temporary position has no independent CT/failure, use `Buffer` or a zero-CT/no-failure station-like helper. - If route differs by model variant and the software cannot easily switch connections dynamically, document options: separate variant routes, conditional bypass/skip logic, connection open/close logic, or unified route with model-dependent CT. - Use `SetupLoss` or equivalent logic for batch/model switching when the loss is mainly timing; use route-control logic when model switching changes physical path. - Represent APC/AGV/EMS only to the degree needed for the study: as a buffer, transfer, source/drain interface, route network, dispatch resource, or external assumption. ## Output Formats Markdown is the default because it is compact and easy to review. HTML can be better for visual summaries, tables, and clickable sections, but can consume more context and authoring effort. Mermaid or SVG can be used for diagrams when useful. Recommended structure: - `Data Check`: supplied, missing, not applicable, assumptions. - `Naming and Granularity`: station-numbering rule, line aliases, variant aliases. - `Process Summary`: one table per line/variant. - `Simulation Object Flow`: textual arrows and/or diagram. - `Object Assignment`: object type, CT source, availability source, failure/PL handling. - `Variant Logic`: route differences, setup/batch logic, skip/bypass rules. - `Safety/Zone Logic`: zone grouping assumptions when relevant. - `Open Questions`: unresolved evidence conflicts and decisions needed. ## Quality Rules - Separate confirmed facts from assumptions. - Preserve source traceability using exact file/sheet/page/row names when possible. - Do not over-model details that do not affect the study objective. - Do not hide side inputs inside station descriptions when they matter for synchronization; draw them as separate branches. - Do not duplicate one physical robot/object as multiple independent failure/CT contributors unless the model intentionally needs that abstraction. - Prefer clear, reviewable diagrams over visually complex floorplan-like drawings. - Mark uncertain mappings with `待确认` or `Review`.