# Twinworks Line Balancing

> Assign assembly tasks to stations for a takt time: efficiency, balance delay and smoothness index. Solves the simple assembly line balancing problem (type 1) with the Ranked Positional Weight (Helgeson–Birnie) and Largest Candidate rules: tasks with times and precedence relations are assigned to as few stations as possible without exceeding the takt time. Returns the station assignment, line efficiency, balance delay, smoothness index and a ready-to-simulate Twinworks model of the balanced line.

Part of usefulweb.site (https://usefulweb.site/). Results are deterministic calculations from the inputs and the listed assumptions; the API returns exactly what the web page shows.

## Use it
- Human interface: https://twinworkslinebalance.usefulweb.site/ (interactive tool at https://twinworkslinebalance.usefulweb.site/tool)
- JSON API: POST https://twinworkslinebalance.usefulweb.site/api/v1/calculate with a JSON object of inputs (schema: https://twinworkslinebalance.usefulweb.site/api/v1/tool)
- OpenAPI: https://twinworkslinebalance.usefulweb.site/openapi.json
- MCP (Streamable HTTP): https://twinworkslinebalance.usefulweb.site/mcp - tool `twinworks_line_balance`

## Inputs
- tasks (table, required): Tasks
- taktTime (number): Takt time - Leave empty to compute it from available time ÷ demand.
- availableMinutes (number): Available production time - Per period, e.g. 450 min per shift.
- demand (number): Demand in that period
- method (select): Heuristic

## Method
- Takt time = available production time ÷ demand (or as given). Theoretical minimum stations = ⌈Σ task times ÷ takt⌉.
- RPW: positional weight = own time + times of all direct and indirect successors; LCR: task time. Stations are filled one by one with the highest-priority task whose predecessors are all assigned and that still fits into the remaining takt time.
- Efficiency = Σ task times ÷ (stations × takt); balance delay = 1 − efficiency; smoothness index = √Σ(max station time − station time)².
- The result includes a Twinworks model (one constant-time machine per station, single-place buffers) so the balance can be simulated or opened in the app.

## Limits
- Heuristics give good but not always optimal balances. Zoning constraints, operator skills and task variability are not considered.