# Twinworks Buffer & Parameter Sweep

> Sweep a buffer size (or any parameter) and get the throughput curve with confidence intervals. Runs a Twinworks factory model repeatedly while one parameter changes — typically the capacity of a buffer from 1 to 10 places, but also servers, cycle times, MTTR, batch sizes, conveyor speed or the number of operators. Returns throughput, lead time and WIP per value with confidence intervals, plus the knee point where extra capacity stops paying off.

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://twinworksbuffersweep.usefulweb.site/ (interactive tool at https://twinworksbuffersweep.usefulweb.site/tool)
- JSON API: POST https://twinworksbuffersweep.usefulweb.site/api/v1/calculate with a JSON object of inputs (schema: https://twinworksbuffersweep.usefulweb.site/api/v1/tool)
- OpenAPI: https://twinworksbuffersweep.usefulweb.site/openapi.json
- MCP (Streamable HTTP): https://twinworksbuffersweep.usefulweb.site/mcp - tool `twinworks_buffer_sweep`

## Inputs
- model (textarea): Model or project JSON - A Twinworks project (.twinworks.json, format "twinworks-project") or a bare model with "nodes" and "links". Times in seconds, lengths in metres. Leave empty and pick a template instead.
- template (select): Or start from a template
- target (text): Object id - Id of the buffer/machine/conveyor/source to change, or pool:<id> for an operator pool. Default for the bottleneck template: B2.
- parameter (select): Parameter
- from (number): From
- to (number): To
- step (number): Step
- replications (integer): Replications per value
- horizonHours (number): Run length per replication - Overrides the model setting.

## Method
- For each parameter value the model is copied, the parameter is set and the model is simulated with the same seeds (common random numbers), so differences between values are not masked by random noise.
- Throughput, lead time and WIP are reported as Student-t intervals across replications.
- Knee point = smallest value whose mean throughput is at least 98 % of the best mean throughput in the sweep: beyond it, more buffer mostly adds WIP and lead time (Little’s law).

## Limits
- The sweep varies one parameter at a time; interactions between buffers are not optimised jointly.