About the lab / evidence you can inspect
Understand the model. Reproduce the result.
Datacenter Twin Lab is a local-first power-continuity what-if simulator and a twelve-lesson course for datacenter engineers.
Created and maintained by Mohammad Rezwan Khan. The project is open source under Apache-2.0. It runs on your device without an account or shared simulation backend.
How long does 100 kWh support a 1 MW load?
For this fixed-load example, 100 kWh × 0.90 discharge efficiency × 0.95 distribution efficiency = 85.5 kWh delivered. That is 307.8 seconds at 1,000 kW. With 50 kWh, it is 153.9 seconds. Charging is disabled; utility and generator fail at 300 seconds and recover at 900 seconds.
What the simulator does
It accounts for finite battery energy, charging and losses, generator startup and failure, surviving-path capacity, shared failure domains, timed recovery and unserved energy. The fast JavaScript path can be checked on demand against the Python reference. Eighteen presets and twelve lessons use explicit inputs and unit-labelled results.
What the evidence means
The tests and published packets establish reproducible software behavior for the declared inputs. They do not establish facility calibration, AC transients, protection coordination, GPU workload performance, grid adequacy, certified uptime or equipment safety. This software does not control equipment. Independent external reproduction has not yet been obtained.
Inspect or cite the work
- Version 1.0.0 source, packages and release evidence
- Canonical input equations and reconstruction instructions
- Versioned evidence manifest with per-file hashes
- Author and software citation record
- External reproduction protocol and scope
- Install and run with Python
How to contribute
Try one lesson, note anything confusing and describe the steps to reproduce it. A clearer explanation, a documentation fix or an independently derived example can help the next learner. Read the contribution guide or open an issue.