Why static scheduling always breaks down
Every production schedule starts optimistic. It assumes machines run at full capacity, operators are always present, and every job takes exactly as long as planned. Within hours of the shift starting, reality diverges — and the schedule becomes a fiction everyone politely ignores.
The real cost isn't just missed delivery dates. It's the constant firefighting: shifting jobs, renegotiating deadlines, and making prioritisation decisions under pressure without reliable data on which orders are actually at risk. Production managers become human schedulers — spending their cognitive capacity on crisis management instead of improvement.
The core problem: Traditional scheduling tools optimise for a world where nothing goes wrong. A digital twin schedules for the world as it actually is — with variability, failures, and uncertainty built in from the start.
What a digital twin production plan looks like
A SimulateFirst digital twin is a Simio simulation model connected live to your ERP or MES. Each morning, it reads the current state of your production: open orders, their routing and process times, available machines, staff, tooling, and any constraints. It then runs thousands of simulated production days — each with slightly different variability — to produce a schedule grounded in probability, not optimism.
The result isn't just a Gantt chart. It's a schedule that tells you: this order has an 87% probability of shipping on time; this one has 43% and needs attention now. You see the risks before they become emergencies.
ERP / MES data integration
We connect the simulation model to your data source — SAP, custom MES, or file export. Orders, routings, process times, machine availability, and constraints are read automatically on a defined schedule (typically each morning).
Simulation model build & validation
We build the Simio model of your production process — machines, workers, routings, buffers, setup times, and failure distributions. It's validated against historical data before going live.
Probabilistic scheduling run
The model runs multiple replications of the coming production period. Each replication uses a different random seed for machine failures and variability. Results are aggregated to produce on-time probabilities and expected completion times per order.
Schedule output & dashboard
Results are written back to your system or displayed in a dashboard — sorted by risk level, flagging orders that need intervention. Manual overrides and what-if scenarios can be run on demand.
What you get at the end
Deliverables include: the live Simio model and integration scripts, a dashboard or data output format your team can use daily, documentation for IT handover, and training for the production planning team. The model is fully yours — no ongoing licence fees to SimulateFirst.
Digital twin vs conventional scheduling tools
| Capability | Digital twin (Simio) | ERP scheduler / Gantt tool |
|---|---|---|
| Accounts for machine failure probability | ✓ Modelled from historical MTBF data | ✗ Assumes 100% availability |
| Delivery probability per order | ✓ Expressed as % confidence | ✗ Binary on-time / late |
| Handles variability in process times | ✓ Distribution-based, not fixed values | ✗ Fixed standard times only |
| Supports what-if scenario testing | ✓ Run on demand, results in minutes | Requires manual re-scheduling |
| Adapts to live ERP data automatically | ✓ Reads fresh data each run | Manual import or static snapshot |
| Reverse planning (deadline-back) | ✓ Calculates latest start to meet due date | Forward planning only in most tools |
Tools & technology
Our digital twin scheduling work is built on Simio — chosen because its native risk-based scheduling engine is designed specifically for probabilistic planning, unlike general-purpose simulation tools that require significant customisation to produce scheduling outputs.
For complex scheduling problems with hard constraints — machine groups, shared tooling, sequence dependencies, calendar restrictions — we implement proprietary optimisation logic using Google OR-Tools and IBM CPLEX constraint programming components. This gives us full control over the scheduling algorithm, allowing tight integration with the simulation and custom objective functions specific to each client's planning priorities.
See it in practice
Steel construction plant — ERP-connected scheduling twin
AnyLogic · Manufacturing
Stochastic production model — scheduling with random failures & variability
Simio · Manufacturing
