Best Practice: Controlling Sim Minutes
Overview
Sim minutes represent the total wall-clock time consumed by:
-
Simulation (including scheduling and optimization)
-
Post-processing (KPI generation, aggregation, and exports)
In Shoresim, users cannot:
-
Select which KPIs are generated
-
Disable post-processing
Because of this, runtime must be managed through how you configure your simulation inputs.
This article explains the three levers that directly control sim minutes and how to use them effectively across different project phases.
Understanding What Drives Sim Minutes
Sim minutes increase with:
-
Longer simulation horizons
-
More replications (runs)
-
Higher model complexity
The key is not to eliminate runtime — it is to spend sim minutes intentionally based on your current objective.
The Three Runtime Levers You Control
1) Simulation Duration (Years Simulated)
This is the single biggest driver of sim minutes.
The number of simulated years directly impacts:
-
Total events generated
-
Number of rescheduling cycles
-
Volume of output data
-
Post-processing workload
Recommended Usage
Experimentation phase:
-
5–10 simulated years
Deep analysis / delivery:
-
Full lifecycle (e.g., 25 years), when lifecycle results are required
Shortening duration reduces runtime significantly and is the fastest way to control sim minutes.
2) Number of Runs (Replications)
This is the second most impactful runtime lever.
Each run repeats the entire simulation and post-processing cycle.
Recommended Usage
Experimentation phase:
-
5–10 runs
Deep analysis / delivery:
-
Determined by required statistical precision
Avoid running large numbers of replications while still exploring solution space. You can increase runs once your scenario direction is stable.
3) Model Complexity (Combinatorics)
Complexity increases the workload of the optimizer and post-processing.
Complexity grows with:
-
Number of tasks
-
Logistics units
-
Bases
-
Personnel groups
-
Constraints
More combinations create more scheduling alternatives for the optimizer to evaluate.
Important Notes
-
Shoresim’s optimizer intelligently prunes the search space and stops once an optimum is found.
-
Simulation time often grows sublinearly as models scale.
-
However, post-processing must iterate over all generated events, meaning post-processing time grows more directly with model size.
In short:
-
Optimization cost is managed.
-
Post-processing cost scales more predictably with model size.
Recommended Settings by Project Phase
Experimentation Phase
Goal: Narrow solution space and gain directional insight.
Recommended configuration:
-
5–10 simulated years
-
5–10 runs
-
Simplified model complexity
-
Accept some statistical noise if the directional signal is clear
This phase should be fast and iterative. Avoid spending full lifecycle sim minutes while still exploring.
Deep Analysis Phase
Goal: Quantify performance with statistical confidence.
Recommended configuration:
-
Full model complexity
-
Full simulation horizon (if lifecycle results are required)
-
Enough runs to meet a defined statistical precision target
At this stage, runtime investment is justified.
Delivery Phase
Goal: Produce decision-grade, defensible results.
Recommended configuration:
-
Fixed simulation duration across all scenarios
-
Fixed number of runs across all scenarios
-
Sufficient runs to produce confidence intervals and defensible claims
Consistency is critical at this stage.
Key Warning
Do not spend delivery-level sim minutes while still exploring.
-
Be fast and aggressive early.
-
Be statistically disciplined late.
Managing sim minutes well is not about minimizing runtime — it is about matching runtime investment to decision maturity.