Time & Distributions
TLDR: Entry and Processor nodes share the same timing controls. Set a simple rate, fit real data automatically, define a time-of-day schedule, or replay exact timestamps. This page covers all four modes.
Simple Mode
The default mode. You type a single number and pick a time unit.
On an Entry node
You see: "[number] arrivals per [second / minute / hour]"
This sets how frequently new agents appear. For example, 12 arrivals per hour means one new agent roughly every 5 minutes.
On a Processor node
You see: "Each takes about [number] ± [variation] [seconds / minutes / hours]"
The first number is the average service time. The ± value adds natural variation — real tasks are never perfectly consistent. For example, 5 ± 2 minutes means most agents take 3–7 minutes to process.
The variation value controls how spread out service times are. A small ± means almost every agent takes close to the average. A large ± means some agents will be much faster or slower. Set it to 0 for perfectly consistent times (like a machine cycle).
Fitted Mode
If you have real data (timestamps, durations, or a list of numbers), click "Have real data? Fit my arrivals" (Entry) or "Have real data? Fit my timing" (Processor).
This opens the Distribution Detective, which:
- Accepts pasted data (one value per line, or comma-separated)
- Tests all 9 distribution types against your data
- Picks the best fit automatically and shows a fit score
- Lets you view ranked alternatives in the Technical Details drawer
Once fitted, the timing card changes to show:
- A green checkmark with "Using your arrival data" (or "Using your timing data")
- How many data points were used (e.g., "Based on 247 timestamps")
- A "Re-analyze" button to open the Detective again
- A "Use simple rate instead" link to go back to Simple mode
Fitting real data always produces more accurate simulations than guessing a rate. If you have historical records, use them.
Schedule Mode
Entry nodes only.
Click "Arrivals vary by time of day?" to open the arrival schedule editor. This lets you define different arrival rates for different time periods — perfect for modeling morning rushes, lunch peaks, or overnight lulls.
The editor shows a grid of time periods. The period length scales automatically based on your simulation duration:
- Short simulations: periods in minutes
- Medium simulations: periods in hours
- Long simulations: periods in days
Type the number of arrivals you expect in each period. Leave a period at 0 for no arrivals (e.g., overnight closure).
Replay Mode
Entry nodes only.
When you fit timestamp data through the Distribution Detective, you may be offered a Replay option. This replays your exact timestamps rather than generating random arrivals from a fitted distribution.
Use Replay when:
- You want to recreate a specific historical day exactly
- You're validating your model against known outcomes
- The arrival pattern is too complex for any distribution to capture
Distribution Types Reference
When the Distribution Detective fits your data, it tests these 9 distributions. All are available to every user.
| Distribution | Shape | Good For |
|---|---|---|
| Deterministic | Exactly the same every time | Machine cycles, fixed intervals |
| Exponential | Many short, few long | Random arrivals (customers, calls, orders) |
| Normal | Bell curve around the average | Human task times, natural variation |
| Uniform | Equally likely anywhere in a range | When you only know min and max |
| Triangular | Peak at most likely, with min/max bounds | Expert estimates ("usually 5, but between 2 and 10") |
| Lognormal | Skewed right — most short, tail of long | Repair times, hospital stays |
| Weibull | Flexible shape (increasing or decreasing hazard) | Equipment failure, reliability |
| Gamma | Sum of multiple exponential stages | Multi-step processes, total wait times |
| Beta | Bounded between min and max | Percentages, completion rates |
In Simple mode, the system handles distribution selection internally. In Fitted mode, the Detective picks for you. You only need this table if you're reviewing Technical Details or writing Custom node code.
Time Units
The time unit dropdown appears on Entry and Processor timing cards. Available units:
- Seconds — for fast processes (machine cycles, automated steps)
- Minutes — most common for service operations
- Hours — for long processes (hospital stays, shipping)
All values are converted internally so you can mix units across nodes. An Entry generating arrivals "per hour" feeding a Processor measured in "minutes" works seamlessly.