Choose Compatible Time Steps​#

Use this page when a simulation rejects how often you want a model to run, or when several models need different time steps. To choose how those models share values between updates, start with Different model cadences.

Find a common base step​#

The simulation advances by a fixed interval, called the base step. Set it with the environment's duration. A model's every value must be a positive whole number of base steps. PlantSimEngine does not add smaller steps for you.

A smaller step may also work, but increases the number of simulation steps. Choose one that suits the equations and your weather data. If you need values between measurements, choose how to estimate them. If you need less frequent values, choose how to combine the measurements, for example by averaging. Changing duration alone does not calculate these values.

Example: hourly and 90-minute sampling​#

Two simple models read temperature at different intervals. Both run at the start of the simulation, then one runs every hour and the other every 90 minutes:

using PlantSimEngine, Dates, DataFrames
using PlantSimEngine.Examples

model = CompositeModel(
    Object(:hourly; scale=:Sensor, name=:hourly),
    Object(:ninety_minutes; scale=:Sensor, name=:ninety_minutes);
    applications=(
        ModelSpec(ToyEnvironmentReaderModel(); name=:hourly,
            on=One(name=:hourly), every=Hour(1)),
        ModelSpec(ToyEnvironmentReaderModel(); name=:ninety_minutes,
            on=One(name=:ninety_minutes), every=Minute(90)),
    ),
    environment=(T=20.0, duration=Minute(30)),
)
simulation = run!(model; steps=7, outputs=:all)
rows = collect_outputs(simulation; sink=DataFrame)
combine(groupby(rows, :application_id), nrow => :samples)

The seven time points cover 0 to 3 hours. The hourly reader records four results, and the other records three. Both record their first result at time zero.

Check when models run and where inputs come from​#

Use these tables to check the configuration:

• Diagnostics.explain_schedule(model) shows when each model starts and how often it runs.

• Diagnostics.explain_bindings(model) shows which model supplies each input and how earlier values are used, including the rule and time window.

• Diagnostics.explain_environment_bindings(model) shows the weather or spatial data each model reads and how values are combined over time.

Setting every replaces the model's default cadence from timespec. If you leave the cadence to the environment's base step, the model can check it with timestep_hint. If you set every yourself, you must check that the equations support that interval.

Fixed periods such as Day(1) are supported. Calendar months vary in length, so Month(1) is rejected. A time window looks back over its specified duration. For example, Day(1) does not automatically mean the preceding midnight-to-midnight day.