Set Up and Run a Simulation

What files are needed for a simulation?

Scenario script

Start the simulators, build the models, create the connections and start mosaik. Exemplary scripts for building, grid etc. can be found in the examples folder. The setup of these scripts is explained below in the Scenario Script Configuration section.

Model data file

Information on number of models and their parameterization.

This is stored as a dict with fields for each model type, named by the class of the model. E.g. the field “HouseholdCSV” contains a list of all household baseloads that read from a csv file. In the example below there is just one baseload, and if more of these should be integrated, we could add a second dict with initialization parameters for this model inside of the given list.

One exception is the Energy Management System (EMS), as they are all sorted within the “ems” section. Their differentiation is coming from the field “strategy”, which is referring to the chosen operating strategy of the EMS.

examples/data/model_data_scenario/model_data_building.json (01/24)

{
   "ems": [
      {
            "strategy": "HEMS_default",
            "cs_strategy": "balanced"
      }
   ],
   "HouseholdCSV": [
      {
            "p_rated": 4500,
            "p_rated_profile": 4000,
            "cos_phi": 1.0,
            "datafile": "examples/data/load/4_persons_profile/load_34.csv",
            "date_format": "YYYY-MM-DD HH:mm:ss",
            "header_rows": 2,
            "start_time": "2014-01-01 00:00:00"
      }
   ],
...

Model connections

Information on the linking of model entities in the simulation. This includes the connections between grid buses, ems models, and the prosumer devices. It is stated, which devices are located at which grid connection point (by the naming of the corresponding load bus) and whether there is an EMS used. If no EMS is given at a loadbus (see “0-load_1_2” below), the devices are directly connection to the grid, but this should only be possible for some of the devices, as e.g. a single heat pump makes no sense.

examples/data/grid/grid_model_config.json (01/24)

{
   "0-load_1_1": {
      "ems": "HEMS_default_0",
      "load": [
            "HouseholdCSV_0"
      ],
      "household_thermal": [],
      "pv": [],
      "bss": [],
      "hp": [],
      "cs": [
            "ChargingStation_0"
      ],
      "ev": [
            "EV_0"
      ]
   },
   "0-load_1_2": {
      "ems": "",
      "load": [
            "HouseholdCSV_1"
      ],
...

Grid file

In .json format (possibly created via pandapower). Provides detail on the given buses and their connections via grid lines (plus their parameterization).

examples/data/grid/example_grid_kerber.json (01/24)

{
  "_module": "pandapower.auxiliary",
  "_class": "pandapowerNet",
  "_object": {
    "bus": {
      "_module": "pandas.core.frame",
      "_class": "DataFrame",
      "_object": "{\"columns\":[\"name\",\"vn_kv\",\"type\",\"zone\",\"in_service\"],\"index\":[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17],\"data\":[[\"Trafostation_OS\",10.0,\"b\",null,true],[\"main_busbar\",0.4,\"b\",null,true],[\"MUF_1_1\",0.4,\"n\",null,true],[\"loadbus_1_1\",0.4,\"b\",null,true],[\"KV_1_2\",0.4,\"b\",null,true],[\"loadbus_1_2\",0.4,\"b\",null,true],[\"MUF_1_3\",0.4,\"n\",null,true],[\"loadbus_1_3\",0.4,\"b\",null,true],[\"KV_1_4\",0.4,\"b\",null,true],[\"loadbus_1_4\",0.4,\"b\",null,true],[\"MUF_1_5\",0.4,\"n\",null,true],[\"loadbus_1_5\",0.4,\"b\",null,true],[\"KV_1_6\",0.4,\"b\",null,true],[\"loadbus_1_6\",0.4,\"b\",null,true],[\"MUF_2_1\",0.4,\"n\",null,true],[\"loadbus_2_1\",0.4,\"b\",null,true],[\"KV_2_2\",0.4,\"b\",null,true],[\"loadbus_2_2\",0.4,\"b\",null,true]]}",
      "orient": "split",
      "dtype": {
        "name": "object",
        "vn_kv": "float64",
        "type": "object",
        "zone": "object",
        "in_service": "bool"
      }
    },
...

Tip

All files can be created (more easily) with a Scenario Configurator (.ipynb) via an excel file. Or use existing files and adapt the parameterization.

Configuration of a Scenario Script

Note

All of these code-blocks derive from examples/test_scenario_building.py as of (01/24) if not stated otherwise.

Setup

import of used packages (examples, like mosaik or logging during the simulation)

import os
import mosaik
import eelib.utils.simulation_helper as sim_help
from eelib.model_connections.connections import get_default_connections
import logging

Setting of paths for simulation data and used model simulators

# define paths and filenames
DIR = sim_help.get_default_dirs(
    os.path.realpath(os.path.dirname(__file__)), scenario="building", grid=None, format_db="hdf5"
)

Define simulators and models for the simulation

# Sim config.: Simulators and their used model types with the properties to store into DB
SIM_CONFIG = {
   # used database, will be left out for model creation and connections
   "DBSim": {"python": "eelib.data.database.hdf5:Hdf5Database"},
   # all the used simulators and their models for this simulation
   "EMSSim": {
      "python": "eelib.core.control.EMS.EMS_simulator:Sim",
      "models": {"ems": ["p_balance", "q_balance", "p_th_balance", "p_th_dem"]},
   },
   "CSVSim": {
      "python": "eelib.data.csv_reader.csv_reader_simulator:Sim",
      "models": {
            "HouseholdCSV": ["p", "q"],
            "PvCSV": ["p", "q"],
            "ChargingStationCSV": ["p", "q"],
            "HeatpumpCSV": ["p_el", "q_el"],
            "HouseholdThermalCSV": ["p_th_room", "p_th_water"],
      },
   },
   ...

This is done in a format that fits both mosaik for orchestrating the simulation and also the handling of data and simulation setup. So in the “SIM_CONFIG” dict we first need to give all used simulators as the keys (e.g. “CSVSim” for all used csv readers). With the sub-key “python” we then provide the path to this simulator in our library (e.g. “eelib.data.csv_reader.csv_reader_simulator:Sim”). Under “models” we list all of the used model types within the simulation for this simulator (e.g. “HouseholdCSV”, “PvCSV”) and the attributes of them that should be stored within the database (e.g. [“p”, “q”] for active and reactive power of the PV system).

Configure start and end time of the simulation, step length, and whether to use forecasts. Model data and connections are read and SIM_CONFIG is handed to mosaik.

# Configuration of scenario: time and granularity
START = "2020-01-01 00:00:00"
END = "2020-01-04 00:00:00"
USE_FORECAST = False
STEP_SIZE_IN_SECONDS = 900  # 1=sec-steps, 3600=hour-steps, 900=15min-steps, 600=10min-steps
N_SECONDS = int(
    (
        arrow.get(END, "YYYY-MM-DD HH:mm:ss") - arrow.get(START, "YYYY-MM-DD HH:mm:ss")
    ).total_seconds()
)
N_STEPS = int(N_SECONDS / STEP_SIZE_IN_SECONDS)
scenario_config = {
    "start": START,  # time of beginning for simulation
    "end": END,  # time of ending
    "step_size": STEP_SIZE_IN_SECONDS,
    "n_steps": N_STEPS,
    "use_forcast": USE_FORECAST,
    "bool_plot": False,
}

# Read Scenario file with data for model entities
with open(DIR["MODEL_DATA"]) as f:
    model_data = json.load(f)

# Read configuration file with data for connections between prosumer devices
model_connect_config = get_default_connections()

# Create world
world = mosaik.World(SIM_CONFIG, debug=True)

With this mosaik.World we start the orchestrator mosaik and afterwards set up the simulation using mosaik. For this, we mostly use mosaik functionalities within the helper functions of the eelib.utils.simulation_helper folder. Regarding the order, we mostly first handle the exceptions of the models, like the output database, and afterwards use the helper functions for all of the other models. Fore a deeper understanding of what is done with these steps, you should have a look at the implementation of these helper functions and possibly even the setup of mosaik for creating the simulation.

Start Simulators for all used Models

# start all simulators/model factories with mosaik for data given in SIM_CONFIG
dict_simulators = sim_help.start_simulators(
    sim_config=SIM_CONFIG, world=world, scenario_config=scenario_config
)

Initiate Model Entities using the created Simulators for each Device Type

# create all models based on given SIM_CONFIG
dict_entities = sim_help.create_entities(
    sim_config=SIM_CONFIG, model_data=model_data, dict_simulators=dict_simulators
)

Connect Entities

The connections for each entity are listed in the model_connect_config data. Now tell mosaik about the and let mosaik create graphs for how the calculation procedure of the simulation is to be executed.

# connect all models to each other
sim_help.connect_entities(
    world=world,
    dict_entities=dict_entities,
    model_connect_config=model_connect_config,
    dict_simulators=dict_simulators,
)

Run Simulation

world.run(until=scenario_config["n_steps"], print_progress=True)

Just go Ahead and Run a Simulation

You can run one of the test_scenario s in the examples folder
building: Just one single building with a bunch of different devices to see the operation of devices inside the household.

grid: Simple low voltage grid (2 feeders with six resp. two household connection points) to get an estimation of the impact of different operating strategies on the local grid.
While running the simulation, some stuff is logged like creation of simulators, entities, and connections. Additionally, mosaik plots the progress of the whole simulation (like how many time steps are finished calculating).
Adapting the parameterization in the simulation files can yield quite different results.
Running one of the simulations will create a .hdf5 results data file in the folder /examples/results.
You can view the information of this file via the H5Web Extension in Microsoft VSC and plot the profiles (stored under Series and the name of the corresponding simulator) of the used devices.
In case you want, you can set the parameter bool_plot in the scenario_config dict to true ( only do this for small simulations!), mosaik will then present 4 plots for the connections, the simulation process and the timely duration during the simulation.

Create your own simulation

Copy one of the test scenarios and delete all of the redundant simulators/devices/connections.
Set up corresponding model (and grid) data as well as a model connection file.