Flexible Load Simulator

Flexible Load Simulator (FLS) for the Automatic Trading Platform (ATP) of the GOFLEX Project. FLS allows to create and manage a variety of simulated loads.

Note: the following commands should be executed from within the root of this repository.

API

Detailed documentation of the API can be found by navigating to the root of the server. For example, if server is running on port 5000 of localhost, then API docs can be accessed by navigating to http://localhost:5000.

Generate Static API Docs

# generate static api docs which will be copied to the docker image during next command
cd apidocs && bundle exec middleman build --clean && cd -

Installation

We describe how to setup and launch FLS using a)- docker, b)- from scratch.

Installation using docker

It's very easy to setup and run the app with docker. All you need to do is build a docker image, then either run it in a container locally or push the image to a remote server and run it there. You must install docker and docker-compose both locally and on remote server if not installed already.

• Install Docker on Ubuntu 16 or Ubuntu 18

• Install Docker Compose

  sudo curl -L "https://github.com/docker/compose/releases/download/1.23.1/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
  sudo chmod +x /usr/local/bin/docker-compose
  docker-compose --version

Build docker image

docker-compose build

Push docker image to a remote private docker registry

# one-time login to private docker registry
docker login repo.treescale.com # it will ask for username and password

docker-compose push

On remote server, pull and run docker image from private registry

# copy docker-compose.yml to remote server
scp docker-compose.yml <username>@<remote-server-address>:/remote/dir/path

# login to remote server and navigate to the directory containing docker-compose.yml
ssh <username>@<remote-server-address>
cd /remote/dir/path

# one-time login to private docker registry
docker login repo.treescale.com # you will need to install docker if not installed already

docker-compose pull

Alternatively, you can perform one-time login to the private docker registry on the remote server as shown above, then directly deploy container from your local machine as follows:

# Note: you must be logged in to the private docker registry to run these commands
cat docker-compose | ssh <username>@<remote-server-address> "docker-compose -f - pull"
cat docker-compose | ssh <username>@<remote-server-address> "docker-compose -f - up -d"

Remove unnecessary docker images from remote server

ssh <username>@<remote-server-address> "docker system prune -f"

Run docker Image in a container

# launch a container by navigating to the directory containing docker-compose.yml and running this command
docker-compose -d up

# attach to the running container
docker attach fls-server-container

# detach from the running container
CTRL-p, CTRL-q

After the container is up and running, the FLS server can be accessed on port 5000.

Manually run with uwsgi

Start uwsgi from command line

uwsgi -s flsserver.sock --http 0.0.0.0:5000 --module main --callable app

Start uwsgi with ini file

uwsgi --ini uwsgi-http.ini

Installation from scratch

First, we need to install JModelica from source. From within the root of this repository, run the following commands to install JModelica into ~/JModelica.

cd scripts && bash installJModelica.sh /path/to/install/dir

When testing JModelica, you might get the following error

 AttributeError: KeepLastStream instance has no attribute 'close'

To fix this problem add the following method to KeepLastStream class in /path/to/JModelica/Python/pymodelica/compiler_logging.py (ref):

def close(self):
  self.stream.close()

Setup virtual environment

sudo pip install virtualenv
virtualenv .venv

Set JModelica environment variables in virtualenvironment. Please set JMODELICA_HOME to the correct install location.

tee -a .venv/bin/activate << END

# JModelica
export JMODELICA_HOME=/path/to/JModelica
export JAVA_HOME=/usr/lib/jvm/java-8-oracle
export IPOPT_HOME=\$JMODELICA_HOME/CoinIpoptBuild
export SUNDIALS_HOME=\$JMODELICA_HOME/ThirdParty/Sundials
export PYTHONPATH=\$PYTHONPATH:\$JMODELICA_HOME/Python
export LD_LIBRARY_PATH=\$LD_LIBRARY_PATH:\$IPOPT_HOME/lib:\$SUNDIALS_HOME/lib:\$JMODELICA_HOME/ThirdParty/CasADi/lib
export SEPARATE_PROCESS_JVM=\$JAVA_HOME
END

Also unset JModelica environment variables on deactivate

sed -i '/deactivate () {/a \
    # JModelica \
    unset JAVA_HOME \
    unset JMODELICA_HOME \
    unset IPOPT_HOME \
    unset SUNDIALS_HOME \
    unset PYTHONPATH \
    unset LD_LIBRARY_PATH \
    unset SEPARATE_PROCESS_JVM \
' .venv/bin/activate

Activate virtual environment

sudo apt install libmysqlclient-dev
source .venv/bin/activate

Install required packages

pip install -r requirements.txt

Create the required MySQL database. Run this inside MySQL shell

create database flsdb;

Usage

Start the simulator program

python app.py

# access over internet
python app.py --host 0.0.0.0

# full list of command line arguments
python app.py --profile prod --host 0.0.0.0 --port 5000 --db-url mysql://username:password@host:port/db --create-fmus

Here --host defaults to localhost, and --port defaults to 5000.

Tips

How to de-activate virtual environment

deactivate

Set environment variables in pycharm (because pycharm is a desktop app which is not a child of shell and env vars are defined in shell only)

JAVA_HOME=/usr/lib/jvm/java-8-oracle
JMODELICA_HOME=/home/username/JModelica
IPOPT_HOME=/home/username/JModelica/CoinIpoptBuild
SUNDIALS_HOME=/home/username/JModelica/ThirdParty/Sundials
PYTHONPATH=:/home/username/JModelica/Python
LD_LIBRARY_PATH=:/home/username/JModelica/CoinIpoptBuild/lib:/home/username/JModelica/ThirdParty/Sundials/lib:/home/username/JModelica/ThirdParty/CasADi/lib
SEPARATE_PROCESS_JVM=/usr/lib/jvm/java-8-oracle

Theory behind simulation of electric loads

The following appliance models have been implemented in Modelica:

Basic On-Off Models

1. On-off Model: This model has a certain fixed power consumption rate p_on when active.

2. Exponential Decay Model: The power consumption rate follows an exponential decay curve, dropping from the initial surge power p_peak to a stable power p_active at a decay rate λ.

   p(t) = p_active + (p_peak - p_active)e^-λt

3. Logarithmic Growth Model: The power consumption rate follows a logarithmic growth curve, starting with a level p_base and a growth rate λ.

   p(t) = p_base + λ · lnt

The table below lists several example loads for simulation in our app. Other loads can be crated by varying the parameter combination within the allowed range.

Load Name	Modelica Model	Parameters
Lamp	On-Off	pon=40
Toaster	Exponential Decay	ppeak=1470, pactive=1433, λ=0.02
Coffee Maker	Exponential Decay	ppeak=990, pactive=905, λ=0.045
Refrigerator	Exponential Decay	ppeak=650.5, pactive=126.19, λ=0.27
Air Conditioner	Logarithmic Growth	pbase=2120.46, λ=13.78
Heat Pump	SISO Linear System	A=-0.01, B=0.002, C=1, D=0

References

• http://sites.psu.edu/symbolcodes/codehtml/#math
• https://www.keynotesupport.com/websites/greek-letters-symbols.shtml
• https://github.com/miguelgrinberg/REST-auth