Second generation tap optimisation algorithm for the eWaterPay project.
See the results here: https://conorwilliams.github.io.
Verify your python and pip distributions are correctly installed. You can test this by running python -V and pip -V. On some machines with both python 2 & 3 these commands are replaced with python3 -V and pip3 -V respectively.
On windows you will need a terminal with root access to install Taptimise i.e you must search cmd into your search bar, right click and press "run as administrator".
If Git is installed and accesable from the command line (test with git --version). Install Taptimise using pip with: pip install git+https://github.com/ConorWilliams/taptimise verify the installation by running: taptimise -V.
Press on the "Clone or download" button and "Download ZIP". Extract the ZIP archive. cd into the extracted folder (probably taptimise-master) and install Taptimise with pip install . (including the .) verify the installation by running: taptimise -V.
Still within the extracted Taptimise folder, to run a full test on one of the example villages run taptimise test/e1.csv --tap-capacity 1000 and view the generated report test/e1_report.html in your web browser.
To run Taptimise with the default setting use:
taptimise path/to/file.csv --tap-capacity tap_capacity. The csv containing the house positions
should be formatted exactly as per the
example csv's.
I.e comma no space, newline separates houses. The path to the csv file can be
relative or absolute. Taptimise should produce a report
(pat/to/file_report.html) containing the optimised tap positions.
If using scribble maps .csv pass the flag --scribble x where x is the daily amount of water consumed PER house. Taptimise will then compute the required number of taps automatically.
Taptimise can accept several command line flags to tweak the optimisation.
To get a complete list of the avaliable flags run taptimise --help.
Setting a maximum separation with -m will trigger automatic reruns each using more taps until a solution is found. This can take a very long time.
Increasing the number of simulation steps with -s will improve the result at the expense of longer compute time.
An effective 'auto' mode can be enabled by passing the flags --steps 50 --scales 20. This will cause taptimise to loop through its cooling stage until it detects a stationary state. This is useful if you don't know how many Monte-Carlo steps to use and you don't want to risk overestimating.
Setting the overload fraction with -o much less than 1.15 will cause non deterministic results.