biketour - plan your bike trip

This tool allows to choose a perfect day to go for a long bike trip.

What is this all about?

Imagine you want to make to long cycle trip in one day. Say, you want to cover over 200 miles by travelling, say, from Aachen to Calais.

As an experienced cyclist, you know that there are several obstacles, that can harden your journey: climbs and wind. You cannot do anything about climbs (luckily, from Aachen to Calais there are non), but wind can be a real trouble-maker. Especially, heading West in West Europe, strong wind from Atlantic can slower you down.

Can we compute how bad the actual wind and weather situation is? This tool is designed precisely for that job.

Physical model

Currently, the physical model being used in calculations is borrowed from the bicycle performance Wikipedia page. The model is used to compute the speed of a cycle at any point along the route given a rider cycle power input. The computed speed at any time point along the route allows to compute a time plan for the trip.

The following physical parameters are used in the model:

• input power of the rider in Watts. Currently, the model assumes a constant power input along the journey. This can be easily generalised to a other power input models (unfortunately, I don't have a powermeter and have no idea how rider power changes along the journey).

• total mass of the rider in kg.

• wind speed and wind bearing. The wind speed and direction is used to compute power needed to overcome the aerodynamic force.

  Here is used a simple model, assuming a fixed drag coefficient. In reality, the drag coefficient should depend on the wind direction. There are plans to extend the model that takes it into account.

• drag coefficient. A parameter that hard to measure physically. As said above, the parameter is assumed to be fixed.

• air pressure, temperature and relative humidity. Those parameters are used to compute the air density, that affects the power to overcome the air drag.

• rolling resistance coefficient. This parameter is affected by your tires, air pressure in them, the wheel weight and the quality of the bearings you have.

• drivetrain efficiency. This parameter describes how efficient your drivetrain is. Assuming to be fixed along the trip.

Weather data source

For computing the trip time plan, we need to know the weather conditions at each point along the route. Fortunately, there is a good service that can provide us with that sort of data.

The service is called Dark Sky and data is queried from them. To make if work you need to get an API-key.

The service is not free, but it is quiet cheap. I queried historical weather data for 500 different days in the last 30 years for the route from Aachen to Calais in ~20000 query requests. That costs ~2 EUR.

Accuracy of the model; those silly physical parameters

It is not claimed that the implemented physical model gives accurate estimates for your journey longevity. But I believe that it should correlate with the true values.

The accuracy of the model depends on the weather data accuracy and the accuracy of the physical parameters you have provided.

It is planned to have a feature that estimates your physical parameters given a real journey trip (a data with coordinates, elevation and input power).

The route coordinates

The cycling routes are generated by cycle travel service. However, the program can take any gpx file. Elevation data is preferable, but not obligatory.

Application example: Aachen -> Calais

The route from Aachen to Calais is approximately 228 miles. The queried historical weather data allows to compute the travel time plans. The trip longevity distribution based on 500 historical data point is shown in the figure below. The historical weather is queried for 150 days of the summer months during the last 20 years.

Now, in order to decide which day to pick for a travel, I compute plans using the current weather forecast for the next few days and see how good it is with respect to the observed historical weather distribution.

Suggestions and critic

Suggestions, critic (raise an issue) and pull requests are welcome!

The program is written to satisfy my needs to ease my double century ride. Currently, it can be used only as a python library. There is even no command-line interface. However, this is planned in future.

ToDO features:

• [DONE] query historical weather database

  for a given route create and populate database with weather historical records along the journey

• [DONE] physical model based on a constant power output

  Implement a physical model that assumes that a cyclist cycles with a constant power. This model takes into account wind, temperature, pressure, humidity, climbs, efficiency of drive_train, etc

• [DONE] compute a trip plan

  The trip plan is computed using a physical model

• [DONE] compute plan characteristics for different data

• estimate physical model parameters using the travel data

  yet to be implemented: the travel data should include power-meter measurements as well as gps coordinates, time and elevation

• compute how good the next days forecast for taking the journey

• include regular pauses in a ride plan

• graphs, plots

• command line interface, GUI?

Attribution

The weather data is Powered by Dark Sky.

The cycling routes are generated by cycle.travel.