Interpretation of speed table

[pwesolek]

Hello,

I'm currently in a process of tuning the matrix settings for our customer needs and have a question concerning speedTable parameter.

What is the intention of having min and max speed for each segment category? I understand the possibility of interval calculations, but in the final run, for each pair of locations we obtain a single number as a value of travel time. How is this value obtained in regard to the configured interval?

Of course I can use the min=max for singularity, but maybe there is another business value in the interval I'm not seeing, which can be of importance to our customer?

Due to our usage of high performance network it is unfortunately efficient to experiment with the interval, as the speed table is "baked into" the HPRN structure obtained in the long-running process...

[Bernd Welter]

Hello Przemyslaw,

I attached an image that explains the mechanism behind the static speed values.

Each one of the 8 so-called NETWORK CLASSES is divided into 8 sub-categories labeled as SPEED CLASSES. So the complete set of street segments of a map can be cut into 64 independent sets.
Defining the MAX and MIN speed for 2 of the 8 speed classes enables us to interpolate the other 6 classes values.

So when our data department receives the raw data from the providers TOMTOM and HERE we map the input data into PTVs proprietary structure and during this process we determine the network and speed classes for all the segments. While some countries have low limits (such as France, limited to 130) others have higher speed classes (at least partial) such as Germany (STRIKE ).

I'm not sure how to understand the other parts of the post? Could you explain more in detail what you need?

Best regards Bernd

[pwesolek]

Hi Bernd,

Thank you very much for the instructive infographics! I didn't know there are 8 speed classes used internally!

As to my general question, I want to know, how the travel time is calculated for a route/map segment.

Previously I assumed that underlying map exposes only NC (network class) for segments, and xServer somehow (magically) mapped [min;max] interval of the specific NC to a driving time along the map segment.

Do I understand correctly, that the segment information from the map is more robust and also provides SC (speed class) of each segment? And this allows for a concrete speed calculation? For example if a segment has NC=2 and SC=2, the speed for this segment will be something like 25 + (85-25)*5/7 = 68 (i.e. somewhere between 25 for SC=7 and 85 for SC=0, proportionally to SC value).

[Bernd Welter]

There you go!

Shows the interpolation

Shows the computation of travel time and virtual costs

After all this explains how we compute a best route: it is the path that produces the least amount of costs!

[pwesolek]

Thank you very much, that answers all my questions!