How does the backtracking strategy with 2-axis trackers work ?

[André Mermoud]

The backtracking for 2-axis trackers is a very complex problem, and there are several possible different strategies.

In the present time, the PVsyst algorithm is quite insufficient, and you are not advised to use it. It only performs the backtracking in azimuth (with tilt set according to the sun's height), and without going to the "north", i.e. blocking the azimuth to values < 90°.

When the sun goes to higher azimuths, the backtracking should switch to a backtracing in tilt (i.e. the next tracker sees "above" the preceding one). This is not done in PVsyst as we could not find a suited analytical algorithm.

At this stage, one of the problems is: when should the system switch between the azimuth and the tilt strategy? This will induce a discontinuity on the tracking orientation: at a given time, the trackers will suddenly pass from an orientation to another one.

With such a behavior, the system should come back to the Azimuth strategy (at values higher than 90°) symmetrically with respect to azimuth = 90° (West or East).

Now a second problem arises with multi-rows arrays. In this case we should also implement the backtracking between 2 rows (one behind the other), which even may have a shift in E-W direction. Mixing this with the E-W BT strategy becomes a diabolic problem.

We don't see any way of modelling all these behaviours analytically. The only solution would be to implement an ajustment by trys-and-errors, which may become time consuming during the simulation. We did not yet develop such an algorithm in PVsyst up to now.