Backtracking Strategy

Mutual shadings

The layout of tracker arrays should be carefully optimized with respect to mutual shadings. Constraints are much more critical than for the sheds disposition, as a significant yield may be expected even when the sun is very low on the horizon.

The mutual shading problem is accentuated by the electrical behavior of the strings under partial shadings. As for sheds arrangements, identical shadings appear simultaneously on each tracker and may block the yield of many strings at a time.

Backtracking strategy

Backtracking is now a widely used strategy for tracking arrays: when the mutual shadings begin, the tracking angle does not follow the sun anymore, but it instead goes back (decreases) so that no shading occurs.

Let's analyze the case of a simple Horizontal N/S axis system.

When the backtracking is activated, the collector plane is not perpendicular to the the sun profile angle anymore; we have a loss of irradiance, due to the cosine (or transposition) effect.

Backtracking angle calculation

The collector tilt angle is closely linked to the relative parameters between two neighboring rows of trackers.

In PVsyst, the tracking angle is computed at each time step for a reference definition of the collector width and pitch, or more exactly of the width/pitch ratio. This tracking angle is applied identically to all trackers of the scene. If all the trackers of the arrays do not have an identical width/pitch ratio, the backtracking becomes inefficient:

• A lower pitch will induce mutual shadings, and therefore possible electrical shading losses,
• A higher pitch will start the back-tracking before it is necessary, therefore losing some irradiance via the cosine effect.

If you have an occasionally larger pitch between some similar arrays (e.g. passageways between groups), you will not have mutual shadings. However the backtracking is necessary for all trackers, including the extremity ones, due to mutual shadings.

If you have adjacent arrays, and the trackers of both groups are staggered, you may have mutual shadings from one group to the other one in some seasons.

If you have height differences between trackers, a "pure" backtracking is not possible, nether in PVsyst, nor in the reality. See "Backtracking on Hill".

NB: With systems imported from other software, the trackers are usually all independent. PVsyst has to check the pitches uniformity, and choose a pair of trackers as reference.
You have a tool for this in the 3D editor "Tools > Backtracking management"

Other Tracking configurations

We have just explained the backtracking for horizontal axis.

Similar ideas can also be applied to other tracking configurations. However the analytic calculation becomes sometimes very difficult.

Presently, PVsyst has elaborated algorithms and proposes the backtracking for:

• Horizontal N/S axis, as explained above.
• Tilted axis, as far as there is no misalignment between trackers of a group (i.e. "rectangular" arrays).
• Sun-shields: when the sun is high in the sky, the sun shields become highly tilted. This is also the case for sun's east/west orientations. The compatibility with comfort conditions has to be studied.
• Two-axis frames: the backtracking has been implemented only for the angle of the tables tilt within the frames, i.e. a backtracking from table to table. There is no backtracking between frames.
• Two-axis trackers: This is a very difficult problem. There may be several backtracking strategies.
  • In the present time, the backtracking mechanism between neighbor trackers is supposed to be a tilt of the plane following the sun's height, and a rotation around the vertical axis ensuring no mutual shadings. It doesn't involve passing to the north of the East-west line. It doesn't take the shadings from one row to the other into account.
  • Even if implemented, nothing ensures that this will be used identically by the control system on the field !

Performance calculations

The backtracking strategy avoids mutual shadings between trackers, for the beam component.

However the diffuse and albedo received by the trackers are affected by the shadings of the neighbors. Namely the albedo is completely lost except for the first and last trackers. This explains that even with backtracking, you always have a Near shading loss in the final results. This is usually of the order of 2-3%.

It should be noted that the Backtracking doesn't increase the total irradiance received. It only improves the electrical loss effects of the shadings. The total irradiance reaching the modules is the same as if there were shadings: it corresponds to the total sun energy intercepted for this given sun direction, by the field area "seen" from the sun. Therefore a simulation with "Linear shadings" (not electrically realistic) and another one with backtracking should give about the same results. See Backtracking performance.