Backtracking parameter tuning

Backtracking parameter tuning means refining a single-axis tracker's backtracking settings. On rolling sites the settings are adjusted for each terrain zone so that recoverable energy is not lost. It is a listed mitigation for Terrain Undulation Yield Losses. For Performance Loss from Tracker Twist, checking the backtracking settings is one step of the repair sequence. Backtracking assumes a fixed geometry between row pitch, tracker angle, module plane and the horizon formed by neighbouring rows. Uneven ground disturbs that geometry, so a row may shade its neighbour or turn further off sun than it needs to. A tracker that follows its command is not necessarily performing well. On undulating ground the command can itself cause the loss when the terrain model, the backtracking algorithm or the motor block layout is too coarse. One way to check the settings is to compare the ideal tilt angle that monitoring software calculates from the tracker configuration, including ground coverage ratio, against the tracker's own setpoints and positions. A separate west-side ratio can tune the afternoon and evening backtracking when it differs from the morning, for example on a hill or where row spacing differs east and west. Losses usually repeat with sun position and are largest at low sun angles. Compare normalized row or inverter output in the morning and afternoon windows before and after tuning. Tuning has limits. A large motor block can force one command angle onto rows that need different backtracking positions, so check motor block sizing as well. Where twist is suspected, check the mechanical geometry before blaming the controls. After any repair, repeat the angle survey and compare normalized power against unaffected rows.

<!-- review notes

• sentence 6: support check: partial - Repeatability by sun position, strongest at low sun, and comparing normalized output across morning/afternoon windows are stated, but 'before and after tuning' is not explicitly stated.
• sentence 8: support check: partial - u11 supports verifying mechanical geometry before blaming controls and repeat angle survey with normalized power comparison against unaffected rows, but it doesn't mention twist suspicion specifically and adds 'under stable irradiance' condition omitted here. -->