The Challenge
Row spacing in utility-scale solar is not a shading parameter. It is a financial engineering decision that simultaneously affects DC installed power, shading losses, cable lengths, and annual energy production.
On a 130 MWp plant with a north-facing slope, the design team needed to evaluate how three spacing scenarios affected these competing variables across a single 5 MWe transformer area. The standard approach of using a single GCR and simulating in PVsyst could not capture the terrain-driven variation.
Design Constraints
All three scenarios held the following parameters constant:
- Same transformer, inverter, and DC combiner box positions
- Same module type, string power, and electrical configuration
- Same site boundary and terrain data
- Shading analyses used identical date and sun position assumptions
- Cable routes calculated in AutoCAD using actual site corridors (not bird’s-eye distance)
Only the row spacing parameter changed: 3.5 m, 4.5 m, and 5.5 m.
Three Spacing Scenarios
3.5 m Spacing (Maximum Density)
The tightest configuration maximized DC installed power at 5.48 MWp within the transformer area. The compact layout produced the shortest total cable length at 55,472 m and lowest cable CAPEX at $1,386,800.
The tradeoff: inter-row shading reached 20.0%, the highest of all three scenarios. Annual energy loss from shading: 1,862 MWh.
Cross-section verification confirmed that at 3.5 m spacing on a north-facing slope, front rows cast significant shadows on rear rows during critical sun angles, particularly around the winter solstice.
4.5 m Spacing (Balanced)
Increasing spacing to 4.5 m reduced shading loss to 14.5% (a 27.5% improvement). DC installed power dropped to 5.00 MWp (8.8% less than 3.5 m) as fewer rows fit within the same boundary.
Cable length increased to 56,741 m as the layout spread out, and cable CAPEX rose to $1,418,500. Annual energy loss from shading fell to 1,233 MWh, saving 629 MWh/year versus the 3.5 m scenario.
The 4.5 m scenario represented the engineering balance point: moderate shading reduction without excessive capacity or cable cost penalty.
5.5 m Spacing (Minimum Shading)
The widest configuration pushed shading loss down to 10.8% (a 46% improvement versus 3.5 m). Annual energy loss from shading fell to 865 MWh, the lowest of all scenarios.
DC installed power dropped to 4.71 MWp (14.1% less than 3.5 m). Cable length reached 59,685 m and cable CAPEX hit $1,492,100.
PVsyst shading analysis confirmed that at 5.5 m on a north-facing slope, inter-row shadow effects at critical dates and times were minimal.
Full Comparison
| Spacing | DC Power (MWp) | Shading Loss (%) | Cable Length (m) | Cable CAPEX ($) | Annual Energy Loss (MWh) |
|---|---|---|---|---|---|
| 3.5 m | 5.48 | 20.0% | 55,472 | $1,386,800 | 1,862 |
| 4.5 m | 5.00 | 14.5% | 56,741 | $1,418,500 | 1,233 |
| 5.5 m | 4.71 | 10.8% | 59,685 | $1,492,100 | 865 |
DC Power Change
| Spacing | DC Power | Change from 3.5 m |
|---|---|---|
| 3.5 m | 5.48 MWp | Reference |
| 4.5 m | 5.00 MWp | -0.48 MWp (-8.8%) |
| 5.5 m | 4.71 MWp | -0.77 MWp (-14.1%) |
Shading Loss Change
| Spacing | Shading Loss | Reduction |
|---|---|---|
| 3.5 m | 20.0% | Reference |
| 4.5 m | 14.5% | -27.5% |
| 5.5 m | 10.8% | -46.0% |
Key Findings
- Row spacing choice swings shading loss by 46% between the tightest and widest scenarios (20.0% to 10.8%).
- Annual energy loss difference: 997 MWh between 3.5 m and 5.5 m spacing.
- Cable CAPEX spread: $105,300 across scenarios for a single transformer area.
- 4.5 m spacing represented the optimal balance between shading performance, installed capacity, and cable cost for this north-facing site.
- DC cable routes were calculated along real site corridors including trench lines and cable turns, not theoretical straight-line distances.
- The shading-spacing tradeoff is terrain-dependent. Flat-field simulation tools miss the slope-driven component of inter-row shading that PVX.Cad captures.
- Row spacing is simultaneously a shading decision, a capacity decision, and a cable CAPEX decision. Optimizing for one variable without modeling the other two produces suboptimal outcomes.