So far I’ve shown a basic iso-IRR curve (in the earlier post “Low-efficiency penalty”) to get a sense for the penalty that low-efficiency modules have to pay. (The quick overview: this is a way of showing the tradeoff between efficiency and price, for the developer. Solar modules on the curve would provide the same financial returns for the project. Lower-efficient panels require more racking and labor, so they have to lower their price to be equally competitive.)

However, not all applications are the same. Specifically, it is a commonly-accepted claim that crystalline silicon is better suited to residential installations, while thin-film products are better for larger projects. **I am going to prove that quantitatively, using the IRR curves. **

The initial assumptions I used (for the curve above) were closer to a ground-mounted, utility-scale project. Smaller, residential projects have different economics. Specifically, they have higher fixed costs (the trouble of getting an installation team on the roof is larger, proportionally, than having a team show up at an open-field site). Also, the labor and racking costs are higher (higher-skilled labor, and more expensive racking to protect the roof). By changing these assumptions, we can see how it changes the shape of the curve. See below*:

The constraints of the residential system make it less forgiving to the lower-efficiency modules. One way to look at it: in my previous ground-mounted example, a 10% module paid a penalty of $0.30/watt versus a 15% module. In a residential equivalent, the penalty grows to $0.67. In other words, residential projects are more willing to pay for higher efficiency, as a way of minimizing the more expensive labor/racking, and covering the larger fixed costs.

So, ground-mounted systems value cheaper price over good efficiency, and residential systems are more willing to pay more for high efficiency.

As a final step, let’s overlay the market prices of current technologies. Again, a big caveat: all of these numbers are very rough, and prices will always change quickly. These are just directional values.

Technology |
Efficiency |
Cost ($/watt) |

Amorphous silicon | 7% | $1.50 |

Cadmium telluride | 10% | $1.75 |

Poly-crystalline | 15% | $2.50 |

Mono-crystalline | 18% | $3.25 |

Plotting these values on our graph, you can see the cost-performance horizon:

Theoretically, customers are always looking for something that is up and to the right on this graph (high efficiency and low cost). A ground-mount (large scale) customer is looking to optimize a curve that is shaped like the blue graph – and so you can see they will be more interested in using thin-film technologies. On the other hand, a residential customer, optimizing the pink curve, will be more interested in going with crystalline.

Finally: I’ve posted this model (you should see it on the box.net sidebar, “PV IRR Model II”). Feel free to download and play with it. Let me know what you think.

** Note on the precision of the numbers***: The numbers here are only for directional takeaways. I have not gotten exact quotes on anything, and I am dramatically simplifying the cost model. I am accounting for only six line items (the largest ones), while a real system will have hundreds of components. Also, the fact that the curves cross at 10% efficiency is not relevant. 10% is not a magical number – in fact as I was fine-tuning the assumptions, I tried to keep the 10% efficiency & $2.05 point at the same IRR. But the real takeaway is that these curves will cross somewhere, wherever that point is.