What a Grid Is Made Of

Ore mined → material refined → components → the generation it becomes. Pick a scenario.

This is a teaching toy, not research. It runs in one direction with fixed coefficients and nothing in it is optimized: you hand it a generation mix and it tells you what that mix is made of. There is no cost, no siting, no lead time, no substitution and no recycling. Roughly a third of the coefficients are order-of-magnitude estimates rather than published figures, and each one says which it is. No number here should be cited.

What this is showing

There are two multipliers in this chain and they sit at opposite ends of it. Neither is an addition, which is why both are easy to miss.

Capacity factor, at the top of the page. Demand is energy, but metal is bought per unit of capacity. Solar runs about a quarter of the hours in a year and a gas plant better than half, so the same terawatt-hour asks for more than twice the nameplate from one as from the other — before a single kilogram of anything has been counted. Move the gas share and watch the capacity number move faster than the demand number does. Storage doesn't appear in that chart, and the caption below it says why: it has no capacity factor of its own, so it can't be a share of a total the other two rows are shares of.

Ore grade, where the diagram begins. Copper ore runs near half a percent, so a tonne of copper is a couple of hundred tonnes of rock. Iron ore runs above fifty percent, so a tonne of steel is under two. The first gap in the diagram — ore into material — is where a modest difference in metal becomes an enormous difference in mining, and reading the diagram left to right now means meeting that gap first, before anything else.

Between them they produce the one result on this page worth remembering, and it holds in every scenario here: copper is a few percent of the metal and the large majority of the rock. Steel dominates the tonnage — a wind tower is a very heavy object — while copper, which nobody weighs, dominates the mining. Those are two true sentences about the same build, and a model that stops at the metal column only tells you the first one.

Try it — each of these is a link that sets the controls for you. Start on Wind-heavy and note the metal, in the amber-gold "Wind" band at the diagram's right end. Switch to Solar + storage. The metal falls and the rock goes up, because wind's mass is steel and steel ore is rich. Then take gas to ninety percent and watch the copper refuse to leave the leftmost column — the grid connection is charged on every megawatt regardless of what built it, so there is no mix that routes around it. Finally set LFP to zero and watch a nickel band appear out of the storage column that was not there a moment ago. Copy link beside the controls will hand you the URL for whatever you have set.

What it leaves out

Everything that makes the real problem hard. There is no recycling here, no substitution when a metal gets expensive, no lead time between deciding to open a mine and the metal arriving twelve years later, and no cost at all — the mix is given rather than chosen. The companion toy on the recovery-loop model covers one of those omissions, the decade-scale lag on end-of-life recovery, and ignores everything this one shows.

Putting them together — a mix that is chosen rather than given, under capacity limits, lead times and recovery — is the actual research, and it does not fit on a web page. That work lives in the lab and in the papers.

Coefficients: src/scripts/materialData.js, one table, every entry labelled with its source. Model: src/scripts/materialModel.js, about two hundred lines and no dependencies. Capacity factors from EIA fleet averages, mineral intensities from the IEA's 2021 critical minerals annex, grades from USGS; bulk steel, aluminum and the grid allowance are engineering estimates made for this page.