How to read this site
Every page here shows plansOne solution of the model: what it builds, where, and how it runs, under one set of assumptions. for Africa's power system in 2040, made by a capacity expansion modelA model that chooses what power plants, storage and power lines to build, and how to run them, to meet demand at the lowest total cost.: a program that decides what power plants, storageBatteries and pumped hydro, which take in energy when it's plentiful and give it back later. and cross-border power lines (corridorsA cross-border power line between two neighbouring countries, which the model can build or expand.) to build, and how to run them, so that every country's demand is met at the lowest yearly cost. It covers 48 countries, each treated as one point, a nodeA point in the model's network: here each country is one node, with its demand, plants and storage in one place.. There are 209 plans, and each differs from the others in one assumption.
The assumption the papers care most about is the cost of capitalThe yearly return investors and lenders need for the money they put into a project, a rate like an interest rate., the yearly return investors and lenders need for the money that builds a plant. Most models give every country one uniform rateOne cost of capital applied to every country and technology alike.. The papers instead give each country and technology its own estimated rate, the country ratesThe estimated 2025 rates: a cost of capital for each country and technology, so a solar farm in one country can cost more to finance than in another., and compare the two plans, which share every other input. The plan at country rates is the treated armThe plan solved at country rates, each country and technology financed at its own estimated cost of capital., the plan at one rate the uniform armThe plan solved at one uniform rate, the comparator a planner who ignores finance differences would build., and how much more the treated arm costs is the premiumHow much more the treated arm's system cost is than its uniform arm's, as a percentage..
Paper A asks what country rates do to the cost of the system and to where it builds. Its 128 plans include the base pair, the frontierThe system cost of plans solved at a series of uniform rates, showing how cost rises with the level of the cost of capital. of uniform rates, the sensitivities that test whether the result holds, and the near-optimal envelopeThe range a quantity can take among plans that cost at most 1% more than the cheapest plan.. Paper B asks what de-riskingLowering a country's cost of capital, for example through guarantees or cheaper public finance, modelled by moving its rates to the benchmark., lowering countries' rates to the benchmark rateThe papers' reference cost of capital, 7%, the uniform rate most research models assume., would save, one country at a time and all together, in 81 plans.
Pick a plan with the menu on the left. Each menu changes one assumption, and the page loads the solved plan that matches. The papers changed one assumption at a time, so most pairs of changes were never solved, and the page says so rather than inventing a plan between two.
The map in the middle shows where the plan builds. The panel on the right shows what it costs and what it builds, and clicking a country opens its own figures. Every underlined term opens its definition when you hover over it or tab to it, and leads to the glossary. Every view is a link, so the address in your browser shares exactly what you see.
Nothing here is a forecast. Each plan is what the model finds cheapest under its assumptions, and every number comes from a solved runOne solve of the model, stored with its inputs, its results and the code it was solved with., named beside it.