Most demand side energy models follow a similar structure. They start with input energy use and a projection of population and economic growth. Then they project energy use by sector and by fuel type. This is done by using a series of equations that describe the relationship between energy use and the factors that drive it. These factors are usually activity, efficiency, and fuel mix.
To improve these models the main methods are to increase the detail of the model, which then means increasing the number of equations and the number of inputs/parameters. As an example, this model now relies on base year values for vehicle stocks, activity, and efficiency by vehicle type and drive type. Then the parameters (or assumptions) are used to project these values into the future. They are set by the user and interact with the equations and inputs to produce the final output. The most important parameter in the model is vehicle sales shares which determines the uptake of electric vehicles, which is the main driver of the transition to low carbon transport.
Within this model there are many other parameters and it can be quite hard to understand how they all interact. I have included guides on how to fill out most of the inputs, such as vehicle sales shares, fuel mixing and other inputs. However, these inputs provide just as much information about what is happening as the outputs do - you can leave them as they are and see what the model projects, or you can adjust them to see how the model responds.
If you are looking for a specific methology document, then that is here (Google Drive link). However, those kinds of documents can be quite hard to read (and write!). Instead, I will provide a brief overview of the main processes in the model below:
Basically the model starts with base year data, macro projections and the user inputs which are assumptions for each modelling year. Then, one year at a time, the model calculates the energy use and emissions for each sector and fuel type, while taking into account the assumptions which guide how the fuel mix, stocks and activity might change during the year.
The main equations to keep in mind are:
activity km = stocks × mileage × occupancy_or_load
new activity km = activity km × activity growth
energy use = ( travel_km ) / efficiency
stocks = stocks - turnover + new stocks
new stocks = ( vehicle sales share × new activity km ) / ( mileage × occupancy )
Given these equations you could probably build a simple model yourself. However, the model is more complex than this because it has to account for the different types of vehicles and drives, and the different types of fuels. It also has to account for dynamics such as turnover, efficiency degradation, and the uptake of new technologies. Basically, for every parameter in the equation, you have to decide if you want to model that parameter too, since it is expected to change over time.
You can read about some isses with historical data here. But basically the model has access to historical energy data but only uses the data from the base year, with all the other input data to do the projection. This other input data includes the values for the equations specified above, such as stocks, activity, efficiency, mileage and occupancy. These are all gathered from various sources and is usually the most time consuming part of building the model.
You can see the historical energy trends below, which are included in the results dashboards. Note that these are only used for qualitative assesments within the model, not as inputs for the projections: