Why Sankey diagrams are so great for understanding energy data
Maybe you have heard about this old diagram invented by an Irish engineer at a time where steam engines were all the rage, but you are not quite sure why you should use them today? Here are five reasons why:
They make energy flows easily understandable.
A Sankey diagram makes it clear how subsystems are connected with each other and from where to where energy flows, which takes you a long way towards understanding a system. Take a random system with combined heat and power (CHP) plant and heat pump for district heating: what is happening there?
The Sankey diagram will not explain the magic of heat pumps or cogeneration, but it will give you a good high-level understanding of the system, and more succinctly than any textual description, including the following: the combined heat and power plant transforms gas into electrical energy, whereby it also generates usable heat as well as non reusable heat. The electrical energy is used by the heat pump to bring ambient heat to a useful temperature. Usable heat from the two previous components is brought into the district heating (DH) system, which delivers it to consumers over substations, but also incurs some distribution losses…
They make proportions visible.
Energy per se being such an odorless, weightless and invisible quantity, it is often difficult to relate to amounts of energy. Do 1000 kWh represent an awful lot of energy? Are 40 BTUs negligible? Well it depends what these amounts of energy relate to and how they compare to other amounts of energy in the same system. Displaying amounts of energy as widths happens to make comparisons very easy for us humans, much easier than when the same quantities are numbers in a spreadsheet or colors in another diagram. Taking the same example as before, we see that the heat pump extract significantly more ambient energy than it requires electrical energy from the CHP plant (this is what makes it interesting). We also see that substation losses are rather negligible compared to distribution losses, which may help prioritizing improvements to the system (it appears more advisable to invest in reducing distribution losses or increasing CHP efficiency than in making the substation more efficient).
They allow balances and imbalances to checked and detected in the blink of an eye.
This makes an indispensable step effortless and intuitive. Are the widths on the left side and right side of an energy conversion node equal? If so, energy conservation is respected, as it should be. If not, it will immediately jump to your eye that something is off in the current data. Also, thanks to the proportion-visualizing property of Sankey diagrams discussed above, the magnitude of the difference could be identified as more or less significant, from slight deviations which may be caused by rounding errors to major deviations hinting at an unbalanced system description.
For instance, this is a Sankey diagram as one could obtain it if one were to forget CHP losses and distribution losses in our example system:
Not so nice, right? Well, it just reflects some imbalance in the data, so find the issue and make a new Sankey diagram, a more beautiful one!
They are beautiful
Ok, this one may be more subjective, but I need to acknowledge it for myself: not only can a good Sankey diagram be more effective at conveying the organization of a system and the proportions of energy flows than columns of numbers or pie charts, but it is also much more pleasing to the eye, isn’t it? Dancing arrows, symmetric branches, traversing flows. Some have the organic beauty of trees, others rather look like uncanny arachnids or colorful balls of yarn. Overall, they are a spectacular kind of visualization.
They allow for many variations
Building upon the basic idea of displaying energy flows as a network of arrows of proportional widths, innumerable variations can be used to create rich visualizations. Just to give you a few ideas:
- they can be colored to identify subsystem types and/or energy carriers
- they can be animated (e.g. to display the evolution of a system over time)
- they can be made interactive (dragging nodes, showing detailed information on hover etc.)
- they can be embedded in a map or another system visualization (think Charles Minard’s famous representation of Napoleon’s Russian Campaign).
