The e!Sankey alternative that computes your missing flows

Who e!Sankey is for

Anyone who has to produce, on a Windows workstation, a carefully crafted flow diagram under tight graphical control: engineering consultancies, process engineering, corporate energy and environment departments. Its home turf is industrial processes and report or presentation deliverables. It is a mature product, with a long track record on that market.

What we learned from e!Sankey

Immediate manual drawing: you drop a process, pull an arrow, type a value. Graphical control is fine-grained — an unlimited number of hand-placed fold points, accentuated tips that keep very small flows visible, a color gradient along a flow, a bundled clipart library. Its keyword label templates ({EntryName} {Quantity} {UnitName}, with prefixes, suffixes and replacements) are one of the features its users appreciate most, and we have nothing as free-form yet. Its balance warning marker on a process lets you check a diagram at a glance, without running any calculation. And its live link to Excel cells updates the diagram when the workbook changes.

What we chose to do differently

e!Sankey is a drawing tool: a flow value is an input, never a result. Every arrow carries the number you typed in. Its calculation-oriented edition adds unit operations — merge, sum, difference, split, transform — that recombine values you already bring along. None of them solves an under-determined system: if you do not know the value of half your flows, no function in the software computes them for you.

That limit is not an oversight, it follows from the model: an e!Sankey diagram carries one value per flow entry, with no propagated uncertainty, no constraint between distant flows, no result interval. And it brings a second one along: since a diagram holds a single set of values, a second year or a second scenario means a second file — and collapsible node hierarchies, which let you present the same model at several levels of detail, do not exist.

What we add: a balance from about 10% of known data

Take a waste flow study over a territory. You have the tonnages entering the civic amenity site (weighbridge readings, reliable to within a few percent), two invoices from downstream operators, a sorting reject rate announced by a plant operator, and an expert range for home composting. You are missing the value of some forty of the forty-five flows in the diagram.

In a drawing tool, the study stops there. You have to estimate the forty missing values by hand, on a calculator, hoping that inputs and outputs add up at every node — then start over the day one more data point arrives.

A material flow analysis engine takes the problem from the other end. Every node is an equation: what comes in equals what goes out. Every known ratio, every range, every uncertainty becomes one more constraint. The five measured values and all of those constraints form a system that the solver resolves: it computes the unknowns, adjusts the measurements that contradict each other along the way — within the uncertainty you gave them — and returns an interval rather than a single value wherever the information remains insufficient.

That is what "producing a balance from about 10% of known data" means: not guessing, but letting the structure of the system carry the information the measurements do not provide. The counterpart is real, and worth stating: you have to describe that structure — the nodes, the flows, the constraints — and a system with too few constraints returns wide intervals. But a wide interval is honest information, where an invented value is not.

Bring your e!Sankey diagrams with you

The editor opens .sankey files directly: File → Open, pick your diagram, done: no conversion step, no server upload, the file is read entirely in your browser. Processes keep their positions and colors, arrows carrying several materials become parallel flows, and each e!Sankey entry becomes a flow tag you can filter on.

When to stay on e!Sankey

If your deliverable is a report or presentation figure drawn to the pixel, if you rely on the color gradient along a flow, on free-text label templates, on the free line as a graphic element, on searching for an element inside a very large diagram, or on the live link to an Excel workbook: those features exist there and not yet here, and the table below lists them honestly. If your workstation is a locked-down Windows PC and your data is already complete and consistent, switching will not buy you much.