New: the AI firewall watches the whole platform and explains each incident in plain words. See how it works

Multiple time series on one chart, or several charts?

Overlay series that share a unit and a question, use a second y-axis rarely, and stack small charts on one time axis for everything else.

Put multiple time series on one chart when they share a unit and a scale and the question is comparison, such as the discharge pressures of three identical pumps. Use a second y-axis only for two series whose timing matters together, like flow and pressure. For more series or mixed units, stack small charts that share one time axis.

Before choosing, I write down the question the chart answers. “Which pump runs hottest?” and “Did pressure drop when flow rose?” need different charts, even with the same data behind them.

SituationChart
2–4 series, same unit, compare levelsOne chart, overlaid
Two series, different units, the question is timingSecond y-axis, with care
Many series, or mixed unitsSmall multiples on one time axis
Many identical devices, find the odd oneOverlay in grey, highlight one

When should you use a second y-axis?

Rarely, and only for two series in different units when you need to see how they move together in time. The relative height of the two lines is set by the axis ranges you choose.

Change the right axis from 0–100 to 0–200 and its series drops to half its height. Where the two lines cross, and which one looks bigger, depends on that choice. So I give each axis its unit, label it in the colour of its series, and never read anything into a crossing point.

How do you compare sensors with different units?

Stack them as small charts on the same time axis, or convert each one to percent of its range. The first keeps real units; the second lets you overlay them.

Percent of range is (value − range low) / (range high − range low) × 100. A tank at 4 m on a 0–5 m gauge reads 80 %, and a room at 20 °C on a 0–50 °C scale reads 40 %. Now both fit one axis. The price is that nobody reads metres or degrees off that chart any more, so I keep the real value in the tooltip and pick ranges that mean something to the operator, such as the normal operating band.

Small multiples or an overlay chart: which one?

An overlay compares levels at the same moment; small multiples compare shapes and timing across many series. Past four or five lines, an overlay turns the legend into a lookup task.

Small multiples work when they line up. Use the same time window on every chart, stack them vertically so a moment sits on one vertical line, and give charts with the same unit the same y-scale. Different scales on neighbouring charts make a small wobble look as large as a real swing. Over long windows, also check that every chart reads the same summary (what an hourly average hides). When one chart stands out, check whether the sensor or the process changed (telling a faulty sensor from a real change).

Overlays have a colour problem too, because not everyone sees colours the same way. WCAG 1.4.1 says colour must not be the only visual means of conveying information, so I label lines at their ends or vary the dash style. I apply the same rule to process screens, and it is why I keep most of a board grey, as in why most of an operator screen should be grey.

Utility rooms are a good test case, with compressors, dryers and chillers that report pressures, temperatures and power side by side; see compressed air, chillers and plant utilities monitoring.

Frequently asked questions

Should a time series chart start its value axis at zero?

Not always. A bar’s length stands for its value, so bar charts should start at zero. A line chart of a process value can zoom in on its working band, because the reader judges position and slope, not length. Label the axis so nobody mistakes a narrow band for a big swing.

Can series with different sample rates share one chart?

Yes, on a shared time axis each point sits at its own timestamp. Check that both series are summarised the same way, though. A raw series next to one averaged into hourly buckets will show peaks on one line that the other has smoothed away.

How do I highlight one device among many on a chart?

Draw every device as a thin grey line and the one you care about in a single strong colour on top. The grey lines show the normal spread, and the highlighted line shows where this device sits inside it. Add a direct label to the highlighted line rather than relying on the legend.

Share

See your field data live this week

Start a trial in your region with your first devices, or book a demo and bring your own use case.