Reading climate data

A year of logged data arrives as tens of thousands of rows, and the temptation is to summarise it into an average and a compliance percentage. Both of those numbers can be excellent while the collection is being damaged, so it is worth knowing how to read the record properly.

First, look at the shape

Before any statistics, plot temperature and relative humidity on the same time axis and just look at it. Almost every important finding is visible by eye.

A slow annual arc with humidity high in summer and low in winter is a passive building following the outside. A daily sawtooth is a control system cycling, or a heating timer. A sharp step that never recovers is usually an instrument that was moved, a door that was propped open, or a plant change nobody mentioned. Weekly patterns follow occupancy. A spike at the same clock time every day is nearly always solar gain or a lighting circuit.

Then untangle humidity from temperature

This is the step that changes conclusions most often. Relative humidity is a ratio between the water in the air and the maximum the air could hold at that temperature. Change the temperature and the ratio moves even though nothing got wetter.

The consequence: a store that swings from 45 to 65 percent relative humidity may have a completely constant amount of water in it, with the swing driven entirely by a heating cycle. The fix for that is temperature management, and buying a dehumidifier would be money spent on the wrong problem.

Dew point, or mixing ratio, strips the temperature dependence out. Plot dew point alongside relative humidity, and the diagnosis becomes obvious:

• Relative humidity moves, dew point flat: temperature is driving it. Look at heating, solar gain, ventilation.
• Relative humidity and dew point both move together: water is entering or leaving. Look at leaks, ground moisture, ventilation with damper outside air, visitor numbers.
• Dew point indoors tracking dew point outdoors: the space is effectively ventilated by outside air, whether or not that was the intention.

Most logger software will calculate dew point. If yours does not, a psychrometric formula in a spreadsheet takes ten minutes to set up once.

Statistics that describe risk

Averages flatter. These do not.

Time outside band. The percentage of readings above and below your chosen limits, reported separately. Fifteen percent of the year above 65 percent and fifteen percent below 40 is a very different collection risk from thirty percent above 65.

Duration of excursions. An hour above 70 percent is a curiosity. Three continuous weeks above 70 percent is a mould risk. Report the longest continuous excursion, not just the total.

Rate of change. The largest change within a rolling 24-hour window is the number that best correlates with mechanical stress on hygroscopic material. A room with a maximum daily swing of 4 percent is behaving very differently from one with a maximum of 25 percent, even if their annual ranges are identical.

Seasonal split. Report the heating season and the rest of the year separately. They are different problems with different causes and usually different fixes.

Compare against outdoors

An indoor record without a matching outdoor record is very hard to interpret. Outdoor data tells you whether an indoor rise was the building failing or the weather arriving, and it is the only way to see how much the building is damping the outside at all.

The damping factor, meaning how much of the outdoor swing survives indoors, is one of the most useful things a year of data can tell you. A building that reduces a 40 point outdoor swing to a 10 point indoor one is doing real work passively, and that is worth knowing before anyone proposes replacing it with mechanical plant.

What to do with the conclusion

If the record shows a genuine, sustained problem, target humidity ranges is where the decision about what to aim for gets made, and humidity control covers the equipment. If it shows short sharp swings on a stable baseline, the answer is more often buffering and sealing than plant, and showcase microclimates is the relevant page.