Reading the comfort compass

Reading the comfort compass#

plot_comfort_compass() sums up how a room’s days split up around a comfort zone: how many days were in range, and how many were too warm, too humid, too dry or too cold, or a mix of two. The counts come from comfort_compass_distribution(), which takes the daily means of temperature and relative humidity and compares every day with the comfort zone. The compass shows how often; the h,x diagram (plot_mollier_hx()) shows where and why.

from importlib import resources

import pandas as pd

from pyedautils.comfort import comfort_compass_distribution
from pyedautils.plots import plot_comfort_compass
from pyedautils.psychrometrics import pressure_from_altitude

df = pd.read_csv(resources.files("pyedautils") / "data" / "flat_temp_hum.csv",
                 sep=";", parse_dates=["timestamp"], index_col="timestamp")

dist = comfort_compass_distribution(
    df,
    temp_band=(20, 26),           # °C
    hum_band=(30, 65),            # % relative humidity
    hum_abs_band=(0, 0.012),      # kg/kg, optional absolute-humidity limit
    pressure=pressure_from_altitude(450),
)
fig = plot_comfort_compass(dist, title="Living room")
fig.show()

Why temperature and humidity belong together#

Comfort, health and the risk of mould always depend on temperature and humidity at the same time. Looking at only one of them, or at two separate graphs, can easily mislead you:

  • Relative humidity changes with temperature. Air at 21 °C and 60 % holds about 9.8 g of water per kg of dry air (at 450 m a.s.l.). Warm the same air to 26 °C and it reads only 44 %. Cool it at a 15 °C outside wall and it reaches about 87 %, enough for mould. The amount of water did not change.

  • The same reading means different things. 23 °C feels pleasant at 45 %, but sticky at 70 %.

That is why the compass judges every day on both together. A day only counts as in range when temperature and humidity are both in the comfort zone, and the direction of an arm tells you which combination was off. Warm + humid calls for cooling or ventilation, cold + humid points to condensation and mould risk, warm + dry to overheated, dry winter air.

Comfort#

The comfort zone is the range of room climate in which most people feel comfortable and the building stays healthy. For the compass it is given as a temperature band (temp_band), a relative-humidity band (hum_band) and, optionally, an absolute-humidity band (hum_abs_band). Typical limits for living rooms and offices are 20–26 °C and 30–65 %, with an upper limit of about 12 g/kg (hum_abs_band=(0, 0.012)). The exact values depend on the standard, the use of the room and the season; a cellar or a store room gets its own band.

With hum_abs_band, a day counts as humid when it is above the relative or the absolute limit, so a warm day above 12 g/kg counts as warm + humid even when its relative humidity is still in range (at 450 m a.s.l., 12 g/kg at 26 °C is only 54 %). Without hum_abs_band there is no absolute limit and humidity is judged by relative humidity only. The absolute humidity depends on the air pressure, so pass pressure= for your site, for example pressure_from_altitude(450) (96 035 Pa).

Warning bands grade how far a day was off. The stages come from ± offsets around the comfort zone: mild is up to the first limit, moderate up to the second, severe beyond, and the worse of temperature and humidity decides. The defaults of comfort_compass_distribution are 1 / 2.5 K (temp_stage_limits), 5 / 10 %-points (hum_stage_limits) and 1 / 2.5 g/kg (hum_abs_stage_limits, used only with hum_abs_band). Use the same offsets for the orange and red warning bands of plot_mollier_hx (comfort_zone_orange, comfort_zone_red), so the diagram and the compass tell the same story. More on the comfort zone in the h,x guide.

Reading the compass#

The dashed circle has the area of 100 % of the days. The green disc and the arms always add up to exactly that area:

  • All days in range: the green disc fills the dashed circle exactly, and there are no arms.

  • Some days off: the disc shrinks by the share of those days, and the same area reappears in the arms. A long arm is therefore not “more than 100 %”; it is area moved out of the disc.

  • Each arm points in the direction of the deviation, as in the h,x diagram: warm up, cold down, humid right, dry left, mixes diagonally. From the inside out its segments are mild, moderate and severe, each with an area equal to its share of days.

Hover over a segment to see its number of days and its share.

Common misreadings#

  • Compare areas, not arm lengths. An arm’s area is its share of days. A thin, long arm and a short, wide one can stand for the same number of days.

  • Compare compasses by area only with fixed_scale=True (the default), see Comparing rooms.

  • The compass counts days, not hours. By default every day is reduced to its daily mean. A day in range can still have had a humid hour after a shower, or a hot afternoon; use plot_mollier_hx() with hourly data to see those.

  • “Humid” is about the room air, mould is about surfaces. A room in range at 21 °C and 50 % can still have mould at an outside wall corner below 13.6 °C. The compass tells you how often; for where and why see Common misreadings and Dew point: where condensation and mould start.

  • The result depends on the zone. Each compass is only as meaningful as the bands it was computed with; state them next to the chart.

Comparing rooms#

By default (fixed_scale=True) the radial axis is fixed to the largest possible arm. The dashed 100 % circle then has the same size in every chart, and equal areas mean equal numbers of days, so compasses of different rooms or periods can be compared directly. Draw them at the same height so they come out at the same size. With mostly good days the glyph is small in this mode.

With fixed_scale=False every compass is zoomed to its own glyph, so it fills the figure. The dashed circle then has a different size in each chart, and two compasses cannot be compared by area, only by their percentages.

rooms = {"Living room": df_living, "Bedroom": df_bedroom}
for name, room in rooms.items():
    dist = comfort_compass_distribution(room, temp_band=(20, 26),
                                        hum_band=(30, 65))
    plot_comfort_compass(dist, title=name, height=460).show()  # fixed_scale=True

Saving energy#

Good for people and the environment: the h,x diagram explains why the compass looks the way it does, and what you can do about it, often while saving energy (all numbers at 450 m a.s.l., 96 035 Pa; see The h,x diagram in five minutes for how they were computed):

  • Too dry in winter (left arms) usually comes from ventilating with cold outside air: at 0 °C and 80 % it carries only about 3.2 g/kg of water, which is just 20 % relative humidity once heated to 21 °C. Ventilate briefly and intensively (a few minutes, windows wide open) instead of keeping windows tilted for hours, and do not over-humidify: evaporating water takes about 0.7 kWh per kg.

  • Warm + dry (upper left) points to overheating. Heating less moves days towards in range: the water content stays, the relative humidity rises. Air with 6.4 g/kg reads about 35 % at 23 °C, but 39 % at 21 °C. As a rule of thumb, every degree less saves around 6 % of heating energy, good for comfort, the bill and the climate.

  • Too humid or cold + humid (right and lower right arms) means a risk of condensation and mould at cold surfaces: at 21 °C and 60 % the dew point is 12.9 °C, and mould can grow on surfaces below about 16 °C, such as outside wall corners in winter. Short, wide-open airing helps, as long as the outside air is drier (holds fewer g/kg) than the room air. In a cool cellar in summer that is often only the case at night or early in the morning.

  • Too warm or warm + humid in summer (upper and upper right arms): cool with night air instead of an air conditioner, as long as the night air holds no more water than the room air, and shade the windows during the day.

The saving-energy section of the h,x guide has the worked examples; more background on Wikipedia.