Cooling and Ventilation Strategies for a sendom.pl House — What the Sensors Reveal

In the previous post we modelled how this Sendom house retains heat. Here we look at the opposite problem: how it gets rid of heat — and what you can do to help it.

Cooling a well-insulated house is harder than it sounds. The same thick walls and airtight envelope that keep warmth in during winter also trap heat during summer. When outdoor temperatures hit 30–35 °C, the house becomes a slow-cooking oven. The question is not whether to ventilate, but when, how much, and with what strategy.

We analysed six months of sensor data, ran a calibrated thermal model, and simulated heatwave scenarios up to 2050. Here is what actually works.

The Sensors: Three Temperatures, Three Stories

The main bedroom has three temperature sensors. They do not agree. Understanding why is the key to interpreting everything else.

The high-mounted sensors are in the warm air layer that accumulates near the ceiling. In a heated room, the vertical temperature gradient is typically 1–2 °C per metre. A sensor at 2.5 m measures the ceiling temperature, not the temperature where people actually live.

For ventilation analysis, we use the bedside sensor as ground truth. It is the only one that captured what actually happened during our ventilation experiments.

Experiment 1: Cross-Ventilation on a Winter Afternoon

On 27 February 2026, we opened both bedroom windows at 16:45. Outdoor temperature was ~8 °C. Wind was 5–10 km/h. The windows stayed open until ~18:35.

What the sensors recorded

The bedside sensor dropped 5 °C in 53 minutes. The Netatmo, mounted high and further from the window, barely noticed — it lost only 1.2 °C and only reached its minimum after the windows were already closed.

This is not a sensor failure. It is thermal mass lag and stratification in action. The high sensor is embedded in the room’s thermal mass (walls, furniture, warm air layer). The bedside sensor is in the air stream entering through the window. It measures what the occupants feel.

CO₂ flush

CO₂ dropped from 1,374 ppm to 428 ppm in 110 minutes — a rate of −8.6 ppm/min at peak. Cross-ventilation with two windows open is brutally effective at clearing stale air.

The humidity puzzle

Relative humidity spiked from 45% to 68% during ventilation. This looks like the room got more humid. It did not. Cold outdoor air (8 °C, 80% RH) has low absolute moisture content. When it enters the warm room, its relative humidity drops — but because the room itself is cooling, the relative humidity rises even as the absolute humidity falls.

Absolute humidity dropped from 9.2 g/m³ to 7.8 g/m³. The “humidity spike” is a temperature illusion.

Experiment 2: Mechanical Ventilation Speed Matters

The house has a Zehnder ComfoAir HRV unit. We compared two empty-room days at different speeds:

Speed 3 clears CO₂ three times faster. But here is the surprise: indoor temperature stayed stable at both speeds. The HRV recovers heat from exhaust air, so even at high speed the thermal penalty is small. This is exactly what the HRV is for — you can ventilate aggressively without freezing the house.

The limitation is noise and power consumption, not temperature. For overnight sleeping, Speed 1 is probably the right compromise.

Experiment 3: Morning Ventilation Strategy

On hot summer days, the classic Polish strategy is: open windows at night, close them in the morning, keep shutters down during the day. We wanted to know: what is the optimal closing time?

We ran a calibrated thermal model (validated against real data — see below) with actual weather from 25 May 2026. Morning low was 13.7 °C, afternoon high reached 25.9 °C.

Strategy comparison

The optimal strategy is: open at 06:00, close when outdoor temperature catches up to indoor — around 10:00–10:30 on that day. Earlier opening gives more cooling but requires getting up earlier. Later opening is safer but less effective.

Validation against reality

On 25 May 2026, the actual window was opened at 06:07 and closed at 08:27. The bedside sensor recorded:

• 06:00: 21.8 °C
• Minimum: 18.8 °C at ~07:00
• 08:30: 20.0 °C

The model predicted 19.0 °C minimum and 19.5 °C at 08:30 — within 0.5 °C of reality. The main discrepancy on earlier runs was using a weather forecast that underestimated morning temperatures by 3–4 °C. Once we fed the model actual outdoor sensor data instead of the forecast, accuracy improved dramatically.

Experiment 4: Heatwave Analysis — What Happens at 40 °C?

We extended the model to simulate extreme heatwaves: outdoor temperatures from 30 °C to 40 °C, with morning lows from 15 °C to 25 °C. We tested six strategies:

1. Baseline: windows closed, shutters up
2. Night ventilation: open 22:00–06:00, shutters up
3. Night ventilation + shutters down: open at night, block solar gain during day
4. Morning ventilation: open 06:00 until outdoor ≈ indoor, then close + shutters down
5. Zehnder 40%: mechanical ventilation + shutters down
6. Cross-ventilation AM: open 06:00–08:00, then close + shutters down

Results for an extreme day (morning 15 °C, noon 40 °C)

Shutters dominate. Without them, even night ventilation leaves the room at 35.5 °C. With shutters down, the same ventilation strategy hits 24.7 °C — a 17.5 °C improvement.

The ground floor shutters block 95% of solar gain. The bedroom shutters, in “blocking mode,” block 70%. This is the single most important factor for summer comfort.

Strategy ranking across all scenarios

“Night + shutters” won 100% of the 25 temperature combinations we tested. It is never worse than any other strategy. The ranking is:

1. Night + shutters (always best)
2. Morning vent
3. Cross-vent AM
4. Zehnder 40%
5. Night vent (no shutters)
6. Baseline (always worst)

Future projections

Using IPCC AR6 projections for Central Europe (SSP2-4.5 scenario), we estimated how heatwave sunset temperatures will evolve:

By the 2050s, even with optimal cooling, the bedroom will average 29 °C at sunset during heatwaves. This is uncomfortable. The house will need active cooling — a heat pump in cooling mode, or at minimum, a ceiling fan — for the 10–15 hottest days of the year.

Practical Recommendations

Based on the data, here is what actually works:

For normal summer days (max 25–30 °C)

• Open windows at 06:00, close when outdoor temp catches up (~10:00–10:30)
• Keep bedroom shutters in blocking mode during the day
• Run Zehnder at Speed 1 overnight for fresh air without noise

For heatwave days (max > 30 °C)

• Open windows at 22:00, close at 06:00
• Ground floor shutters down 100% all day
• Bedroom shutters in blocking mode all day
• Run Zehnder at Speed 3 continuously (smart bypass handles the rest — see below)
• Do not open windows during the day — even briefly. The solar gain penalty outweighs any ventilation benefit.

The Zehnder Smart Bypass

Our Zehnder ComfoAir has a smart bypass that automatically opens when outdoor temperature exceeds indoor temperature. This changes the heatwave strategy completely:

Why this matters: Without the bypass, running Speed 3 on a 35 °C day would add ~0.1–0.5 °C of unwanted heat. With the smart bypass, that penalty drops to effectively zero. You get the full CO₂-clearing benefit of Speed 3 (6 ppm/min vs 2 ppm/min) without warming the house.

We simulated this explicitly — same heatwave day, same starting conditions, only the Zehnder strategy changes:

The smart bypass makes Speed 3 a free lunch: 3× faster air clearing at no thermal cost.

What does NOT work

• Opening windows during the afternoon “to create a breeze” — outdoor air is hotter than indoor, and solar gain is extreme
• Relying on the Netatmo sensor to judge comfort — it reads 3–5 °C cooler than the bedside sensor during ventilation events
• Night ventilation without shutters — you get fresh air, but the room still hits 35 °C by sunset

The Model

All simulations use a physics-based thermal model calibrated against real sensor data:

• Thermal mass: 4.0 MJ/°C for the bedroom (calibrated from bedside sensor data)
• Ventilation conductance: 1,000 W/°C for wide-open cross-ventilation (both windows)
• Solar gain: 2.2 W per W/m² of irradiance, reduced by shutter state
• Inter-zone heat transfer: estimated from adjacent room temperatures

The model is intentionally simple — it runs in milliseconds and can be swept across hundreds of scenarios. Its predictions match reality within 0.5 °C when fed actual weather data.

What’s Next

Future posts in this series will cover:

• Installing HRV duct sensors to measure true heat recovery efficiency directly
• Model Predictive Control (MPC): using the thermal model to optimise heat pump scheduling for cost and comfort
• PV-aware heating: coupling the thermal model with solar production forecasts
• Shutter automation: quantifying how much external shutters actually save, and when to open/close them automatically

This analysis was conducted independently, without involvement or knowledge from Sendom. All data was collected and processed with the homeowner’s consent.