How Many Temperature Sensors Do I Need?

What is the temperature in the room you’re in right now? Take a guess, you’ll be correct within a few degrees. Now, what is the temperature of the room? Don’t bother answering that, it’s a trick question because in fact there is no “temperature of the room”, the temperature of the room is a 3D matrix that likely varies by up to 3°C (5.4°F) from one point to another.

In spaces as different as commercial refrigerators and data centers, temperature differences can be even greater. Computer modeling demonstrates how, in a data center, server racks can be cool at the bottom and hot near the top. Commercial refrigerators can have very cold areas near the chilled air outlets.  Whether or not the temperature variations are meaningful depends on what they impact. Consider the last time you turned your refrigerator down a little and noticed the next morning the milk container in the direct blast from the cooled air outlet was partially frozen.

Our latest white paper is titled, “Why isn’t the sensor reading the same as my thermostat?” The paper shows a room cycling through a twenty-four hour cycle in a second floor, sunny bedroom temperature differences at the floor and 6-feet from the floor can be as much as 5°F, and are never equal.  MIT’s Building Technology Group is explores design, technology and implementation of environmentally responsive urban housing in China. Figure 1 shows temperature variations from room to room in a sustainably designed apartment. This one plane model shows a 1°C (1.8°F) temperature difference in rooms with heat sources.

Figure 1: Modeling temperature variations in an environmentally responsive urban home shows average of 24°C and high of 25°C. Source: MIT Chinahousing Research

To make informed decisions about how many sensors to deploy, consider whether or not the heating and cooling sources are in direct line with sensitive materials. Enough sensors will be needed to insure the warmest and coolest locations are within established parameters. Too many sensors can lead to “sensor data fatigue”, having too much data. If you’re unsure, experimenting with a few in different locations is a good start. A balance of protecting valuable materials, cost, and variability within the space being monitored will insure that when problems occur, they are noticed.