Freezers are an important part of the laboratory. They store important samples - biological, chemical, or otherwise - and ensure that they are held at cold temperatures. By doing so, the rate of degradation of invaluable materials slows and enables recovery and use at a later point in time. Therefore, the health of a freezer is critical to the safe storage of samples; a deviation of temperature, no matter how brief or how small, may ruin one’s life’s work.
At TetraScience, we help laboratories remotely monitor their freezers and incubators. Having protected the freezers in pharma, biotech, and academic labs, we have gleaned tremendous insights into the behavior of these freezers. Our technology can provide a window into the health of a freezer simply by examining temperature data over the course of a day. Figure 1 is a great illustration of a freezer’s performance.
In Figure 1A, you will see the canonical temperature versus time plot for a ‘healthy’ -20C freezer. A non-obvious observation is that freezers are never at one temperature - their internal temperatures actually fluctuate as the compressor turns on (cooling) and the freezer warms. Furthermore, larger temperature increases (deviations) occur when a door is opened, for instance. The cycling of the compressor, which is observed via temperature fluctuations, occurs approximately every hour. Figure 1A is the temperature plot for a properly-functioning freezer.
In comparison, Figure 1B, we see the plot for a -20C freezer that is about to fail. The compressor is turned on far less frequently and more sporadically compared to it’s normal counterpart. Further analysis reveals an interesting trend: over the course of several days, the struggling compressor has a natural drift in baseline temperature.
In Figure 2, the trend line plotted along with the temperature data illustrates an upward trajectory indicative of a warming freezer. Based on the slope of the trend line, the freezer is warming more than 1/2 a degree every five days. In comparison, the trend line plotted for the normally-functioning freezer shows an average that warms at 1/3 of a degree every five days. The difference in rates of warming is indicative of a steadily rising baseline that may portend compressor failure as the freezer is unable to reach the desired set point.
As one further digs in to the data, analytics and machine learning can be applied to predict failures in the future. At TetraScience, we have built the infrastructure to monitor these instruments and capture the data produced. In time, we hope to leverage modern technologies, like artificial intelligence, to help solve current problems by leveraging knowledge from the past.