How to Scientifically make the coffee machine explode

Wednesday, April 4, 2018

The coffee-machine. A very non-intelligent tool to magically transform plain water into a delicious, nice smelling extraction of beans and spices. As well as the beauty of the coffee itself, the beauty of the machine is yet to be found! With the introduction of IoT, not only the result counts, but the path to receive the result is evenly important, maybe even more important.

Below are 2 identical pictures. When looking purely at the result, these two pictures display a cup of coffee.


Surprising measurments

Measuring the power consumption of a cup of coffee is pretty straightforward using a multimeter. A multimeter offers a good starting tool to measure basic power consumption, but once we add some tracking, and analytics, the experiment becomes more interesting.

Firstly, we need some tools to measure the actual power consumption and store them into a database. In this project a Raspberry Pi was used. The raspberry pi with the addition of several electronical components serves as a gateway to a SQL database. Every second, a sample from the coffee measurement was send to the cloud in an endless loop, measuring sample after sample.

While one sample itself is already representative, a measurement the total coffee (like the picture above) tells a lot about the behavior of the machine before, during and after operation. First we see that there is a startup period of several minutes before even the coffee machine is ready to produce some coffee. This takes several minutes and consumes almost the same energy as making a coffee itself. One coffee is made in one minute, and even after the coffee is ready, some peaks in power consumption can be measured, even though the machine is into his idle state. Insight is key. What we expect isn’t always what we see. To produce one cup of coffee, we have to multiply the energy consumption of the machine during operation times three!

Endless possibilities

The measurement of the coffee machine is just an implementation and a Proof of Concept in the world of IoT. It goes beyond the measurement of power consumption. Every device connected to the power outlet will be monitored, not only the power consumption, but also to other factors such as temperature, humidity, operating time, cost predictions,  monitoring, ….  While nowadays electrical bills are calculated on the global power consumption of different devices, in the future every single device will be monitored and the share of the bill can be calculated individually.

The technology for implementing IoT already has been developed and redeveloped. The necessary components to start with IoT had already been foreseen in the early 2000’s. IPv6 was introduced in 1998, which was necessary because ‘only’ 3,706,452,992 addresses were defined by the precursor IPv4. To put this into perspective, this is the equivalent of only 7 addresses for each km² of the earth. IPv6 enlarged this amount to 2128, which is approximately 1 trillion addresses for each mm².

The options for machine learning and Big Data analysis are widely open. What is the influence of the temperature of the surroundings on the coffee machine? What happens to the power consumption of a fridge while the humidity raises? What can we do to minimize costs while keeping the operation optimized, and what will happen the next 2 years with my microwave if a certain trend will continue? Those are questions which will probably be answered in the next few years.


While I think that there is a future for IoT, first the mindset of the customers need to be changed in order to make full use of the potential of IoT. Once the GDPR regulations are set and people are getting used to share their data in a secure way, I’m sure a wide and broad band of IoT devices will pop up and a limitless amount of features will be introduced. Give it a few years, but the future is bright, thanks to Science!