...
Data analysis was performed in a similar way as was performed in the Experiment 1. Figure 1. depicts the initial and final water level in the bubble collector during each of the data collection period ("runs"). Each run represents a specific time period during which the water level in bubble collector gradually sinks down from its maximum to the set minimum point. This period is represented on the graph when the line slants downward. Once the minimum water level is reached, the system has to refill with water in order to continue the runs. For this reason, the water outflow valve is closed until the water level reaches the set maximum point. This period is represented on the graph by the vertical lines.
Figure 1. also shows that the bubble collector didn't finish the whole cycle during the third period. The measured water level in the bubble collector governs when the bubble collector should start a new cycle by filling up with water. It is possible that the cause was occasional noise in data. Since the data averaging period is 10 seconds, there might have been spikes in data collected in this particular averaging period. The multiple spikes would could have caused the average to be lower than the minimum water level set point. Essentially, this would fulfill the condition for the process to change to a refill state. More detailed information on the bubble collector setup can be found here.
...