Table 16.

Sources of experimental error and mitigation strategies.

Source	Description and mitigation steps
Change in ambient test conditions	Changes in temperature and humidity in the laboratory used for testing could impact results, including the density of the testing solution (and recorded mass). Temperature and relative humidity were measured every 5  min over a period of 24  h to quantify the fluctuations in ambient conditions over the length of a flow-rate test. The average temperature was measured as 20.4  ° C , with variation $<$2  ° C . The average relative humidity was 58.3%, with variation $<$13.5%. As the water used for testing was left to equilibrate to room temperature, volume was calculated using the density of water at 20.4  ° C . The variation on calculated volume due to temperature changes ($<$2  ° C ) is $\le$0.05%, and thus considered negligible.
External vibration	External vibrations could have significant impact on the performance of the balance, by introducing loads which interfere with the weight measurement mechanism. Therefore, the balance was placed on an active vibration isolating table (Vibraplane, Boston, MA, USA).
Eccentric loading of balance	If the load is not centred on the weighing pan, eccentricity error could impact the measurement. To reduce potential eccentricity error, the beaker was centred on the weighing pan. The level of the balance was also monitored throughout testing.
Evaporation	Evaporation of water from the beaker could impact the measured mass, which is of particular concern in flow-rate tests which run for 24  h . To reduce likelihood of evaporation, a 1.5  mm layer of paraffin oil was placed on top of the water in the weighing beaker. The use of paraffin oil is common in studies investigating the accuracy of commercial insulin pumps, and is proven to improve test accuracy [19], [20], [21], [22].
Air in testing solution	Air in the testing solution would impact the measured mass. A testing solution of ISO 3696:1987 class 3 water was used, as per the IEC standard. The water was de-gassed to reduce bubble formation, and limit the resulting variation in mass.
Siphoning effects	Siphoning effects could occur if the pump is not vertically aligned with the cannula tip. As shown in Fig. 11, an adjustable table was used to ensure the pump centre-line was level with the cannula tip.
Blockage in the infusion line	If a blockage developed in the infusion line, the test would be compromised. As per the IEC standard, a new infusion set was used for every flow-rate test. Frequent changes of the infusion line limits the chance of blockages developing.