4.3 Accuracy and long term stability

In order to investigate the accuracy of the measurements end stability of the mechanism, the instrument was programmed to take 128 readings at half-hourly intervals over a continuous 64 hour peried. Dow Corning DC200 1000 cst lubricant was used as the test fluid, and the calibration performed for 1000 cst at 24.1 degrees centigrade. The experiment was carried out at room temperature with no oil bath.

The instrument satisfactorily made all the measurements, a full listing of which appear in appendix D; there was no substantial drift in the instrument between start and finish. A routine statistical analysis of the data revealed a mean value of 948 cst and standard deviation of 37 cst corresponding to 4.0 % of the measured value.

The low mean value clearly demonstrates that the accuracy of all measarements cannot be better than that of the calibration: effectively the precision of the measurements is halved. In this case the calibration seems to have been high, resulting in slightly low figures for all subsequent readings.

It is thought that most of the 4 % random errors arise from the non-linearities of the spring already discussed. However viscosity is also highly affected by variations in temperature, and it could certainly be argued that blame for the observed errors could be apportioned to variations in the laboratory temperature during the day/night cycle. In order to calculate the temperature drift that would cause a +/- 4 % viscosity deviation, the coefficients obtained from the temperature experiments of section 4.2 can be used. The constants from the exponential fit of figure 13 imply that a 37 cst change on 1000 cst would require a temperature change of just 4%. Such an amount may seem excessive but certainly a substantial fraction of the deviations could be due to temperature variations of one or two degrees.

In order to determine the true effect of temperature variations on the figures. future experiments could involve accurate temperature control during the measurements, and/or logging of temperature as well as viscosity.
 
 

4.2 Dependance of viscosity on temperature

Contents

5: Discussion