Statistical Analysis Of Time Of Water Outflow On The Soil Surface After The Failure Of A Buried Water Pipe
Free (open access)
39 - 49
M. Iwanek, P. Suchorab, A. Skrzypek, M. Budzioch
Descriptive and inferential statistics are two main methodologies used in statistical data analysis. Descriptive statistics summarize data sets obtained as a result of an experiment, whereas inferential statistics enables generalization of observed data properties on a larger population. The present article is the outcome of the statistical analysis of results of time measurements of water outflow on the soil surface after the failure of a buried water pipe during laboratory investigations. During each of the 28 experiment variants, the pipe under internal water pressure was intentionally damaged causing the leakage. The time between the moment of the breakage and the moment of the water outflow on the soil surface was measured. A pressure head in the pipe and a leak area were two parameters varying during investigations. The first part of the statistical analysis included calculations of a mean, a mode, quartiles, a range, a standard deviation, a dispersion, a skewness and an excess kurtosis for data obtained in laboratory investigations. The normality of the results distribution was verified with Kolmogorov–Smirnov tests – original and modified by Lilliefors as well as the Shapiro–Wilk test. Relationships between varying analysed time and parameters during investigations were estimated on the basis of the regression analysis. All needed parameters were calculated with the Statistica 12 (StatSoft, Inc.) and MS Excel software. Results of analyses enabled to determine average values of time for investigated conditions, necessary in further studies. Laboratory investigations indicated that higher pressure head in a pipe results in tendency of time of water outflow to be lower (as per expectations), but no regularity between time and the leak area was proved during the analysis.
statistical analysis, water pipe failure, water outflow time