Typically, the statistical unit of one-square foot is selected based on maximizing the accuracy of the statistical analyses and this size unit is preferred based on experience in applying extreme values statistical analysis.  Ideally, a sufficient number of locations are selected to be representative of the corrosion conditions affecting the structure throughout the study area.  The size and number of inspection locations will vary with the length of the structure, the nature of the corrosion conditions to which it is subjected and cost considerations.

The pipe must be excavated at one (1) or more locations, with ten linear feet of pipe available for inspection at each location.  The size of each statistical unit is further reduced by dividing the pipe within the excavation into small statistical units. Therefore, a ten-foot long excavation for a 36-inch pipe can lead to the division of the pipe within this excavation into one hundred (100) sections of one (1) square foot each.  Having established the size of the statistical unit, the depth of the deepest pit in a given statistical unit is measured and recorded. In practice it is necessary to measure a number of the apparently deepest pits to determine which is the deepest. 

When the values of the deepest pits for all units selected for a given structure or section of structure have been recorded, these values are then arranged and plotted on extreme value probability paper.  This paper is similar in nature to normal probability paper, with the probability scale distorted.  Extreme value probability paper and extreme value probability statistical methods must be used when the area of concern is not near the middle of the range (mean and mode), but at one end.  Since we are not concerned with pits to some median depth, but rather with the deepest pits on the structure, extreme value statistics are used.

By a similar analysis, a prediction can be made, with a relatively high correlation, regarding the total number of penetrations that will occur by some future date.  The number of pits to a given depth that exist on the entire structure is determined using the extreme probability graphs.  The annual corrosion rate is calculated by dividing the pit depth by the age of the pipe.  An assumption is made that the corrosion rate is linear (i.e. a pit of 0.1 inches in ten years will progress to 0.2 inches in 20 years).  This result is typically expressed in mils (0.001 inches) per year.

The depth of the existing pits is subtracted from the wall thickness to yield the remaining thickness.  The remaining thickness is then divided by the corrosion rate to yield the number of years to penetration. Since the rate of corrosion is not a linear function of time and the wall thickness is not absolutely uniform, adjustments must be made.  Specifically, the analysis would deal with a range, or band of time.  If the contents of the pipe are under pressure, penetration can occur once the remaining wall thickness becomes insufficient to handle the stress.

PAGE 1 2 3 4 5 6 7 8 9 next