You are on page 1of 2

Analysis techniques

The Analyser should consider the following techniques when analysing instrumentation data. (1) Compare current data with the most recent data set to detect anomalies, discernible shortterm behavioural changes, and instrument malfunctions. If the deviation exceeds established statistical limit and falls under the outlier group, consider unusual anthropogenic activity or environmental conditions such as temperature, rainfall, snow, sun and shade at the data collection period to explain the deviation. Verify other adjacent instruments of same type and related complementary instruments to see an effect or correlation. Checking with the data collector by personal contact will help identify the possibility of collection, recoding or conversion error as well as equipment malfunctioning. Probability plot or box plot of data could be a method of identifying outliers. (1) Compare the current data point with historical performance over a significant period of time to ascertain consistency of instrument performance for the monitored feature. This can also indicate compliance of the new information with an established trend. Trend could be seasonal, over all or for a particular time span. Data fluctuation could be a problem to choose trend period. If the data fluctuate with reservoir level or season wise, then splitting the data set as per inflection points would be good to establish the trends accordingly. (2) Compare current data points with the initial reading for that point to determine the magnitude of change over time. This can indicate instrument drift or fundamental characteristic behaviour of the structure. Keep an eye to fluctuations at other data points to see the behaviour of fluctuation. Usually the piezometers located upstream of grout curtain show fluctuation with reservoir level. Precipitation may have influence too, specially to abutment piezometers. Extensometers reading response may sometimes be accompanied by inclinometer reading response. Inclinometers movement magnitude is evaluated in conjunction with direction. Rate of movement is also important information as it changes over time. (3) Compare trends of behaviour over time with trends predicted during design, with values relating to calculated factors of safety, and/or with any other predicted behaviour. Note that the historic behaviour of a structure becomes the base for comparison of future behaviour and the performance predicted during design becomes less relevant. Sometimes data might exceed the original designed caution limit. Revised Surveillance plan would establish new set of limits to compare with. Search for cause and effect should be the main aim of the analysis. The analyser will correlate the instrumentation data to the behaviour or performance of the physical component of the infrastructure. Knowledge of construction and

design details, geology and other subsurface conditions as well as environmental factors would be an essential tool for the analyser. (4) Compare the results of one instrument system with those of complementary systems to confirm or deny an implied physical change (e.g., consolidation settlements with dissipation of pore water pressure, pore water pressures with functioning of drains). Correlation of piezometer reading with reservoir may be confirmed by computing statistical parameters, i.e., correlation coefficient rather than cursory scanning. Correlation between piezometer reading and seepage weir or relief well drain reading may also be valuable information. (5) Use statistical analyses to assess the performance of instruments. Statistics like mean, standard deviations and variance are good indication of instrument response. This is also helpful in determining calibration frequency. This type of analysis would be good to perform periodically, say, annually rather than doing at every cycle.

You might also like