 
          
            
              22
            
          
        
        
          
            Geotechnical News • June 2016
          
        
        
        
          
            GEOTECHNICAL INSTRUMENTATION NEWS
          
        
        
          the project designer, coming from a
        
        
          geotechnical or civil engineering back-
        
        
          ground. The quality of the monitoring
        
        
          service is significantly increased when
        
        
          what is being measured is understood.
        
        
          See Figure 2.
        
        
          
            QA/QC
          
        
        
          
            
              Rule number 7: Ensure quality
            
          
        
        
          
            
              control on the measurements, and
            
          
        
        
          
            
              actively maintain the monitoring
            
          
        
        
          
            
              systems.
            
          
        
        
          First, instruments must be installed
        
        
          properly. For many instruments, poor
        
        
          quality of installation will render
        
        
          future measurements impossible or
        
        
          will deliver very poor quality data.
        
        
          Ideally “final control sheets” are put in
        
        
          place, which list, for each instrument,
        
        
          the quality control to be carried out.
        
        
          When possible the control consists of
        
        
          applying a known variation to what is
        
        
          being measured, and checking on the
        
        
          final output (the report or the moni-
        
        
          toring database screen) whether the
        
        
          variation is correct. It is surprising
        
        
          how many mistakes can be detected
        
        
          using this method. Typically, these
        
        
          include factors of 10 or 1000, inverted
        
        
          axes, etc.
        
        
          After installation comes the monitor-
        
        
          ing. Many clients question the reason
        
        
          for having the expense of data manag-
        
        
          ers and data control on site. A common
        
        
          comment is “the instruments are auto-
        
        
          matic, so you do not need anybody on
        
        
          site”. But without
        
        
          continuous
        
        
          quality control,
        
        
          the systems,
        
        
          whether manual
        
        
          or automatic,
        
        
          will quickly drift.
        
        
          Such control can
        
        
          be automatic
        
        
          though data anal-
        
        
          ysis algorithms,
        
        
          but human brain
        
        
          power is also
        
        
          necessary. Data
        
        
          managers analyse
        
        
          the alarms and
        
        
          conduct corrective actions if neces-
        
        
          sary, they check the manually acquired
        
        
          data, and they are in charge of car-
        
        
          rying out detailed quality checks on
        
        
          selected instruments.
        
        
          Finally, depending on the accepted
        
        
          level of risk, sufficient spares parts
        
        
          and redundancy must be provided and
        
        
          included in the budget.
        
        
          
            Data to information
          
        
        
          
            
              Rule number 8: Include added value
            
          
        
        
          
            
              tools to maximize the use of the
            
          
        
        
          
            
              monitoring data.
            
          
        
        
          The primary deliverable of any moni-
        
        
          toring system are valid measurement
        
        
          data. This is a major achievement in
        
        
          itself. But the next question to address
        
        
          is: how can the usefulness of the
        
        
          monitoring data be maximised for the
        
        
          users, considering that data are useless
        
        
          if not understood?
        
        
          With this objective in mind, all the fol-
        
        
          lowing are important features:
        
        
          • Data integration (all data, from all
        
        
          sources, in a single system)
        
        
          • Data fusion (cross correlation of in-
        
        
          formation from different sources)
        
        
          • Alarm velocity and data velocity
        
        
          (rapid delivery of the alarm and
        
        
          rapid analysis of its causes)
        
        
          • Alarm management (acknowledge-
        
        
          ment, by whom, why, etc.),
        
        
          • Weather map approach or dash-
        
        
          board. This is the ability to display
        
        
          in a very simple and effective way
        
        
          a huge volume of more or less
        
        
          complex data, so one can under-
        
        
          stand what’s going on at a glance,
        
        
          in a similar way to a meteo map on
        
        
          your TV screen summarizing the
        
        
          calculations of some of the biggest
        
        
          computers on earth. See Figure 3
        
        
          for example, where three sites are
        
        
          summarized on one page, show-
        
        
          ing for each site the number and
        
        
          percentage of sensors not reading
        
        
          (for example disconnected if au-
        
        
          tomatic, or the planned frequency
        
        
          is not respected if manual), the
        
        
          number and percentage of alarms
        
        
          of type 1 and type 2 (count L1
        
        
          and count L2), and the number of
        
        
          alarms that have been acknowl-
        
        
          edged (i.e. controlled and com-
        
        
          mented by an operator). The colors
        
        
          of the squares and side bars help to
        
        
          understand at a glance the status at
        
        
          the monitoring site.
        
        
          • Journal (the monitoring system
        
        
          records a journal of internal or
        
        
          external events that is presented
        
        
          alongside the data to help the
        
        
          analysis)
        
        
          
            Conclusion
          
        
        
          The factors influencing the suc-
        
        
          cess, partial failure, or total failure
        
        
          of a monitoring project are numer-
        
        
          ous, starting from the design phase,
        
        
          through to procurement of the
        
        
          monitoring contractor, installation of
        
        
          the instruments, to data collection,
        
        
          pre-analysis, data presentation and
        
        
          reporting.
        
        
          
            
              Martin Beth,
            
          
        
        
          
            Soldata Group
          
        
        
          
            3120 Route d’Avignon,
          
        
        
          
            13 090 Aix en Provence
          
        
        
          
            France.
          
        
        
          
            Tel +(33) 4 42 21 72 11
          
        
        
          
            E-mail:martin.beth@soldatagroup.
          
        
        
          
            com
          
        
        
          
            Figure 3. Weather map approach: “board” showing a
          
        
        
          
            summary of three sites on one page.