Predicting structural disasters with Radar interferometry

Dal 23-25 2015, a Ginevra in Svizzera si è svolto lo IABSE Conference: Structural Engineering: Providing Solutions to Global Challenges. RIprendiamo qui l'Abstract dell'articolo rimandando con un link al sito del congresso per leggere l'articolo in modo completo.

Predicting structural disasters with Radar interferometry

Karel Terwel, Ramon Hanssen
Delft University of Technology, Delft, the Netherlands
Henk van Waning
Koninklijke Volker Wessels Stevin N.V., the Netherlands

Radar interferometry is a technique which can observe the earth’s surface during day and night time. It makes uses of thousands of pulses per second that are transmitted by satellites and reflected by the surface of, for instance, structures. By analyzing the data it is possible to measure displacements of the surface within millimeter precision. For the building industry this might be a promising technique, for monitoring buildings or for forensic engineering. Variables that might be monitored are the displacements of roofs, of balconies or the settlements of buildings. In addition, the technique might be a tool which can be used for forensic investigations. This paper will discuss the possibilities and limitations of radar interferometry for both building monitoring and forensic engineering. The method is expected to be especially useful for measuring soil displacements and the resulting settlements of structures.

Keywords: radar interferometry, structural health monitoring, forensic engineering, new technologies

1 Radar interferometry and structural failures
Satellite radar interferometry (or InSAR: Interferometric synthetic aperture radar) is a technique to observe the geometry, and geometry changes, of the earth’s surface from an orbiting satellite [1]. The satellite orbits are designed such that they repeat exactly after a given repeat interval, typically in the order of one or two weeks. The radar samples the earth’s surface and all structures on it with a spatial resolution in the order of meters. In this way, time series are constructed with a typical length of a decade per satellite mission [2]. Using radar phase measurements, the precision of measuring geometry changes is in the order of millimeters, and the estimation of strain rates can achieve precisions of better than 1 mm per year [3]. This
routine form of sensing holds great promise to assess and monitor the health of structures, anywhere on earth, and independent of weather conditions or solar illumination.
Although a limited number of the resolution cells in the images is useable, given that a typical radar image has ten thousand to a million resolution cells per square kilometer, this ensures that there remain many observation points, especially over structures. Furthermore, since radar images have been archived from many satellite missions since the early 1990’s, there is a huge potential for retrospective analysis.
The investigation and determination of the causes of structural failures of buildings, bridges, and other constructed facilities is called forensic structural engineering [4]. A structural failure can be defined as the inability of a structure or a structural member to fulfill the specified requirements [5]. A structural failure can manifest itself in a (partial) collapse, structural damage, material deterioration, insufficient functionality or no damage (when structures cannot fulfill the requirements, but no damage has occurred yet).
The characteristics of radar interferometry create the possibility to use it for ‘forensic structural engineering’, analyzing in hindsight whether a known structural failure has had precursory deformation. In addition, radar interferometry can be used for ‘structural health monitoring’, predicting and avoiding or mitigating incumbent structural disasters by detecting anomalous motion.
The possibility to monitor the deformation of a building night and day for a period of time without being on site, on difficult-to-access areas, is a unique feature of InSAR as structural health monitoring tool. Another advantage of InSAR is its ability to completely monitor large structures with a single measurement. For onsite measurement techniques, this is often impossible or else very time-consuming and expensive.
However, there are some requirements and limitations on the use of InSAR for forensic structural engineering or structural health monitoring.
First, the deformations should be visible at the outer shell of a structure. Failures within a building, without effects on the outer shells cannot be analyzed with InSAR.
In addition, a strict requirement to achieve reliable (‘coherent’) time series is that the reflective characteristics of the structure do not change significantly over time. This condition of coherent, or persistent, scattering is in practice very hard to achieve. In fact, perhaps more than 90% of all resolution cells does not satisfy this condition.
Furthermore, it is important to stress that the technique is opportunistic, in the sense that one cannot control the location of the radar reflections, as this is the result of an intricate combination of the object geometry, shape, and orientation, and the satellite’s orbit, viewing geometry and wavelength.
Another important aspect is the sensitivity to the radar viewing direction. Whereas the deformation vector of a (part of a) structure is effectively three-dimensional, the radar can only measure geometric changes in the line-of-sight from the object to the satellite. By combining two viewing geometries, the rank defect can be reduced from two to one, but this still implies that not all deformation components are equally well observable. Particularly the north-south component is difficult to observe.
For these reasons, the technique should be seen as a valuable complementary component to other analyses and structural health assessment techniques, particularly to get a quick and wide- range overview over a certain area. It provides a sample of structures, rather than a dedicated all- inclusive coverage.
The question now is to know what are the possibilities of radar interferometry for forensic engineering and structural health monitoring, given the opportunities and limitations of this method?
Therefore, in this paper an analysis is made of a database with 401 structural failure cases [6] to determine InSAR’s potential as a forensic engineering tool. In addition, the use of InSAR for forensic engineering and structural health monitoring is illustrated with a case study about large soil deformations of a shopping mall ‘t Loon. Finally, a brief comparison with other measuring techniques is provided.

2 Prediction of failures with InSAR
To investigate the potential of INSAR predicting failure cases, a database with structural failures has been examined.
This database consists of 401 Dutch failure cases from 1993 until 2009 which were published in a Dutch newspaper for the building industry, called Cobouw [6]. In this database general information of the failure cases is provided, such as location, number of stories and building type. Furthermore, information about the damage is provided, like type of damage, involved materials, involved structural parts and technical cause. In addition, information on possible physical warning signs is included. Finally, information on underlying causes is presented, like the building phase in which the failure was caused and the possible human error that was made.
For the selection of cases the focus has been on buildings. The criteria that were used to determine if a case potentially could be determined by InSAR data, are

  • damage should be visible on the outer shell, and
  • damage should occur gradually, over a longer period of time, because the satellite only provides data every 11 days (with a sudden collapse it will be hard to compare reflection points).

These criteria result in the following subset of cases.

  1. Only buildings (238 out of 401 cases, 1 redundant case has been removed).
  2. Damage should be visible (so a few cases with for instance erroneous calculations possibly leading to failures of the construction are out of scope)
  3. Buildings during use, during construction they usually do change too rapidly (189 out of 238 cases left
  4. No temporary structures (these are usually built and used during a too limited time, 181 cases left)
  5. No accidental loads (these usually result in a very brief development time of the damage (169 cases left)
  6. The damage should be visible at the outer shell. This can be directly visible (for instance with facades), or indirectly visible when the underlying structure was displaced (142 cases left)
  7. Deformation in the outer shell should be visible (71 cases left)
  8. Deformation in the outer shell must take more than 11 days to develop (50 cases left)

From the 50 cases that may be detectable InSAR there are 35 roof cases, 7 fac?ade cases and 8 balcony cases.
From these 50 cases, for 18 cases it is reasonable that the damage may be detectable with InSAR, when adequate data points would be available. For 32 cases this may be possible, but the case descriptions did not provide sufficient information to give a final answer.
The analysis of failure cases from the database gives insight in the potential of InSAR data to be helpful for analyzing, or even predicting, forensic cases. Approximately 10-20% of the reported failure cases with damage (18-50 out of 238) might be analyzed with help of InSAR data, if adequate data with useful reflection points would be available. The most probable situations are damage to roofs, fac?ades or balconies.
Deformation of roofs can be visible with InSAR. Deformation due to snow and instantaneous wind load might be hard to detect, because of limited visibility of reflection points. Displacements of complete buildings (observed via deformation of the roof), for instance caused by settlements of the foundation, are easier to detect with InSAR data.
Although facade damage can be visible, many failures did occur suddenly, for instance in the case of fac?ade elements where the connection failed during strong wind loads.
Damage of balconies revealed by excessive deformations, may be visible. For instance, it was shown by InSAR data that deformations of balconies in summer and winter can be different due to temperature.
From this first inventory, it appeared that structural failures caused by soil settlements are especially promising to analyze with InSAR, although the Cobouw might pay more attention to more ‘spectacular’ and visible failure causes. To illustrate the potential use of InSAR for forensic structural engineering and predicting disasters, a case study has been performed on a shopping mall in Heerlen that suffered large damage due to soil settlements.