Japanese tunnel collapse; Uncertainty in the forensic engineering investigation of foundation failures

(This is one in a series of articles on the investigative methods used in forensic engineering)

Japanese tunnel collapse

The recent highway tunnel failure in Japan (Ref. 1) reminded me of the difficulty in reliably determining the physical, chemical, and mechanical properties of foundation soils and rocks.

These properties are used in design and construction and must be determined for all earth and earth-supported structures resting on or in the ground.  Earth is made up of soil, rock, and water.

Design and construction of the tunnel relied on the properties of the rock the tunnel was in.  The tunnel would be a rock structure, a structure formed of or in rock.

Media reports are that the concrete lining of the tunnel collapsed after the anchor bolts corroded and gave way – more specifically, possibly the heads of the bolts corroded and rusted.  The concrete lining held in the place by the failed anchor bolts would then fall to the floor of tunnel and the vehicles there.

Questions?

  1. Was the corrosiveness of the groundwater and/or the rock reliably determined along the entire length of the proposed tunnel alignment preparatory to tunnel design?  This chemical property of water and rock would be important to anchor bolt design.
  2. Was the degree of fracturing of the rock mass reliably determined in the event it is found that the bolts were not embedded deeply enough and some pulled out?

Unlikely, is the answer to both questions considering the nature of a tunnel.

Difficulty carrying out reliable engineering investigations

A forensic engineering investigation of the cause of a failure or a complete collapse, where the initial hypotheris is that the cause lies in the ground, would check if the physical, chemical, and mechanical properties were reliably determined.  This checking during a forensic engineering investigation would experience similar difficulties to that during a field investigation for original design purposes.

Cause of difficulty

The difficulty in reliably determining the physical, chemical, and mechanical properties of foundation soils and rocks is due to the heterogeneous nature of the ground beneath the structure.

Soil and rock are construction materials like the timber, concrete, and steel, for example, that are elsewhere in a structure.  The difference is that where concrete and steel are very uniform – the same throughout, soil and rock are very non-uniform – not the same throughout; heterogeneous.

The design properties of steel, for example, are selected from a book taken off a shelf in the design office.  The properties are very reliable.  The properties of foundation soils and rocks must be determined for each construction site by means of field and laboratory testing.  The properties as determined can be quite reliable or quite unreliable, and everywhere in between.

I learned a long time ago when practising in England to expect the unexpected when dealing with the ground.

A review of foundation failures in England found that many were due to inadequate determination of the properties of the foundation soils and rocks.

What reliability depends on

The reliability of the properties determined for foundation soils and rocks depends in part on:

  • the nature of the surface of a site – the topography,
  • the degree of heterogeneity of the foundation soils and rocki,
  • the nature of the structure,
  • the thoroughness of the field and laboratory testing,
  • local practice, and,
  • the experience of the professional engineer planning the field work, and interpreting the data.

Examples of variable reliability in engineering investigation

In the case of the tunnel in Japan, the surface of the construction site would be a mountain.  How do you reliability and thoroughly determine the properties of the rock along the alignment of a tunnel beneath a mountain?  There are methods but the costs are very high.  There are less costly methods but the reliability is much lower.

The rock is investigated in advance of the working face of the tunnel during construction.  But this doesn’t give as reliable data on the properties of the rock above and near the crown of the tunnel.

A highway is a linear structure like a tunnel.  The surface of the highway site is relatively level.  Determining reliable physical, chemical, and mechanical properties is easier by comparison to a tunnel.

Mountains are sometimes formed by the upthrusting of rock formations from below.  The mountains on the west coast of North America were formed that way.  This uplifting distorts and fractures the rock – introduces greater heterogeneity into the rock formation.  There is likely to be greater variability in the physical, chemical, and mechanical properties along the tunnel alignment because it is beneath a mountain.

The ground surface at construction sites in Truro and also on the valley floor of the Annapolis Valley in Nova Scotia are level but the foundation soils must be expected to be quite variable.  This because of how they were formed in water and beneath glaciers.

Nature of field testing and the inherent uncertainty

Field testing at construction sites to determine physical, chemical, and mechanical properties is characterized by testing at discrete points.  The judgement call for the professional engineer is how close or far apart those points should be.  Then interpreting and extrapolating the data from the test points to all the soil and rock inbetween the discrete points – the great mass of soil and rock compared to the very small amount of soil that is field tested.  Then assigning physical, chemical, and mechanical properties to these construction materials.

Because of the uncertainty inherent in this process, engineering reports include a section on the limitations of the engineering investigation.  The section states that if changed conditions are encountered during construction – changed with respect to those reported, the engineer who did the field tests must be contacted.  He is given an opportunity to review his interpretation and extrapolation of the data at the discrete field test points, and the properties he assigned to the mass of soil and rock for design purposes.

Forensic engineering investigation of foundation failures is burdened by this same inherent uncertainty.  In spite of this, a forensic investigation would be more thorough and reliable if for no other reason but to avoid making the same mistake twice

References

1. Globe and Mail, December 4, 2012.

 

Leave a Reply

Your email address will not be published. Required fields are marked *