Underground Excavations, Tunnels and Geologic Construction Risks

Underground works reveal a moving sample of the ground model while also changing stress and flow. Rock-support choices require excavation sequence, joint geometry, convergence and observational verification, not only material strength.

Outcomes and a practical engineering question

By the end of this unit, a student should be able to translate the geologic observations in excavations, tunnels and geologic construction risks into a documented alternative ground model, quantify at least one relevant sensitivity, and explain what additional evidence would change the decision. The goal is not memorization of terms but a defendable sequence: observation → interpretation → process model → engineering implication → uncertainty.

A usable field record contains the datum, location, measurement method, orientation or depth, date, observer, evidence strength and confidence. A drawing without those attributes is difficult to audit when field conditions change.

Four connected lessons

7.1 Excavation stability and stress redistribution

Core physical interpretation. An opening changes stresses around its boundary, creating concentrations and unloading. The Kirsch solution is an elastic idealization for a circular opening in a large continuous medium; jointed rock may behave very differently.

Civil-engineering case. A tunnel advance transitions from sound rock to a fractured shear zone, increasing deformation at similar cover depth.

Common error. Do not apply elastic closed-form solutions where failure, anisotropy or support interaction dominates.

Recommended observational method. Specify excavation method, initial stress, monitoring arrangement and permissible model assumptions.

From field observation to a working ground model. Begin by recording what could actually be seen or measured when investigating excavation stability and stress redistribution. In a coastal embankment, the team may encounter salinity-sensitive pore fluid while tidal groundwater variation changes boundary conditions. Divide the record into physical evidence, a preferred interpretation, and at least one plausible competing explanation. For example, a curved reflectivity anomaly, a change in drilling return or a wet joint might be consistent with multiple geological situations. A good model therefore predicts several independent observables rather than treating one indirect signal as a unique material label.

Scale and engineering decision. The scale of the question for excavation stability and stress redistribution must match the scale of the decision. A hand specimen can establish mineral type or intact strength but cannot prove the continuity of a weak layer over a bridge footprint. Conversely, a regional map can reveal structural trends but rarely resolves a narrow fracture zone at a tunnel face. Define whether the decision concerns a meter-scale foundation, a tens-of-meters roadcut, or a kilometre-long alignment, then select mapping, boring, laboratory testing and monitoring accordingly. State the range of plausible boundary positions and explain how each affects stability, excavation, drainage and the construction sequence.

Measurement protocol. For excavation stability and stress redistribution, first state the expected range and diagnostic features, then calibrate instruments and confirm the reference datum. Photograph the observation with a scale and north indication, record adverse as well as favorable evidence, and label samples so the record can be reconstructed. Preserve original orientation data before transformation. Where the first interpretation is uncertain, obtain an independent check rather than selecting a value just because it fits a familiar textbook category. Compare day-to-day conditions when moisture or stress changes could affect what is measured.

Quantitative reasoning. Choose a calculation proportional to the physical process instead of using a formula because it is available. Write the dimensions, units, reference directions and boundary assumptions before substituting numbers. Work with ranges for the parameters least constrained by field evidence. A sensitivity plot can reveal which parameters matter most, but its smoothness is not evidence that the geological variation is smooth. Every plotted result should be traced to the original observation, the simplifying constitutive law, and a check against an independent physical limit.

Interpretation and risk. In coastal embankment settings, a mistaken assessment of salinity-sensitive pore fluid may be more consequential under tidal groundwater variation. Define a failure of the working hypothesis: for example, repeated water-level readings contradict the assumed flow direction, measured joint planes diverge from the mapped structural domain, or a test pit intersects a contact outside its expected depth interval. Explain the difference between an uncertainty that merely widens a construction cost range and one that introduces a new safety-critical failure mechanism. Assign new evidence collection to the latter first.

Alternative hypothesis exercise. Hypothesis A assumes excavation stability and stress redistribution follows the dominant regional trend; hypothesis B assumes a local feature interrupts that trend. Draw a vertical section for both, label what is observed versus extrapolated, and compare the expected outcome of a borehole, scanline or geophysical traverse. Record what measurement would falsify either interpretation. The useful result is a testable prediction—not a forced choice between unverified assumptions.

Field-report exercise. Draft a one-page section called “Evidence and limits for Excavation stability and stress redistribution.” Include the measurement conditions, the most consequential model uncertainty, the observation that would reduce it, and the implication for a civil-engineering decision. Explain why the quantities should not be interpreted as final design parameters without project-specific evidence and qualified review.

Three contrasting investigation scenarios (synthetic input records).

• Scenario 7.1.1: A proposed coastal embankment exposes a geological contact at 18 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 24° relative to the local structural reference and a second observation is 4 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.1.2: A proposed coastal embankment exposes a geological contact at 17 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 43° relative to the local structural reference and a second observation is 9 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.1.3: A proposed coastal embankment exposes a geological contact at 26 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 56° relative to the local structural reference and a second observation is 12 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.

Local checkpoint. Explain the difference between a feature shown on a regional geological map and a site-specific characteristic verified at an excavation. State one parameter that can be calculated from the observations and one that cannot be estimated responsibly.

7.2 Rock support and observational methods

Core physical interpretation. Bolts, shotcrete, ribs and drainage respond to distinct failure modes and construction stages. Observational design depends on defined trigger levels and pre-agreed response actions.

Civil-engineering case. A rock cavern experiences progressive joint opening and local loosening before large convergence occurs.

Common error. Installing support does not eliminate the need to verify bolt anchorage and drainage behavior.

Recommended observational method. Plan inspections, instrumentation frequency and contingency measures alongside support concepts.

From field observation to a working ground model. Begin by recording what could actually be seen or measured when investigating rock support and observational methods. In a hillside road cut, the team may encounter daylighting structural surfaces while short-duration rainfall changes boundary conditions. Divide the record into physical evidence, a preferred interpretation, and at least one plausible competing explanation. For example, a curved reflectivity anomaly, a change in drilling return or a wet joint might be consistent with multiple geological situations. A good model therefore predicts several independent observables rather than treating one indirect signal as a unique material label.

Scale and engineering decision. The scale of the question for rock support and observational methods must match the scale of the decision. A hand specimen can establish mineral type or intact strength but cannot prove the continuity of a weak layer over a bridge footprint. Conversely, a regional map can reveal structural trends but rarely resolves a narrow fracture zone at a tunnel face. Define whether the decision concerns a meter-scale foundation, a tens-of-meters roadcut, or a kilometre-long alignment, then select mapping, boring, laboratory testing and monitoring accordingly. State the range of plausible boundary positions and explain how each affects stability, excavation, drainage and the construction sequence.

Measurement protocol. For rock support and observational methods, first state the expected range and diagnostic features, then calibrate instruments and confirm the reference datum. Photograph the observation with a scale and north indication, record adverse as well as favorable evidence, and label samples so the record can be reconstructed. Preserve original orientation data before transformation. Where the first interpretation is uncertain, obtain an independent check rather than selecting a value just because it fits a familiar textbook category. Compare day-to-day conditions when moisture or stress changes could affect what is measured.

Quantitative reasoning. Choose a calculation proportional to the physical process instead of using a formula because it is available. Write the dimensions, units, reference directions and boundary assumptions before substituting numbers. Work with ranges for the parameters least constrained by field evidence. A sensitivity plot can reveal which parameters matter most, but its smoothness is not evidence that the geological variation is smooth. Every plotted result should be traced to the original observation, the simplifying constitutive law, and a check against an independent physical limit.

Interpretation and risk. In hillside road cut settings, a mistaken assessment of daylighting structural surfaces may be more consequential under short-duration rainfall. Define a failure of the working hypothesis: for example, repeated water-level readings contradict the assumed flow direction, measured joint planes diverge from the mapped structural domain, or a test pit intersects a contact outside its expected depth interval. Explain the difference between an uncertainty that merely widens a construction cost range and one that introduces a new safety-critical failure mechanism. Assign new evidence collection to the latter first.

Alternative hypothesis exercise. Hypothesis A assumes rock support and observational methods follows the dominant regional trend; hypothesis B assumes a local feature interrupts that trend. Draw a vertical section for both, label what is observed versus extrapolated, and compare the expected outcome of a borehole, scanline or geophysical traverse. Record what measurement would falsify either interpretation. The useful result is a testable prediction—not a forced choice between unverified assumptions.

Field-report exercise. Draft a one-page section called “Evidence and limits for Rock support and observational methods.” Include the measurement conditions, the most consequential model uncertainty, the observation that would reduce it, and the implication for a civil-engineering decision. Explain why the quantities should not be interpreted as final design parameters without project-specific evidence and qualified review.

Three contrasting investigation scenarios (synthetic input records).

• Scenario 7.2.1: A proposed hillside road cut exposes a geological contact at 19 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 25° relative to the local structural reference and a second observation is 5 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.2.2: A proposed hillside road cut exposes a geological contact at 18 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 44° relative to the local structural reference and a second observation is 10 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.2.3: A proposed hillside road cut exposes a geological contact at 27 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 57° relative to the local structural reference and a second observation is 13 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.

Local checkpoint. Explain the difference between a feature shown on a regional geological map and a site-specific characteristic verified at an excavation. State one parameter that can be calculated from the observations and one that cannot be estimated responsibly.

7.3 Karst and underground void hazards

Core physical interpretation. Void roofs can span loads for a time and fail when drainage, stresses or weathering change. Conceptual beam analogies help identify which dimensions matter but cannot predict actual collapse of irregular cavities.

Civil-engineering case. A building site lies above buried limestone and geophysical reflections suggest a cavity of uncertain extent.

Common error. An apparent void from geophysics can be a lithologic artifact; intrusive checks remain critical.

Recommended observational method. Combine geomorphology, cavity geometry, hydrogeology and consequence screening.

From field observation to a working ground model. Begin by recording what could actually be seen or measured when investigating karst and underground void hazards. In a urban excavation, the team may encounter buried fill and utility corridors while pumping-induced drawdown changes boundary conditions. Divide the record into physical evidence, a preferred interpretation, and at least one plausible competing explanation. For example, a curved reflectivity anomaly, a change in drilling return or a wet joint might be consistent with multiple geological situations. A good model therefore predicts several independent observables rather than treating one indirect signal as a unique material label.

Scale and engineering decision. The scale of the question for karst and underground void hazards must match the scale of the decision. A hand specimen can establish mineral type or intact strength but cannot prove the continuity of a weak layer over a bridge footprint. Conversely, a regional map can reveal structural trends but rarely resolves a narrow fracture zone at a tunnel face. Define whether the decision concerns a meter-scale foundation, a tens-of-meters roadcut, or a kilometre-long alignment, then select mapping, boring, laboratory testing and monitoring accordingly. State the range of plausible boundary positions and explain how each affects stability, excavation, drainage and the construction sequence.

Measurement protocol. For karst and underground void hazards, first state the expected range and diagnostic features, then calibrate instruments and confirm the reference datum. Photograph the observation with a scale and north indication, record adverse as well as favorable evidence, and label samples so the record can be reconstructed. Preserve original orientation data before transformation. Where the first interpretation is uncertain, obtain an independent check rather than selecting a value just because it fits a familiar textbook category. Compare day-to-day conditions when moisture or stress changes could affect what is measured.

Quantitative reasoning. Choose a calculation proportional to the physical process instead of using a formula because it is available. Write the dimensions, units, reference directions and boundary assumptions before substituting numbers. Work with ranges for the parameters least constrained by field evidence. A sensitivity plot can reveal which parameters matter most, but its smoothness is not evidence that the geological variation is smooth. Every plotted result should be traced to the original observation, the simplifying constitutive law, and a check against an independent physical limit.

Interpretation and risk. In urban excavation settings, a mistaken assessment of buried fill and utility corridors may be more consequential under pumping-induced drawdown. Define a failure of the working hypothesis: for example, repeated water-level readings contradict the assumed flow direction, measured joint planes diverge from the mapped structural domain, or a test pit intersects a contact outside its expected depth interval. Explain the difference between an uncertainty that merely widens a construction cost range and one that introduces a new safety-critical failure mechanism. Assign new evidence collection to the latter first.

Alternative hypothesis exercise. Hypothesis A assumes karst and underground void hazards follows the dominant regional trend; hypothesis B assumes a local feature interrupts that trend. Draw a vertical section for both, label what is observed versus extrapolated, and compare the expected outcome of a borehole, scanline or geophysical traverse. Record what measurement would falsify either interpretation. The useful result is a testable prediction—not a forced choice between unverified assumptions.

Field-report exercise. Draft a one-page section called “Evidence and limits for Karst and underground void hazards.” Include the measurement conditions, the most consequential model uncertainty, the observation that would reduce it, and the implication for a civil-engineering decision. Explain why the quantities should not be interpreted as final design parameters without project-specific evidence and qualified review.

Three contrasting investigation scenarios (synthetic input records).

• Scenario 7.3.1: A proposed urban excavation exposes a geological contact at 11 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 26° relative to the local structural reference and a second observation is 6 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.3.2: A proposed urban excavation exposes a geological contact at 19 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 45° relative to the local structural reference and a second observation is 11 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.3.3: A proposed urban excavation exposes a geological contact at 28 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 58° relative to the local structural reference and a second observation is 14 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.

Local checkpoint. Explain the difference between a feature shown on a regional geological map and a site-specific characteristic verified at an excavation. State one parameter that can be calculated from the observations and one that cannot be estimated responsibly.

7.4 Construction geology and differing site conditions

Core physical interpretation. The encountered ground may depart from the baseline model. Clear reporting, timely logging and consistent observational records support decisions about excavation changes and contractual risk.

Civil-engineering case. A contractor encounters a previously unmapped faulted zone during tunnel advancement and must distinguish geological surprise from investigation limitations.

Common error. Do not silently revise an interpretation without preserving its evidence and date.

Recommended observational method. Maintain versioned ground models and close the loop between predictions and as-built records.

From field observation to a working ground model. Begin by recording what could actually be seen or measured when investigating construction geology and differing site conditions. In a bridge approach, the team may encounter buried channel lenses while seasonal river stage changes boundary conditions. Divide the record into physical evidence, a preferred interpretation, and at least one plausible competing explanation. For example, a curved reflectivity anomaly, a change in drilling return or a wet joint might be consistent with multiple geological situations. A good model therefore predicts several independent observables rather than treating one indirect signal as a unique material label.

Scale and engineering decision. The scale of the question for construction geology and differing site conditions must match the scale of the decision. A hand specimen can establish mineral type or intact strength but cannot prove the continuity of a weak layer over a bridge footprint. Conversely, a regional map can reveal structural trends but rarely resolves a narrow fracture zone at a tunnel face. Define whether the decision concerns a meter-scale foundation, a tens-of-meters roadcut, or a kilometre-long alignment, then select mapping, boring, laboratory testing and monitoring accordingly. State the range of plausible boundary positions and explain how each affects stability, excavation, drainage and the construction sequence.

Measurement protocol. For construction geology and differing site conditions, first state the expected range and diagnostic features, then calibrate instruments and confirm the reference datum. Photograph the observation with a scale and north indication, record adverse as well as favorable evidence, and label samples so the record can be reconstructed. Preserve original orientation data before transformation. Where the first interpretation is uncertain, obtain an independent check rather than selecting a value just because it fits a familiar textbook category. Compare day-to-day conditions when moisture or stress changes could affect what is measured.

Quantitative reasoning. Choose a calculation proportional to the physical process instead of using a formula because it is available. Write the dimensions, units, reference directions and boundary assumptions before substituting numbers. Work with ranges for the parameters least constrained by field evidence. A sensitivity plot can reveal which parameters matter most, but its smoothness is not evidence that the geological variation is smooth. Every plotted result should be traced to the original observation, the simplifying constitutive law, and a check against an independent physical limit.

Interpretation and risk. In bridge approach settings, a mistaken assessment of buried channel lenses may be more consequential under seasonal river stage. Define a failure of the working hypothesis: for example, repeated water-level readings contradict the assumed flow direction, measured joint planes diverge from the mapped structural domain, or a test pit intersects a contact outside its expected depth interval. Explain the difference between an uncertainty that merely widens a construction cost range and one that introduces a new safety-critical failure mechanism. Assign new evidence collection to the latter first.

Alternative hypothesis exercise. Hypothesis A assumes construction geology and differing site conditions follows the dominant regional trend; hypothesis B assumes a local feature interrupts that trend. Draw a vertical section for both, label what is observed versus extrapolated, and compare the expected outcome of a borehole, scanline or geophysical traverse. Record what measurement would falsify either interpretation. The useful result is a testable prediction—not a forced choice between unverified assumptions.

Field-report exercise. Draft a one-page section called “Evidence and limits for Construction geology and differing site conditions.” Include the measurement conditions, the most consequential model uncertainty, the observation that would reduce it, and the implication for a civil-engineering decision. Explain why the quantities should not be interpreted as final design parameters without project-specific evidence and qualified review.

Three contrasting investigation scenarios (synthetic input records).

• Scenario 7.4.1: A proposed bridge approach exposes a geological contact at 12 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 20° relative to the local structural reference and a second observation is 7 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.4.2: A proposed bridge approach exposes a geological contact at 16 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 46° relative to the local structural reference and a second observation is 12 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.
• Scenario 7.4.3: A proposed bridge approach exposes a geological contact at 22 m depth in one investigation point; an adjacent reconnaissance traverse is oriented 55° relative to the local structural reference and a second observation is 15 m higher in elevation. First decide which of these quantities is directly observed and which remains a hypothesis. Construct an A/B ground model and select a targeted follow-up measurement. Describe the effect of a wrong contact projection on construction planning and groundwater observation. Do not calculate apparent dip from these data without a measured true dip and a clear azimuth convention.

Local checkpoint. Explain the difference between a feature shown on a regional geological map and a site-specific characteristic verified at an excavation. State one parameter that can be calculated from the observations and one that cannot be estimated responsibly.

Integration of the two interactive laboratories

This unit uses Lab 13 and Lab 14. Both are stand-alone offline models. Start with the default case, predict the direction of change, perform an input sweep, and export the result. Critically compare the model to field observations: which assumptions are physically justified and which idealizations are merely convenient?

A laboratory is not complete until the student has submitted the predicted trend, the observed numerical response with units, a physical explanation, an estimate of sensitivity, and a paragraph on why the result cannot be applied directly to a real project.

Synthesis case: competing geological explanations

A team is examining a industrial building affected by weathered rockhead relief. The investigation was conducted while heavy equipment vibration was relatively quiet, creating a risk that the observed state is not representative of the critical construction condition. The team has a preliminary map, two discrete observations and a short monitoring record. Neither spatial continuity nor transient behavior is fully constrained.

Construct three competing cross-sections: one with laterally continuous units, one with an abrupt structural boundary, and one with a buried erosional or weathering feature. For each cross-section, list the predicted drill response, the predicted groundwater response, and the construction problem that would arise if it were true. Rank only the value of information, not a numerical probability unsupported by data. Identify the most efficient next test and the observations that would cause you to revise the construction method.

Finally write a limitations paragraph for a geotechnical or engineering-geology report. This paragraph must say what was measured, what was inferred, the dates and groundwater conditions, and which assumptions need checking during construction. Separate a local observation from a regional geological generalization and make all units consistent.

Laboratory and assessment links

• [Simulation 13: Circular tunnel elastic hoop stress](../labs/lab-13.html)
• [Simulation 14: Conceptual cavity-roof beam response](../labs/lab-14.html)
• [Unit 07 assessment](../assessments/unit-07.html)

End-of-unit review

1. Identify two measurements that distinguish the alternative ground models presented in this unit.
2. Demonstrate an order-of-magnitude consistency check on one simulation output and explain each assumption.
3. Explain how sampling orientation, geological discontinuity and time-varying groundwater could each bias a conclusion.
4. Prepare a short site-report recommendation with one action that reduces the greatest uncertainty.