CIVL6001 Chap.1 Decisions When the Past Is Not Stationary
Decisions When the Past Is Not Stationary
Civil infrastructure is commonly designed from observations of past hazards, yet climate change means the future distribution may not match the historical one. This chapter preserves the value of records while narrowing what they can prove. It separates hazard, exposure, threshold and consequence, then shows how scenario conditioned evidence supports an adaptation decision without becoming a false prediction.
Historical observations remain indispensable for describing local variability, extremes, failure mechanisms and the quality of the measurement record. The problem begins when a statistic estimated from that record is treated as fixed throughout an asset life that extends into a changing climate.
A defensible analysis therefore states the observation period, spatial boundary and statistical assumptions before asking what additional evidence is needed for a future planning horizon. Hazard is the potentially damaging physical condition; exposure describes what lies in its path; vulnerability and thresholds determine how exposure becomes consequence.
Keeping these ideas separate prevents a projected change from being presented as a completed risk assessment. The engineer must also distinguish a scenario conditioned projection from a prediction of one inevitable outcome. Multiple plausible futures can inform robust margins, staged investment or monitoring triggers.
The preferred response depends on consequence, reversibility, lead time and the ability to detect change before performance becomes unacceptable. Quantitative comparisons are most useful when every demand and capacity value carries a unit, boundary and time period. A capacity margin ≤ 0 indicates that tested demand has reached or exceeded capacity.
The chapter's decision chain provides a reusable discipline: establish what the record supports, test asset performance across credible conditions, compare feasible responses and state the evidence that would reverse or advance the decision.
What this chapter covers
- 01
nonstationary design assumptions
- 02
historical baseline boundaries
- 03
hazard and exposure separation
- 04
asset performance thresholds
- 05
adaptation and monitoring triggers
- 06
conditional engineering judgement
Worked example · free
Drainage capacity under a changed flow
- 1State both values with units.
- 1Calculate the negative 0.6 capacity margin.
- 1Interpret scenario failure.
- 1Retain evidence conditions.
- 1Compare feasible responses.
Key terms
- Nonstationarity
- Change in the statistical behaviour of a process through time.
- Design threshold
- The value at which an asset no longer meets its required performance.
- Monitoring trigger
- A measurable condition that initiates review or a staged response.
Decisions When the Past Is Not Stationary FAQ
Why is historical data still useful?
It reveals baseline behaviour, natural variability, event mechanisms and measurement limitations. The key is to stop treating the historical distribution as an automatic future distribution and to add scenario conditioned evidence for the asset planning horizon.
When does a projection become a design decision?
A projection becomes decision relevant after it is translated into a hazard metric, compared with a specific asset threshold and linked to consequences. Option selection then depends on robustness, cost, flexibility and the seriousness of failure.
What makes staged adaptation credible?
A staged approach needs an observable trigger, reliable monitoring, enough lead time to act, authority and budget for the later step, and acceptable performance before the trigger is reached. Without these conditions, staging can merely postpone failure.
Exam move
Begin revision by drawing the evidence to hazard to threshold to response chain from memory. Under historical evidence, list what observations genuinely provide: baseline behaviour, natural variability, event mechanisms and measurement limitations. Beside that list, write the claim they cannot support alone: that the same distribution will govern the whole future asset life.
This contrast keeps nonstationarity from becoming a slogan. Next, choose an asset you understand and define one hazard variable, one exposure boundary, one service requirement and one failure threshold. Use units throughout. Write a present baseline statement, then rewrite it as a scenario conditioned future test. Notice how the second sentence needs a planning horizon, pathway and method before its number becomes meaningful.
Practise the drainage example with different capacities and flows. Always calculate threshold minus demand so the sign has a stable interpretation. A negative margin means the tested condition exceeds capacity; it does not automatically identify the preferred adaptation. Build an option table with immediate robustness, staged adjustment and monitoring only.
For every staged option, define the observable trigger, monitoring frequency, implementation lead time and consequence if action is late. Reject staging when those conditions cannot be met. Prepare for oral questioning by giving a ninety second answer in five moves: conclusion, relationship, evidence boundary, consequence and reversal condition.
Ask yourself what new observation or sensitivity result would change the recommendation. If the answer is nothing, decide whether the option is genuinely robust or the analysis has been insulated from evidence. Finally, take one historical design number from a report and annotate its data period, statistical method, spatial boundary and intended service life. Add the future evidence needed before claiming continued safety.
This exercise turns the chapter into a repeatable professional habit: preserve the record, narrow its claim, test the asset and make the action conditional on visible evidence.