Risk Analysis
Open in the Padhaiko app7.1 Origin of risk in projects
Every engineering economic evaluation relies on forecasts of future costs, revenues, interest rates, and service lives. Because the future rarely unfolds exactly as projected, engineering decisions are exposed to risk and uncertainty.
| Aspect | Risk | Uncertainty |
|---|---|---|
| Probability Knowledge | Possible outcomes and their probabilities are known or can be estimated reliably from historical data. | Possible outcomes may be identified, but objective probabilities are unknown. |
| Analytical Tools | Expected Monetary Value (), Variance / Standard Deviation, Decision Trees, Monte Carlo simulation. | Sensitivity analysis, breakeven analysis, scenario analysis, minimax/maximin rules. |
| Example | Annual flood exceedances estimated from years of hydrological gauge records. | Market adoption of a brand-new disruptive AI hardware architecture. |
Major Origins of Risk in Engineering Projects
- Technical & Design Risk: Unproven technology, equipment failure rates, geological surprises (tunneling/foundation conditions), and scope changes.
- Cost & Estimation Risk: Errors in Bills of Materials, subcontractor defaults, and commodity price spikes (steel, copper, silicon).
- Schedule & Delay Risk: Permitting delays, supply-chain bottlenecks, and weather disruptions that postpone revenue while interest accrues.
- Market & Demand Risk: Lower-than-expected sales volume, entry of lower-cost competitors, or customer price sensitivity.
- Macroeconomic & Financial Risk: Interest rate hikes, exchange rate fluctuations on imported machinery, and unexpected inflation.
- Regulatory & Political Risk: Changes in tax law, environmental emission limits, tariff structures, or labor regulations.
7.2 Risk analysis of projects (Sensitivity, breakeven and scenario analyses)
Before committing capital, engineers quantify how sensitive project profitability is to forecasting errors using three complementary methods.
1. Sensitivity Analysis
Sensitivity Analysis
Sensitivity analysis evaluates how a project's economic criterion (, , or ) changes when one input parameter at a time (e.g., initial investment , annual sales volume, unit price, O&M cost, MARR, or project life) is varied across a plausible percentage range (e.g., ) while holding all other inputs at their base-case estimates.
- Spider Plot / Tornado Diagram: Plotting against percentage parameter changes reveals which variables have the steepest slopes. Steep variables are critical risk drivers requiring tighter estimation, hedging, or pilot testing.
- Limitation: Standard one-at-a-time sensitivity analysis ignores correlations among variables (e.g., higher inflation often raises both revenue and operating cost simultaneously).
2. Breakeven Analysis
Breakeven Analysis
Breakeven analysis determines the critical threshold value of an input variable (typically annual production volume , selling price, or service life) at which a project's profit or equivalent worth equals zero ( or ), or where two competing alternatives yield identical costs ( or ).
- Accounting (Linear CVP) Breakeven: With fixed cost , unit selling price , and unit variable cost :
where is the unit contribution margin.
- Economic (Time-Value) Breakeven: Incorporates capital recovery of initial investment (and salvage ) at over life :
- Two-Alternative Crossover Breakeven: Equating total annual cost of Method () and Method ():
Below , choose the lower-fixed-cost option; above , choose the lower-variable-cost option.
3. Scenario Analysis
Unlike one-variable sensitivity testing, scenario analysis evaluates consistent combinations of multiple variables changing together under distinct future economic states—typically Pessimistic (Worst Case), Most Likely (Base Case), and Optimistic (Best Case).
When probabilities () are assigned to each scenario with outcome :
- Expected Net Present Value ():
- Variance and Standard Deviation (Dispersion / Risk Measure):
A smaller Coefficient of Variation () indicates lower risk per rupee of expected return.
7.3 Decision tree
When an engineering decision unfolds over multiple sequential stages—such as conducting a pilot test before full-scale deployment, or sizing a small facility now with an option to expand later if demand surges—decision trees model the sequence of choices and random chance events.
Decision Tree Conventions and EMV
A decision tree is a directed acyclic graph read chronologically from left to right containing three node types:
- Decision Nodes ( Squares): Points where the engineer/manager chooses among controllable alternatives (select the branch with the highest Expected Monetary Value).
- Chance / Event Nodes ( Circles): Points where nature reveals an uncertain outcome with probability (). Its value is the Expected Monetary Value () of outgoing branches:
- Terminal Leaves ( Triangles): Final net payoffs ( or net profit) at the end of each complete path.
Rollback (Backward Induction) Algorithm
To solve any decision tree:
- Construct the chronological tree from left (initial decision) to right (terminal payoffs), labeling all immediate branch costs/revenues and chance probabilities.
- Compute the net cumulative payoff () at every terminal leaf.
- Fold back from right to left:
- At each chance node (), calculate .
- At each decision node (), select the branch yielding the maximum (minimum expected cost) and prune inferior branches (marked with
//).
- The value obtained at the leftmost root decision node is the optimal strategy's overall .
Worked Examples
1
A solar microgrid upgrade has base-case estimates: Initial investment , uniform annual net cash inflow , project life , zero salvage value, and .
Perform a one-at-a-time sensitivity analysis by varying , , and independently by from their base values, and identify the most critical risk factor.
Computing the base-case : Using :
Testing variations on each parameter independently:
- Initial Investment (; ):
- Annual Net Cash Inflow (; ):
- Discount Rate ( where ; where ):
Comparing sensitivity swings:
| Parameter Varied | Value | Base Value | Value | Total Swing |
|---|---|---|---|---|
| Annual Inflow () | ||||
| Initial Investment () | ||||
| Discount Rate () |
Annual Net Cash Inflow () has the widest swing () and is the only parameter whose drop turns negative (). Management must prioritize verifying revenue/demand forecasts.
2
- Economic Breakeven: A molding machine costs , lasts years with zero salvage at , incurs fixed annual O&M of , and produces parts with variable cost sold at . Find the annual economic breakeven volume .
- Scenario Analysis: Market studies yield three scenarios for the line: Pessimistic (, ), Most Likely (, ), and Optimistic (, ). Compute the Expected and Standard Deviation ().
Computing the economic breakeven volume (): Convert initial investment to an equivalent annual cost using :
Dividing by the unit contribution margin :
Computing Expected and Standard Deviation across scenarios:
Compute variance and standard deviation:
3
An electronics manufacturer must decide between building a Large Plant ( outlay) or a Small Plant ( outlay) for a new IoT controller:
- Large Plant: High demand () yields PV inflow (net ); Low demand () yields PV inflow (net ).
- Small Plant:
- If High demand occurs (), management can either Expand (additional PV cost , yielding total PV inflow ) or Do Not Expand (PV inflow ).
- If Low demand occurs (), management does not expand and earns PV inflow (expanding under low demand would yield only inflow for an extra cost).
Draw the decision tree structure, fold back from right to left using Expected Monetary Value (), and determine the optimal initial decision.
Mapping the decision tree and terminal net payoffs:
+--- High Demand (0.6) -------------------> +700k
/
+-- Large (-800k) (O)
/ \
/ +--- Low Demand (0.4) --------------------> -300k
[Root]
\ +--- Expand (-300k) -------> +600k
\ +-- High (0.6) [D2]
\ / +--- No Expand (0) // -----> +350k
+-- Small (-450k) (O)
\ +--- Expand (-300k) // ----> -150k
+-- Low (0.4) [D3]
+--- No Expand (0) --------> +100kTerminal net payoffs for the Small Plant paths:
- High Expand: .
- High No Expand: .
- Low Expand: .
- Low No Expand: .
Folding back secondary decision nodes and :
- At (Small Plant + High Demand): (Choose Expand).
- At (Small Plant + Low Demand): (Choose Do Not Expand).
Computing at the root decision node:
Selecting the higher : Build the Small Plant initially (), and expand only if high demand materializes.
Exercises
added
Distinguish between risk and uncertainty in project analysis, and explain four primary origins of risk in engineering projects.
Distinguishing Risk vs. Uncertainty:
- Risk refers to situations where future states are unknown, yet historical or statistical data allow analysts to assign objective probabilities to each outcome.
- Uncertainty refers to situations where possible future outcomes lack reliable probability estimates.
Listing four primary origins of project risk:
- Technical/Design Risk: Component failures, unforeseen geological conditions, or immature technology.
- Cost & Schedule Risk: Underestimated Bills of Materials, labor shortages, and construction delays.
- Market/Demand Risk: Unanticipated shifts in customer demand, competitor entry, or price erosion.
- Macroeconomic/Regulatory Risk: Inflation surges, interest rate hikes, currency devaluation, and environmental or tax policy changes.
Risk involves known outcome probabilities whereas uncertainty lacks assignable probabilities. Four key origins of project risk are technical/design failures, cost/schedule overruns, market/demand volatility, and macroeconomic/regulatory shifts.
A fabrication workshop compares two production methods over a -year life at :
- Manual Method: No initial machine purchase, annual fixed overhead , variable cost .
- Semi-Automated Machine: Initial cost (zero salvage after years), annual fixed O&M , variable cost .
Determine the annual breakeven production volume above which the semi-automated machine is cheaper.
Computing equivalent annual fixed cost for the semi-automated machine: Using :
Equating Total Annual Costs ():
The breakeven volume is units/year. Above units/year, choose the semi-automated machine.
A grid battery project has three discrete outcomes under different demand scenarios:
- Pessimistic (probability ):
- Most Likely (probability ):
- Optimistic (probability ):
Calculate the Expected () and the Standard Deviation ().
Computing Expected ():
Computing Variance and Standard Deviation:
Expected ; Standard Deviation .
A warehouse automation project requires , runs for years with zero salvage at , and is projected to generate in annual net cash inflows. By what maximum percentage can annual cash inflows fall below before the project breaks even ()?
Finding the minimum breakeven annual cash inflow : Set using :
Computing maximum allowable percentage drop:
If annual cash inflows drop by more than , the project becomes economically unviable ().
Annual cash inflows can fall by at most (down to ) before turns negative.
An engineering startup can spend to develop a sensor prototype. Prototype testing succeeds with probability and fails with probability (zero salvage on failure). If testing succeeds, the startup can either Commercialize (costing for PV revenue) or Sell the Patent outright for . Using a decision tree, determine the optimal post-success action and the overall Expected Monetary Value () of developing the prototype.
Evaluating the post-success decision node:
- Commercialize: Net post-prototype payoff (total path net ).
- Sell Patent: Net post-prototype payoff (total path net ).
At the decision node, choose Commercialize if testing succeeds.
Calculating overall of developing the prototype:
Since , the firm should develop the prototype and commercialize upon success.
If testing succeeds, Commercialize (). The overall of developing the prototype is .