| Vose Software

Industry: Engineering
Product: Tamara
Application: Structural design optimization


Optimisation Pays Only If It Targets the Right Activities. Three Drive This Bridge's Finish

An engineering firm faced a structural-design choice on an urban bridge: deliver the architecturally ambitious baseline structure, or an optimised route — a simpler structural form, phased construction and dual-source procurement — that cost slightly more upfront but promised a calmer risk profile. Optimisation only pays if it attacks the activities that genuinely drive the outcome: the ones that are both frequently on the critical path and high-impact when they slip. The first job, then, is to map exactly those.

Rebuilt in Tamara, Vose Software's Monte Carlo project risk tool, with Beta-PERT durations and costs on every activity and six discrete risk events layered on top, that map is a scatter of each activity's criticality index (how often it lands on the critical path) against its cruciality (how strongly its duration moves the finish). The top-right corner is where optimisation effort pays:

Criticality versus cruciality scatter of bridge activities — the de-risk map

Material procurement leads (cruciality 0.53, critical in 83% of runs) — the prime candidate for dual-sourcing — followed by superstructure assembly (0.44, 100% critical) and detailed structural design (0.44, 100% critical). These three are precisely the activities the optimised design targets through a simpler structure, phased assembly and dual-source procurement. Low-cruciality activities such as inspection and commissioning are not worth re-engineering, however visible they are on the bar chart.

The activities that drive the date

The same ranking, drawn as a tornado, confirms the de-risk shortlist by cruciality:

Schedule tornado ranking activities by correlation with the completion date

Material procurement dominates (cruciality 0.53), followed by superstructure assembly (0.44) and detailed structural design (0.44). Against the deterministic plan of 920 days, the simulated P50 finish is day 1,072 and the P80 sits 231 days (about 7.6 months) beyond the plan — a 3% chance of meeting the headline date. Optimisation attacks the activities the model identifies as the real drivers, not the ones that are merely visible.

Activity-level spread

The same simulation places every activity in time as a band, not a bar:

Stochastic Gantt showing each activity P50 bar with P10 to P90 finish spread

The whiskers widen downstream because uncertainty compounds along the chain: procurement, fabrication and assembly carry the widest whiskers — the same activities the scatter flags — and because they sit in series the spread compounds toward completion.

What the optimised design is worth

The optimised design trades a small upfront premium for a markedly tighter tail. Both routes are simulated on the same network, the optimised case re-running with reduced risk probabilities and impacts:

Project cost distribution for the baseline versus optimised design against the budget

Against a $179M budget, the baseline design carries a mean of $173M, a P90 of $188M and a 30% probability of overrun; the optimised design lands at a $167M mean, a $180M P90 and a 12% overrun probability. Because schedule slip feeds extended general-conditions and urban traffic-management overhead, the optimised design also pulls the P80 finish in by 72 days (1,151 → 1,079 days). The two effects are coupled, and the decision is not about avoiding a breach but about the lower mean and the tighter tail the optimised design buys.

The discrete risks that drive the tail

The six discrete events were modelled as Bernoulli risks. Ranking them by expected schedule impact (probability × delay) gives a clean Pareto:

Pareto of discrete risk events by expected schedule impact

Five of the six events carry ~80% of the expected discrete-event delay — a material-procurement delay (5.1 weeks expected), adverse ground conditions (4.2), a contractor-productivity shortfall (3.5), design revision/rework (3.1) and weather-related disruption (3.1). The optimised design's interventions — dual-sourcing, a simpler form and phased construction — map directly onto the top of this list.

What Tamara changed

  • The design decision was made on full cost-and-schedule distributions, not single-point estimates, exposing the optimised design's tighter tail (overrun 12% vs 30%).
  • Risk-reduction effort was aimed at the three top-cruciality activities — material procurement, superstructure assembly and detailed design — through the design choice itself.
  • The optimised design's upfront premium was justified on its tail-clipping value, not dismissed as a cost increase.
  • A 7.6-month baseline schedule contingency was quantified, and the optimised design shown to cut the P80 finish by 72 days.

Tamara Functionality Used

  • Monte Carlo simulation of cost and schedule over the full activity network, with Beta-PERT durations and costs.
  • Criticality-versus-cruciality mapping identifying the activities that are both frequently critical and high-impact.
  • Discrete risk-event modelling (Bernoulli occurrence × Triangular impact) on durations and costs.
  • Cumulative cost distribution and overrun-probability analysis comparing two design options on the same budget axis.
  • Stochastic Gantt and discrete-risk Pareto for communicating uncertainty and prioritising risk response.
  • Scenario comparison quantifying the optimised design's before/after impact on both the overrun probability and the P80 finish.

Structural design optimisation is a choice between distributions, not point estimates. Tamara is what lets the firm see the tail it is buying down — and prove the optimised design pays for itself.