System Dynamics Modeling for Project Disruption Costs
On heavily disrupted projects the loss is dynamic — it compounds as effects feed back on their causes. System dynamics models stocks, flows and feedback loops to quantify what static methods miss — the most sophisticated, and most demanding, disruption tool. Built on AACE 135R-24.
What system dynamics modeling is
Most disruption methods are static: they compare a clean window to an impacted one and assume the loss is a fixed factor. But on heavily disrupted projects the loss is dynamic — it grows over time as effects feed back on their own causes. System dynamics is a simulation technique, born at MIT, built to capture exactly that.
Feedback loops — reinforcing and balancing
The whole method turns on feedback loops: chains of cause and effect that circle back on themselves. Two kinds matter.
| Reinforcing loop (vicious cycle) | Balancing loop (correction) | |
|---|---|---|
| What it does | Amplifies a change — more leads to more | Counteracts a change — pushes back toward a goal |
| On a disrupted job | Pressure → overtime → fatigue → errors → rework → more pressure | Hiring more crews to recover the schedule |
| The catch | Drives the runaway overrun you're claiming | Often has side effects (dilution, congestion) that feed the vicious loop |
The rework cycle
At the heart of most construction system-dynamics models is the rework cycle. Work flows from "to be done" into "done," but a fraction of "done" work is actually defective — and crucially, that defect is often undiscovered for a while. When it surfaces, it flows back to be redone, consuming hours twice and stealing capacity from new work.
When system dynamics is the right tool
- The loss exceeds what static methods explain — the overrun is far larger than the Measured Mile or comparison studies can account for, pointing to compounding effects.
- Severe, pervasive disruption — many overlapping changes, acceleration, and rework: the conditions where feedback loops dominate.
- Rich project data to calibrate — manpower curves, progress and rework records detailed enough to fit and validate the model's parameters.
- Expert resources are justified — the dispute is large enough to warrant the specialist modelling effort the method requires.
Strengths and cautions
Its great strength is also its great risk: a system-dynamics model can represent almost anything — which means it can be built to produce almost any answer. Credibility therefore rests entirely on calibration and transparency. The model's structure must reflect accepted construction behaviour, its parameters must be fitted to the project's own contemporaneous data, and a neutral expert must be able to follow and reproduce it. A model that only its author can run, tuned to hit a target number, will not survive cross-examination.
Ten things to remember
- System dynamics models disruption as a dynamic, compounding process, not a fixed loss factor.
- It uses stocks, flows, and feedback loops to capture circular causation.
- Reinforcing loops amplify; balancing loops correct — and corrections can backfire.
- The rework cycle — including undiscovered rework — is the classic structure.
- "Fixes that fail": adding labour to recover can worsen productivity.
- Use it when the loss exceeds what static methods explain and disruption is severe.
- It is data-hungry and specialised — not a first resort for routine claims.
- Calibrate to this project's records — assumed parameters are a fatal weakness.
- Transparency is everything — a model only its author can run won't survive scrutiny.
- Reproducibility beats sophistication — a result an independent expert can rebuild wins.
Glossary
- Balancing loop
- A loop that counteracts change, pushing toward a goal.
- Calibration
- Fitting the model's parameters to the project's own contemporaneous data.
- Feedback loop
- A chain of cause and effect that circles back on itself.
- Flow
- A rate that moves quantity between stocks — e.g. progress, rework discovery.
- Reinforcing loop
- A loop that amplifies change — a vicious (or virtuous) cycle.
- Rework cycle
- The structure where defective work, often undiscovered, returns to be redone.
- Stock
- An accumulation in the model — e.g. work to do, work done, undiscovered rework.
- System dynamics
- A simulation method modelling stocks, flows, and feedback loops over time.