Discussion
The findings of this study demonstrate that an integrated maintenance strategy, rooted in an understanding of system relationships, offers a superior framework for infrastructure management compared to traditional, isolated approaches. By prioritising operational dependencies over simple simultaneous scheduling, the station can maintain a robust state of serviceability.
Multiobjective Optimisation and Logical Scheduling
The multiobjective optimisation (MOO) provided a range of efficient alternatives. Given the complex interactions within the system, a single global optimum is not apparent. While the
stochastic genetic algorithm attempts to optimise all events simultaneously, producing lower emissions and costs. The best timeline option offers a more logical approach for a complex systems such as a Train station.
The environmental and financial trade-off for this approach is relatively minimal, with total costs rising only slightly from €3,004,540 to €3,191,061. For decision-makers, this marginal increase is justified by the preservation of full serviceability. The selected maintenance schedule ensures the station remains open and functional for passengers, while the selected intervention frequency do not incur in high costs relative to the alternatives produced by the Pareto Optimization.
| Performance Indicator | Pareto Optimised (Selected) | Scheduling Optimized Timeline |
| Downtime Duration | 494 | – |
| Energy | 14.483.846 | 15.115.395 |
| CO2 | 1.022.232 | 1.087.334 |
| NOx | 2.373 | 2.513 |
| Cost | 1.539.428 | 1.656.143 |
| Total Costs | 3.004.540 | 3.191.061 |
Reflections
Through executing maintenance strategies and conducting a multi-objective optimisation, the following reflections have been made:
- Low-frequency events and high-frequency events create an intricate synergy, where optimising for duration through bundling causes relatively elevated costs. The cause is the high repetition of high-frequency events.
- The decision maker can opt for a balanced solution that seeks a middle point between cost and downtime duration. Optimising completely for downtime duration can come at higher environmental and financial costs, and it is usually linked to more interventions bundled, which add strain to logistical efforts.
- In general, the frequency of energy and CO2-intensive maintenance should be tried to kept as high as possible. At the same time, bundling strategies will reduce disruption for users, as long as the bundled systems do not have a negative relationship.
Limitations and future research
Limitations
The scope of this study is constrained by several factors that impact the total environmental and financial projections. Spatially, the analysis excludes secondary infrastructure such as stairs and access platforms, meaning total impacts are likely higher than reported. Environmentally, the cradle-to-gate boundary ensures data integrity by omitting speculative transport and end-of-life logistics, yet it ignores construction energy, operational consumption (e.g., lighting, HVAC), and recycling impacts.
Maintenance modelling was restricted to critical interventions affecting daytime availability; minor overnight repairs and routine inspections were excluded from both the timeline and the LCI. Furthermore, the analysis assumes linear degradation cycles and current economic/environmental conversion rates, which do not account for non-linear structural wear, 120 years of inflation, or shifting carbon tax regulations. While a longer analysis period might yield different specific solutions, the general trend of staggering high-impact, high-frequency events would likely persist.
Future Research
To build upon the framework established in this report, the following areas have been identified for further investigation:
- Expanded Boundaries: Transitioning to a Cradle-to-Grave or Cradle-to-Cradle assessment to model deconstruction and circular economy benefits.
- Algorithmic Refinement: Developing MOO models that incorporate pre-defined, non-independent system interactions and constraints based on historical budget data, rather than treating all events as stochastic variables.
- Probabilistic Modelling: Replacing fixed intervals with stochastic degradation models to simulate real-world resilience against extreme weather or unexpected structural wear.
- Operational Integration: Incorporating the station’s operational energy demand and overnight maintenance tasks to provide an exhaustive material inventory.
- Geographical Sensitivity: Evaluating supply chain logistics within the LCA.