An MRT train does not steer left or right. The track geometry changes underneath it.
Turnouts, switches and crossovers create the physical junctions that let one wheelset path become another while preserving gauge, guidance, clearance, comfort and maintainability.
A road vehicle turns because the driver changes wheel steering angle.
A conventional steel-wheel MRT train does not do that.
The wheelset is guided by the rails.
To move the train from one track to another, the railway changes the rail path itself.
straight route or diverging route → selected by movable track geometry
The family includes:
- switches or points;
- turnouts;
- crossovers between parallel tracks;
- diamond crossings and more complex junction layouts.
LTA’s current track-alignment design criteria explicitly include a chapter on turnouts and crossovers. The same design document says track alignment should prioritise passenger riding comfort, minimise wear and tear, and reduce maintenance cost while satisfying geometry and clearance requirements.
This article owns the physical junction geometry and its capacity/maintenance consequences. It does not own movement permission, route locking or train protection; those belong to How MRT Signalling and Train Regulation Work Using Mathematics.
The RFE — Why Does a Railway Need Turnouts?
The weak answer is:
to let trains change tracks
True, but incomplete.
Turnouts allow the railway to:
- enter or leave depots;
- reverse or turn trains;
- cross between tracks;
- route around blocked sections where infrastructure allows;
- serve branches and terminals;
- create operational flexibility during maintenance and disruption.
The RFE of a turnout is to introduce controlled route choice into a railway that otherwise physically constrains every train to one line of steel.
Prompt 1 — What Is the Geometry of a Turnout?
A simple turnout has a through route and a diverging route.
The diverging route changes heading over finite distance.
For a circular approximation of radius R, curvature is:
κ = 1/R
Lateral acceleration scale is:
ay = v²/R
If speed doubles, lateral acceleration becomes four times as large for the same radius.
This is why diverging turnout speed cannot be chosen independently of geometry.
Prompt 2 — Why Does the Switch Point Matter?
At the switch, movable rails guide the wheel flange and tread towards one route.
Conceptually, route geometry must satisfy:
correct rail position + correct gauge + sufficient wheel flange clearance + sufficient structural clearance → guided wheelset path
Small geometry errors can create impact, wear or poor guidance.
Let intended lateral rail position be y*(x) and measured be y(x).
e(x)=y(x)−y*(x)
The acceptable envelope is project-specific and is not reproduced here.
Prompt 3 — What Is a Crossover?
A crossover connects two nearby tracks using two turnouts and an intermediate connecting track.
Track A
\ turnout
\ connector
\ turnout
Track B
If track-centre spacing is D and crossing angle relative to the main track is α, a simplified geometric connector length scale is:
L ≈ D/sinα
Real crossover design includes transition geometry, turnout lead length, clearances, structures and signalling overlap.
A small α gives a longer, gentler crossover.
A larger α makes the junction more compact but increases curvature and dynamic consequences.
Turnout geometry trades land and length against speed, comfort, wear and maintainability.
Prompt 4 — How Does a Turnout Become a Capacity Constraint?
A junction is a shared resource.
Two conflicting train movements cannot occupy incompatible paths simultaneously.
Represent each train movement through the junction as an interval:
occupation interval i = [tin,i, tout,i]
For conflicting movements i and j:
intervals must not overlap subject to required safe separation
If one diverging movement takes longer because of lower speed, the junction remains occupied longer.
Junction throughput therefore depends on both signalling rules and physical geometry.
The signalling pillar owns safe authority; this pillar owns why geometry changes traversal time.
Prompt 5 — Why Do Turnouts Wear Differently From Plain Track?
Plain rail provides a comparatively smooth continuous running surface.
Turnouts contain geometric transitions, movable components and specialised crossing regions.
Wheel load distribution and contact location change through the turnout.
A simplified wear model is:
Wear rate ∝ contact pressure × sliding/creep energy × traffic exposure
For cumulative passages N:
D(N)=Σ ΔDpassage
Heavy use, sharper divergence and poorer geometry can accelerate local degradation.
That is why LTA’s current track-design criteria explicitly connect alignment design with passenger comfort, wear and maintenance cost.
Prompt 6 — How Are Turnouts Inspected?
Inspection can include:
- track geometry;
- switch rail position;
- wear and surface condition;
- fasteners and structural components;
- point-machine state as a separate electromechanical system;
- vibration or impact signatures.
Condition vector at turnout j can be represented:
xj(t)=[geometry, wear, actuation behaviour, vibration, inspection findings]
Trend matters more than one measurement.
de/dN
shows how profile or positional error changes with train passages.
This links to the Track Inspection and Predictive Maintenance pillars.
Prompt 7 — How Does a Turnout Affect Resilience?
A crossover can create a route around a blocked track.
But this does not mean full normal capacity remains.
Suppose normal two-track corridor capacity is C0.
During single-track operation through crossovers, effective capacity becomes Cd.
recovery capacity ratio = Cd/C0
The crossover may preserve service while reducing frequency.
That is valuable resilience.
But if a turnout itself fails, the alternative route may disappear.
Flexible infrastructure creates both option value and additional assets to maintain.
Prompt 8 — How Does the Junction Prove It Is Still Correct?
The design predicts:
- route geometry;
- train dynamic response;
- wear rate;
- junction traversal time;
- maintenance interval.
Operation returns:
geometry residual wear residual vibration residual traversal-time residual
If a turnout wears faster than predicted, geometry, traffic exposure, contact or maintenance assumptions need review.
If trains repeatedly require lower speed for comfort than expected, the designed dynamic envelope should be reassessed.
A Fictional Crossover Example
Suppose two parallel tracks are 4.5 m apart and a simplified crossover uses effective angle α=1.5°.
L ≈ 4.5/sin1.5° ≈ 172 m
A larger 3° effective angle gives:
L ≈ 4.5/sin3° ≈ 86 m
The compact geometry roughly halves this simple length scale but implies tighter geometric transition.
Suppose diverging radius is 300 m.
At 10 m/s:
ay = 10²/300 = 0.333 m/s²
At 20 m/s:
ay = 20²/300 = 1.333 m/s²
Four times as large.
The numbers are fictional and not Singapore turnout limits.
Deletion Tests
- Remove turnout geometry: trains can select routes without a physical path.
- Remove speed: tight and gentle turnouts have identical dynamic consequence.
- Remove gauge and clearance: route selection ignores wheel and structure fit.
- Remove junction occupation: conflicting movements can pass simultaneously without capacity penalty.
- Remove wear: junction flexibility has no maintenance cost.
- Remove inspection: geometry remains assumed correct forever.
- Remove resilience: crossover value is judged only in normal operation.
Turnout Paradoxes
- A train changes direction without steering.
- A turnout creates network flexibility by adding a local mechanical complexity.
- A slower diverging route can increase whole-network resilience.
- A more compact crossover can consume less space while demanding tighter dynamic geometry.
- The same junction that rescues service during a disruption can itself become a single point of failure if poorly maintained.
The Junction Audit
- What routes must the turnout physically provide?
- What curve radius and transition geometry result?
- What diverging speed is compatible with ride and wheel–rail dynamics?
- What track centres and clearances constrain the layout?
- What conflicting train movements share the junction?
- How long does each movement occupy it?
- How much network capacity does the junction constrain?
- How does local contact and wear differ from plain track?
- How is switch geometry inspected?
- What degradation trend is observed?
- What degraded service becomes possible if another track is blocked?
- What World Return evidence proves the turnout remains inside its intended geometry?
World Return — The Junction Under Real Trains
design geometry → install turnout → run trains → measure geometry, vibration and wear → maintain → remeasure → update interval and model
In September 2024, the East-West Line incident damaged three point machines used to divert trains to different tracks. That event is a public reminder that junction infrastructure is physically small relative to the railway but can matter disproportionately to route flexibility and recovery.
An MRT turnout works when a few metres of carefully controlled rail geometry can safely change the future path of an entire train—and still do so after thousands of repeated passages.
Key Equations
κ=1/R Curvature ay=v²/R Lateral acceleration scale e(x)=ymeasured−ytarget Track-position error L≈D/sinα Simplified crossover connector length occupation i=[tin,tout] Junction resource interval D(N)=ΣΔDpassage Cumulative degradation recovery capacity ratio=Cd/C0 Degraded-service capacity
Reader-safety note: This article does not reproduce Singapore MRT turnout locations, switch-control logic, point-machine internals, route-locking details, exact operational speed limits, tolerances or security-sensitive junction configurations. Numerical examples are fictional teaching values.