Railways: point heaters, traffic flow and traction power supply

Converters for compensation and rail power need tuned control

What happens. Rail power supplies and grid-interconnection converters feed single-phase traction power from a three-phase grid, and compensation units hold the voltage. The control has to deal with fast, uneven loads from trains and with strict grid rules at the connection.

What it costs. Commissioning takes long, and the control is tuned on site. A badly tuned unit disturbs the grid or the trains, and approval waits for a successful test.

Why the usual fix fails. Copying a controller from another substation ignores the different line, load and grid strength. Tuning on the real installation is slow and cannot reproduce rare faults.

How a fixed-price remote diagnosis checks it. From converter models, measurements from earlier installations and the grid’s requirements, I check how the control behaves over the operating range and where tuning effort could be moved to simulation. The fixed-price remote diagnosis tells you what a model-based design would bring.

Background: Predictive drive control

Timetable recovery after disruptions is done by hand

What happens. When a train is late, the dispatcher decides which train waits, which overtakes and which connection is held. Decisions are made under time pressure from experience, with a view of a few trains. A delay spreads to the next lines.

What it costs. Passengers miss connections, freight waits, track capacity goes unused and delays build up through the network. Dispatchers carry a heavy load and decisions differ from person to person.

Why the usual fix fails. Fixed rules, such as the first train goes first, ignore the knock-on effects. A bigger timetable buffer costs capacity in normal operation.

How a fixed-price remote diagnosis checks it. From timetable data, disruption logs and track layouts, I check how an optimisation could propose the order of trains in real time and how much delay it removes in past disruptions. The fixed-price remote diagnosis tells you what you can expect before a pilot.

Background: Optimal paths and setpoint plans

Point heaters run without knowing how much heat reaches the switch

What happens. Heaters keep rail points free of ice and snow. Most are switched on by a rail temperature sensor at one place, and then run at full power until a timer or the weather ends it. Nobody knows how much heat actually reaches the point.

What it costs. A heated network of this size uses a lot of energy, much of it in hours when snow or ice would not have formed. Where the heater is too weak, a point freezes and the traffic stops.

Why the usual fix fails. Raising the switch-on temperature is the safe way and uses more energy. One sensor stands for hundreds of points in different positions, exposure and wind.

How a fixed-price remote diagnosis checks it. From temperature records, weather forecasts, heater power and fault reports, I check how a heat balance of the point could decide when and how much to heat. The fixed-price remote diagnosis tells you what saving is plausible without more failures.

Background: Economic predictive control