A standardised 'black box' has been used for more than 40 years to record and preserve system data during the catastrophic conditions of an commercial airplane crash. The rail industry, however, has only recently approved standards for data recorders. These standards are still regionally limited. Samuel Kruger reports.

Experience has proved that flight (FDR) and train (TDR) data recorders provide an unparalleled method of learning from transportation system accidents. The collection of a maximum of data from the involved equipment before and during an accident helps in the reconstruction of the scene for investigation.
The rail industry only recently recognised that there was a need to define standards for data recorder attributes, and to unify the controversial requirements of the operators around the globe.
To this end, standards like IEEE1482.1(1999), GM/RT2472(2002), ERTMS/ETCSSUBSET-027(2001) were developed. These still diverse standards include the most practical and realistic solutions for train data recorders so far.

Functional characteristics

The train data recorder has to record and preserve data to be used in the investigation of an accident. At least one operating data recorder should always be available in the lead unit of any kind of powered rail vehicle. IEEE1482.1 allows the recording of events either as a standalone feature, or as part of a compliant train monitoring and diagnostic system.
The advantages of a compliant system are both cost reduction and an increased MTBF, through the multiple uses of all the data available in the system.
The goal is to monitor as many signals as possible. However, each of the before-mentioned standards independently defines which signals are to be captured. If only a joint list of the required signals were agreed upon, specifying which of those signals require physical inputs, the costs of hardware implementations would reduce greatly.

DATA LOGGING

In a compliant train monitoring system, signals not requiring physical inputs can be captured and transmitted by state-of-the-art networks (PROFIBUS, MVB, WTB, CAN, LON, IBIS, FIP, etc). The data logging application may be customised with an IEC61131-3 compliant PLC programming tool.
A joint standard should include a definition on the sampling and storage rate of each input signal. It is already mandatory that all records are date- and time- stamped (eg according to ISO8601). Whether event-driven or at a determined maximum interval, the data also needs to be sampled at a rate, which assures that any significant change is captured.

NON-VOLATILE MEMORY

The minimum storage time before data is overwritten in the non-volatile memory should be specified according to the demands of accident reconstruction. A parallel storage on a PC Card makes data collection easy for maintenance personnel.
Other download technologies (for example remote wireless data transfer, USB Memory sticks) are evolving rapidly.

A unified standard should also include self-test requirements concerning the program storage memory, data memory, other memory, clocks, processor exceptions, watchdog timers and controlled shutdown.
Additionally, error logging procedures, software design practices and documentation, handling of embedded batteries, sensor and input channel protection, and minimum number of outputs should be defined.
Some standards allow data transfer protocols to be defined by the suppliers.
However, ERTMS/ETCSSUBSET-027 defines downloading to a PC, the format and content of the messages exchanged, the communication protocol and the physical interface.
A universal definition on this, as well as on an evaluation tool for analysing the extracted data, would be welcome.

Mechanical and test characteristics

As well as on functional characteristics, the existing standards have diverse requirements on mechanical characteristics. It is important to bear in mind that the crash-hardened portion of the data recorder has to be the memory module, not the entire system.
The sequencing of tests is not yet included in any standard. This should be defined according to the order of a collision or derailment scenario as follows:
1. Impact shock.
2. Penetration.
3. Static crush.
4. Fire.
5. Fluid immersion.
6. Hydrostatic pressure.
In addition to the fire and fluid immersion tests already in the standards, a new standard should include tests with other flammable fuels like gasoline. Collisions at grade crossings including tank lorries have to be taken into account.
A magnetic field test described in GM/RT2472 (valid only in Great Britain) should be included in a universal standard. This requirement is applicable on all railways where track electrification is more common.

The rail industry is on its way to a point where data recorders with downloading and evaluation tools are logically defined under a universal standard. This much called-for standard needs
to evolve with the technology to include all new developments, such as on-board voice recording and web-based monitoring.

Enter 53 or at www.engineerlive.com/rii

Samuel Kruger is project manager, EKE Electronics Ltd, Espoo, Finland. www.eke.com