MRT: Preventive maintenance is key

I agree with Dr Tan Wu Meng’s commentary “Can big data help tackle MRT woes?” (July 15) that the key step to tackle our MRT woes is to get better data and knowledge of the operating system.

Despite inheriting a corporation with nearly 30 years’ operating experience, as well as having the benefit of hindsight from the December 2011 disruptions and the Committee of Inquiry’s resultant recommendations, the new management unfortunately failed to prevent recurrent weaknesses.

As a retired engineer, I believe in scheduled preventive maintenance based on the useful lifespan of components and records of performance data, rather than relying on predictions about mean time between failure.

Preventive maintenance programmes are planned by established manufacturers for servicing or replacing component parts in the system supplied, and maintenance should be done according to the manufacturer’s manual, at recommended periods and in checklists.

Periodic monitoring of the conditions of components and sub-systems must be done faithfully by competent maintenance teams to ensure parts are replaced as recommended, regardless of how well these continue to perform.

There could be hidden problems, however, if maintenance work is based on “mean time between failure” predictions, which are known unknowns. Scheduled preventive maintenance is ultimately more reliable, since it is proven on component lifespan data from manufacturers. In other words, likely failures when reaching the useful lifespan are prevented before they can happen. Post-mortem details of replaced parts would provide valuable big data for future reference.

In large installations, such as ports, airports and power stations, it is a best practice to not outsource preventive maintenance services to contractors and sub-contractors. Different levels of contractor competency would make tracking of essential data difficult.

Additional monitoring devices would keep tabs on the hardware and software to narrow down problem areas for faster action. This would ensure consistent uptime in train service and the capability to resolve disruptions faster.