Reliability Centered Design: The Cheapest Place to Fix a Failure Is the Drawing

Every article in this series so far has worked on equipment that already exists. Reliability centered design (RCD) works on equipment that doesn't exist yet, which is when you have the most freedom and the lowest cost of change.

This is the fifth article in our Reliability Engineering for Maintenance series, and it returns to Stage 1 of the full P-F curve. It also connects to reliability centered maintenance, which works best when design and strategy are developed together.

The idea

Design determines an asset's inherent reliability. Maintenance can protect that reliability but can't exceed it. If a design makes a failure likely, or makes the equipment hard to inspect and repair, every year of operation pays for that choice.

RCD means treating reliability, maintainability, and operability as design requirements from the start, with the same weight as capacity and cost.

What reliability centered design looks like

• Define reliability requirements in operating terms. "Must run continuously between planned outages" is a better requirement than "must be reliable."
• Analyze failure modes early. FMEA or FMECA at the design stage shows where a design is vulnerable, while changes are still cheap.
• Design in redundancy where consequences justify it. Use criticality to decide where a standby unit or bypass is worth it, and where it is wasted cost.
• Use margin and simplicity. Derating components, reducing part counts, and standardizing on proven equipment lowers failure opportunities and spares complexity.
• Design for maintainability. Provide access for inspection and lubrication, lifting points, isolation points, and clearances for removal.
• Design for condition monitoring. Include sensor mounting points, test ports, and instrumentation so the early warning of Point P is available from day one instead of retrofitted.
• Compare total cost of ownership, not just purchase price. The cheapest unit to buy is often not the cheapest to own.

Designing out maintenance

One of the strongest RCD ideas is eliminating the maintenance task itself. A sealed bearing arrangement, a better coupling, a self-draining layout, or a different material can remove a recurring failure that no PM could fully control. Your bad-actor list and root cause analyses are a catalog of designs worth fixing, and they should feed into specifications for the next project.

Close the loop

Design improves only if lessons make it back. Failure data, RCA findings, and operator and technician feedback should flow into standards, specifications, and design reviews. Without that loop, the same weaknesses get purchased again on the next project.

Include operations and maintenance people in design reviews. They know which valves can't be reached, which instruments can't be read, and which pump always seems to be in a bad spot.

Common mistakes

• Price-driven selection with no look at operating cost and failure history
• No maintenance voice in design reviews
• Treating monitoring as optional and discovering later there's nowhere to mount a sensor
• Ignoring the real duty, such as specifying equipment for a nominal condition the plant rarely operates at
• Never feeding back lessons learned