Maintenance planning

Machines wear out. The question is whether you stop them on your terms or theirs. There are four broad approaches:

• Reactive: run until it breaks, then fix it. Cheap until it happens on the bottleneck during the festive rush.
• Preventive: service on a fixed schedule (every 500 running hours, every month), whether the part needs it or not. Predictable, but you sometimes replace parts that had life left.
• Predictive (condition-based): measure vibration, temperature or oil quality and intervene when the data shows wear. Fewer surprises and fewer wasted services, but it needs sensors and analysis.
• TPM (total productive maintenance): operators own basic care such as cleaning, lubricating and inspecting, so small problems are caught before they become breakdowns.

Two numbers describe reliability:

• MTBF (mean time between failures): how long it runs between breakdowns.
• MTTR (mean time to repair): how long a repair takes.

Availability is roughly MTBF ÷ (MTBF + MTTR). A machine that fails every 190 hours and takes 10 hours to fix is available about 95% of the time. You can improve it by failing less often, or by fixing faster with spare parts on hand and trained technicians.

Planning matters as much as method: schedule stops in low-demand periods or alongside changeovers, protect the bottleneck most carefully, and build stock ahead of long shutdowns.

Availability, worked through

Choosing the approach for each machine

How planned work flows

1. Notification: an operator, an inspection or a sensor reports a problem or a due task.
2. Planning: the planner defines the job, the parts, tools, permits and skills needed, and how long it takes.
3. Scheduling: the job gets a slot agreed with production, ideally during a changeover or a low-demand shift.
4. Execution and closing: technicians do the work and record what failed and why.
5. Analysis: failure records feed improvements and adjust maintenance intervals.

ERP systems, including SAP’s plant maintenance module, manage this flow with maintenance plans, notifications and work orders.

Measures that matter

Planned work as a share of all maintenance work, schedule compliance, backlog in weeks of work, MTBF and MTTR for critical assets, and maintenance cost compared with the value of the equipment.

Most failures aren’t about age

Reliability-centred maintenance (RCM) grew out of airline studies in the 1960s and 1970s, summarized by Nowlan and Heap (1978). A key finding was that most failure modes of complex equipment were not related to age: the chance of failure didn’t rise steadily with time. For those failure modes, replacing parts on a fixed schedule doesn’t prevent failures and can even add early-life failures to freshly installed parts. That shifted maintenance towards condition monitoring and towards asking, for each failure mode, which task actually reduces its risk.

The P-F interval

Many failures announce themselves. The point when a developing failure first becomes detectable (P, potential failure) comes some time before the equipment stops working (F, functional failure). If vibration analysis can detect a bearing defect about six weeks before it seizes, inspecting every three weeks catches it with time to plan the repair. The rule: inspect more often than the P-F interval, and plan the fix inside it.

Predictive maintenance in practice

Sensors measure vibration, temperature, current, pressure or oil condition, and models flag patterns that typically come before failures. It works best where failures have a measurable signature, data is available over enough failures to learn from, and the cost of a surprise failure is high. Pitfalls: too many false alarms (technicians stop trusting them), too few historical failures to train on, and alerts that arrive without a process to plan and schedule the response.

Maintenance and the supply chain

• Spare parts: critical spares are classic intermittent-demand items, so stocking them is a service-level decision weighed against downtime cost.
• Shutdowns: planned plant shutdowns need stock built ahead, customers informed and contractors booked, and are managed like projects with a critical path.
• Bottleneck first: protect the constraint with the best maintenance, spares and response times, since its downtime is the plant’s downtime.

Criticality comes first

Before choosing tasks, rank equipment by what a failure would cost (safety, environment, output, quality, repair cost) and how likely it is. Failure mode and effects analysis (FMEA) does this systematically. High-criticality assets get predictive monitoring and spares; low-criticality ones may simply run to failure.