← All articles Insights

Machine Downtime Tracking: Reason Codes, MTBF and MTTR Explained

18 Nov 2026 · By NKE Team

Most plants know their machines stop too often. Ask which cause costs the most time, though, and the answer is usually a gut feeling. The problem is rarely a lack of data; it is that machine downtime gets recorded without structure, so nobody can analyse it. This article explains how to separate types of downtime, design downtime reason codes, choose a data collection method, and use MTBF and MTTR to attack stops at the root.

What Machine Downtime Is, and Why a Single Total Is Not Enough

Downtime is time when a machine should be producing but isn't. In practice, split it into at least three groups, because each one is fixed in a different way:

  • Planned downtime – breaks, start-of-shift meetings, scheduled PM, or no production plan.
  • Unplanned downtime – breakdowns, waiting for material, waiting for an operator, waiting for QC, or quality problems.
  • Minor stops / micro stops – short, repeated interruptions such as jammed parts or a sensor that misses. Each looks trivial, but over a shift they can add up to a significant amount of time.

A total downtime figure tells you how much time was lost, not where to act. Grouping and assigning causes is what turns the number into an action plan. It is also the basis of the Availability factor in OEE (see What is OEE? How to calculate it and the six big losses).

Designing Downtime Reason Codes People Will Actually Use

Use a two-level hierarchy

A common approach is a top level of broad categories with specific reasons underneath, for example:

  • Machine – spindle fault, hydraulics, control alarm, coolant system.
  • Material / parts – no material, raw part out of tolerance, waiting for transport.
  • People – no operator, waiting for maintenance, training.
  • Method / quality – changeover, tool change, waiting for measurement, offset adjustment.
  • Planned – break, PM, no schedule.

Rules that keep the codes useful

  • Keep the list short. Operators should be able to pick a reason in seconds. If the list scrolls for pages, people will pick whatever appears first.
  • Make codes mutually exclusive. Every event should fit exactly one code.
  • Watch the "Other" code. If its share keeps growing, your list doesn't cover reality and needs a review.
  • Use shop-floor language. Build the list with supervisors and maintenance, not only at a desk.
  • Tailor by machine type. A CNC lathe and a press stop for different reasons and deserve different code sets.

How to Capture Downtime: From Paper to Machine Signals

1. Manual logs

You can start today at no system cost. The drawbacks: times get rounded, short stops rarely get written down, and someone has to type everything into a spreadsheet later. Paper is a good way to trial your code structure before investing in a system.

2. Machine signals

A machine monitoring system reads status from the CNC or PLC, or from the stack light, so start and end times are accurate and micro stops are no longer invisible. But the machine only knows that it stopped, and perhaps the alarm number. It cannot know that it was waiting for material.

3. Hybrid: the machine times it, the operator explains it

Many plants get good results by letting the system detect stops automatically and asking the operator to select the reason on a terminal at the machine. Points to decide:

  • Set a time threshold defined by the plant. Stops shorter than it are classed as minor stops automatically, so operators aren't asked every time.
  • Stops longer than the threshold must be assigned a reason before restarting or within the shift.
  • Link to the shift calendar and production plan so planned downtime is separated automatically.

Analysing Downtime: Pareto, MTBF and MTTR

Run Pareto by duration and by frequency

Rank causes twice. By total duration, you see the big, long stops such as major breakdowns. By count, you see small, frequent stops such as jammed parts, which can often be fixed with a fixture or sensor change. Looking at only one view can hide real opportunities.

MTBF and MTTR

For breakdown-related stops, two widely used metrics are:

  • MTBF (Mean Time Between Failures) = total running time ÷ number of failures. It indicates how reliable the machine is.
  • MTTR (Mean Time To Repair) = total repair time ÷ number of failures. It indicates how quickly the team gets the machine running again.

Hypothetical example: a machine runs 400 hours in a month, fails 8 times and needs 16 hours of repair in total. MTBF is 50 hours and MTTR is 2 hours. A low MTBF points you towards your PM plan and failure root causes; a high MTTR points towards spare parts, repair instructions and how maintenance is called.

Don't ignore changeovers

In high-mix CNC work, setup time is often near the top of the list. Once the data makes that clear, you can launch a focused SMED setup reduction project.

Common Mistakes in Downtime Tracking

  • Collecting data nobody reviews. If it isn't discussed in daily or weekly meetings, entries soon become careless.
  • Using data to blame people. Operators who feel watched will choose "safe" codes instead of the real cause.
  • Changing codes so often that history can't be compared. Review on a fixed cycle and document every change.
  • Mixing planned and unplanned downtime. The numbers end up looking worse, or better, than reality.
  • Not closing the loop on actions. Each top cause needs an owner, a due date and a follow-up check using the same data.

The data system is only a tool. What actually reduces downtime is fixing root causes, sometimes by changing work methods, sometimes by upgrading equipment or adding factory automation at the bottleneck.

FAQ

Should we track downtime on every machine from day one?

Not necessarily. Start with the bottleneck machine or line, trial your reason codes, refine them, then roll out to other machines. You learn faster and avoid confusing the shop floor.

Can we collect data from older machines without a communication port?

Usually, yes. Run/stop status can be taken from stack lights, relays or motor current, and operators enter the reason via a terminal or buttons at the machine. The detail will be lower than on machines connected directly to the controller.

How is downtime different from OEE?

Downtime is the record of stops and their causes. OEE is an overall indicator combining Availability, Performance and Quality. Well-coded downtime data explains why Availability is low and where to act.

If you want to start tracking downtime properly, from designing reason codes and connecting machine signals to OEE and energy monitoring dashboards, the NKE team is happy to review your shop floor with you. Contact NKE, call 02-746-7196 or add LINE @nke.th.

Cover photo: jotoler / Pixabay

Want to talk about automating your plant?

Talk to an engineerChat on LINE