1. How can we organise our field maintenance more effectively?

How does a work order run in your organisation, from planning through to completion? The planner looks for an available technician, the technician drives to site, gathers information, carries out the work and records the findings. At each of those points, time can be lost through awkward planning, extra kilometres, missing asset information or waiting for a specialist.

Suppose you carry out periodic inspections on assets across a large work area. In the system they sit as separate work orders. On the map, three inspections turn out to be only a few kilometres apart. By combining planning and location information, you can put that work into a single route. That saves travel time and stops scarce technicians having to enter the same area several times. For pipeline operators, water boards and infrastructure organisations, that geographical connection carries a lot of weight in planning.

But what is the use of good planning if the right information is missing in the field? On site it should be immediately clear which asset you are dealing with, what happened before and which instructions apply. With that information on a phone or tablet, your technician can see straight away what is needed and can already update the record while the work is being done. If a serial number is missing or a meter reading is wrong, the technician can complete that on the spot, with a photo if needed. Every work order then yields up-to-date data for the next inspection or maintenance decision.

And what happens if a technician sees an anomaly at an installation that cannot be explained straight away? The experienced colleague who knows this type of fault may be working dozens of kilometres away. With remote knowledge sharing, that specialist can look along live via video. The work does not have to wait for a physical visit, the technician gets help with the diagnosis and the findings can be recorded immediately. The same expert can then support several colleagues in a single day.

Planning, execution, registration and knowledge transfer all interlock. Tight planning achieves little if technicians in the field lack the right information. Mobile registration only helps so much if jobs are assigned poorly. Remote expertise has more effect when earlier diagnoses can be found again. Where does your process stall: in planning, travelling, searching for information, calling in expertise or reporting back? That is a useful starting point for improvement.

2. How do we decide whether to invest in maintenance or replace our assets?

Many maintenance programmes run for years on the same intervals. A pump gets annual maintenance because that was once laid down. Meanwhile the load may have changed, the failure rate may be rising, or measurements may show that the technical condition is still perfectly fine. Is that annual maintenance still the best choice?

For a usable condition assessment, you determine which signals say something about wear and failure risk. For one installation that may be running hours and vibration; for another temperature, corrosion, failure frequency or the availability of spare parts. From that follows which data you need and which measurements are still missing.

The quality of that data is often a constraint. Work orders are there, but cause codes are missing. One technician records a defect as a ‘bearing problem’, another as a ‘mechanical fault’. On older assets, serial numbers or links to components are missing. A health score can therefore look convincing, while the underlying data still contain gaps.

Condition based management requires fixed registration agreements and clear responsibilities. Which data do technicians record, and how do they do that? Which fields are mandatory? What should a maintenance partner hand back after the work? Configuration management also counts: asset data need to stay usable throughout the lifecycle. The quality of the registration determines how much trust you can later place in the condition measurement.

With reliable condition information, you can treat assets that look the same on paper in different ways. Take two pumps that are twenty years old. The first runs stably and needs little corrective maintenance. The second fails more often, parts are becoming scarce and repairs take longer. Age alone then does not give you enough to go on. Condition, risk, performance and cost show which pump can continue and which is a candidate for replacement.

That also helps you substantiate investments more clearly. Which assets need budget first? Where is extra maintenance still justified? And when do the costs and risks of carrying on no longer outweigh replacement? Condition information gives you a factual basis for that trade-off.

3. How can we use AI to manage our assets better?

AI quickly comes onto the table once the migration is behind you. Where can AI take concrete work off people’s hands in your maintenance process? Look, for example, at tasks where staff spend a lot of time searching for information, entering data, classifying work orders or preparing analyses.

A planner wants to know which faults have been reported for a particular asset over the past six months. Normally that means searching through different screens and filters. With an AI assistant, the question can be asked in plain language. That shortens the search and helps colleagues who use Maximo less intensively. New colleagues can get support during a task without having to memorise every step in advance.

Registration is another concrete application. One employee writes ‘motor is getting hot’, another ‘overheating’ and a third selects ‘mechanical fault’. For a report, those are three different records of what may be the same problem. AI can interpret the description and suggest a suitable problem code or category. The data become more consistent and analyses rest on a more reliable foundation.

For reliability engineers, AI can take over preparatory work in analyses of failure modes, effects and measures. The system offers a first proposal based on the available information; the engineer tests this against the installation and the maintenance concept. The professional judgement stays with the engineer, while less time is needed for searching and working things out.

Document processing also offers practical applications. Data from contracts, purchasing documents or technical documents can be recognised and structured automatically. Staff have to type over fewer details and the chance of entry errors falls. In IBM MAS 9.2, new capabilities have been added for this.

How do you then know whether such an application actually delivers? Choose one bounded process and agree in advance what you will measure. Compare, for example, search time, the percentage of correctly classified work orders or the number of manual entry actions. Also record which data the application may use, which security requirements apply and who checks the outcomes. That way you can assess the effect on the basis of results from your own process.

Decide where you want to improve first

Field service, condition based management and AI each place different demands on your processes and data. Which choice should take priority for your organisation depends on the maturity of your asset management and on what is already well set up.

With the Gemba Growth Model you map that starting position. You can see which preconditions still need attention and you can substantiate which improvement should come first for your organisation.

Which step fits your organisation? Get in touch with Gerben ter Horst on +31 (0)6 14 57 21 19 or g.terhorst@gemba.nl and find out which options within IBM MAS line up with your asset management goals.