Maintenance · Assets · BMS · AI
Predictive maintenance
Act before the failure, only where it matters.
Most buildings are maintained in one of two ways: on a calendar, even when nothing is needed, or after a failure, when it costs most. With the data the building already produces you can see a problem as it starts and decide when to act. We help you choose the right assets, connect the data and turn alerts into work orders.
Four ways to maintain
| When you act | What it needs | |
|---|---|---|
| Corrective | After the failure | A crew on call and spare parts |
| Preventive | At fixed intervals, in days or running hours | A maintenance plan |
| Condition-based | When a measurement crosses a threshold | Sensors or BMS points and thresholds |
| Predictive | When data trends point to an approaching failure | Data history, models and a process to act |
Definitions from EN 13306.
The P-F curve
How early can you see the failure?
Every failure has a point P, where the defect becomes measurable, and a point F, where the asset stops working. The time between them is the time you have to act. Pick a method and see how much is left.
Time to act
AI on trends
Which assets it is worth it for
Monitoring has a cost. It pays where a failure stops the service, costs a lot or puts people at risk.
- Yes
Air handling units
Filter differential pressure, motor and bearing vibration, power draw.
- Yes
Pumps and motors
Vibration, bearing temperature, running hours, power draw.
- Yes
Chillers and heat pumps
Circuit pressures and temperatures, efficiency over time, number of starts.
- It depends
Switchboards and substations
Periodic thermography and temperatures. Condition-based maintenance is often enough.
- It depends
Lifts and escalators
Data often sits in the lift contractor's system: you need access to the telemetry.
- Usually not
Lighting, windows, finishes
Monitoring them costs more than replacing them: calendar or run-to-failure is better.
The data is almost always there already
BMS
System trends and alarms
Meters
Energy, water, gas, by system or floor
CMMS
History of failures and work orders
Documents
Manuals, datasheets, warranties
BIM
Code and location of each asset
How to start, in five steps
- 1
Critical assets
Assets are ranked by failure impact and likelihood. Start with a few, the ones that matter.
- 2
Data check
Which BMS points exist, how much history there is, whether it is reliable. Where measurements are missing, a few sensors are added.
- 3
One code per asset
BMS, maintenance software and model use the same code. Without it, the data does not connect.
- 4
Thresholds, then models
Simple thresholds first, which work right away. Then trend models, once there is enough history.
- 5
From alert to work order
Every alert opens a work order. You measure what changes: failures, emergencies, spare parts.
What you need to start
- 1The list of your main systems, even in Excel
- 2Access to the BMS or an export of the trends
- 3An export of the history from your maintenance software
- 4The failures that cost you most in recent years
Frequently asked questions
How can I optimise the maintenance of my assets?+
Start with critical assets, where a failure stops the service or costs a lot. For those, use the data you already have, BMS, meters and maintenance history, first with simple thresholds and then with trend models. Every alert becomes a work order. Everything else stays on calendar maintenance.
What is the difference between preventive and predictive maintenance?+
Preventive maintenance acts at fixed intervals, whether the asset needs it or not. Predictive maintenance acts when the data shows a failure is approaching. The first is easy to plan; the second avoids unnecessary work and catches failures that happen between two intervals.
What data does predictive maintenance need?+
BMS trends, consumption from meters, the history of failures and work orders from the maintenance software and, where needed, vibration or temperature sensors. All of it must use the same code for each asset, otherwise the data does not connect.
Do you need a digital twin for predictive maintenance?+
No, you can start from BMS and maintenance data. A digital twin helps when there are many assets: it shows where each asset is, what it serves and which documents relate to it, so the crew arrives with the right information.
How much does predictive maintenance save?+
It depends on the assets and on what failures and calendar work cost you today. That is why we start from your data: the calculator on the site gives a first estimate, then we check it together on a group of assets before extending the work.
BIM consulting for the Italian market
- BIM consultingDigital, at the service of those who build.→
- Outsourced BIM managerRoles covered. Decisions that stay yours.→
- BIM support for contracting authoritiesEvery duty, at the right moment.→
- Information requirements (EIR)Requirements written first. Useful models later.→
- BIM trainingSkills by role, on your own projects.→
- Digital twinThe model that keeps working after handover.→
- CDE managerEvery file in its state. Every step with a name.→
- ESG for buildingsConsumption, emissions and indicators taken from the asset's real data.→
- AI for asset managementAI agents that read the asset's model, systems, maintenance and documents.→
Which assets do we start with?
Tell us about the building, its systems and the failures that weigh on you most. We will reply within one working day.