Analyse and Advise
“The process of analyzing the data sources and using this to advise on next best actions.”
We have now collected a lot of data and are using the tools available to us in core Maximo. Now is the time to start exploring the tools available to us as part of Maximo Asset Performance Management (APM).
Level 1
The first APM product to invest time in will be Maximo APM Asset Health Insights – Maximo Health. Level 1 is setting up health scores for assets and locations. This requires the creation of a scoring method, which has a set of drivers with a weighting set for each driver, for example age, cost, performance. Each driver has a formula which can combine different factors, for example, year to date maintenance costs versus budget, or year to date costs versus replacement cost. Weightings are applied to each factor that is part of the same driver. Each factor has its own formula which derives a value from data anywhere in Maximo, this uses Maximo formulas. A range of percentages is used to create a colour coded baseline, for example 0-50 (poor) – red, 50-85 (fair) – yellow, 85-100 (good) – green.
You perform asset health scoring for sets of assets that you define with a query. Each asset in a set is evaluated against its scoring method to derive a health percentage which is tracked over time. The average and lowest health scores of the assets at the same location are calculated and these values are similarly set at higher levels in the location hierarchy. For example, in a water utility you may be looking at the health score of pumps, at a high level you can see that there is an issue and then drilldown to find it. If maps are set-up you can also visualize this from a map. You can take actions when you find an asset that requires attention.
Level 2
Level 2 also uses Maximo Health, and this is helping to make repair or replacement decisions. You identify assets that are near end of life and make plans for whether they will be refurbished, replaced or decommissioned and include these decisions as part of long-term budget planning.
It is key to have expected end of life dates and to perform condition assessments. You then make a query for those assets which are within the end of life date in the time horizon for which you perform long term planning.
For each asset in the time horizon you review the condition and expected end of life and add a plan and pick the plan type, for example refurbish. Based on this you assess by how much the end of life will be extended and a target date for the refurbishment. You add other details like the job plan to use, labor hours and downtime hours.
When you have made plans for multiple assets, which may need to be adjusted based on available budget, you can then implement the plan by creating work orders.
Level 3
Level 3 uses Maximo Predict or Maximo APM Predictive Maintenance Insights to give it its full title. This is a cloud-based solution with five predictive models. The model scores are integrated with Maximo Health.
Assets are generally maintained on a pre-determined schedule that doesn’t consider how they are being used which can lead to a lot of maintenance efforts as being ineffective. Preventive Maintenance is often performed too frequently, and it doesn’t necessarily achieve increased uptime of those assets.
Predictive maintenance is founded on a condition-based model, ideally the condition would be monitored with IoT sensors, historical maintenance data, operational logs and sometimes historical or current weather conditions. The predictive models use this data and suggests which assets need maintenance and when. Specifically, the five model templates are:
- Days to failure
- Predicted failure probability
- Factors that contribute to a predicted failure
- Anomaly Detection
- Failure Probability Curve
Asset groups are sets of similar assets that can be scored and analysed using the same approach. An asset event data file for the assets in the asset group is compiled containing failure events, maintenance events and operating data. This is analysed, and the models determine whether the asset is being under maintained, over maintained or is well maintained and makes a prediction of the next failure date. You then adjust accordingly.
Level 4
Up to now our focus has been on data that you may or should already have in Maximo. None of it is real-time, it is all historical although some of it may be events occurring in the last 24 hours. Level 4 moves towards the real-time element and captures other data for which Maximo is not the source, data from SCADA, historians and IoT sensors, and weather, all data captured through sensors.
There is a good reason why this data is not in Maximo, sensor data captures a lot of data and it needs analytics to make sense of it. Maximo Monitor is the most recent of the Maximo APM products and should be used as a filter for Maximo Health. The health scores, repair/replace decisions and predictive elements that we have discussed in levels 1 to 3 will all benefit from being enriched by sensor and weather data.
Maximo Monitor has four distinct elements:
- Connect uses data from sensors, SCADA and historians and maps it through to asset data in Maximo
- Prepare uses advanced analytics and anomaly detection algorithms to set meaningful alerts
- Visualise – build custom dashboards with pre-built widgets to show the output from the prepare element. The widgets are used to monitor key drivers, OEE, availability, uptime, etc.
- Investigate – Allows you to respond to the alerts to find root cause and take automated actions by drilling down from a plant, to a system, component, asset or sensor.
Maximo Monitor uses the IBM Watson IoT Platform with its integration into Watson tools, applications, and APIs that provide an array of cognitive services and analytics.
Level 5
This uses Maximo Assist, it has two elements:
- Assist AI ingests data from Maximo and other systems and unstructured data from documents, photos, videos, etc., to provide a virtual assistant to a technician. Natural language is used to ask questions, and recommendations are returned to the technician. This has been integrated into the Work Execution work center and so you will probably want to rollout the use of work centers first or at least in parallel.
- Assist AR is an augmented reality assistant that connects experts/peers to the field technician. It uses the video camera on a smartphone or tablet so that the expert/peer can see what is going on, they then use AR-tags to annotate the video and interact with the technician. The whole becomes an instruction of how to fix an issue with an asset, this may subsequently be found by the Assist AI element at a later point.
Both the Assist AI and Assist AR help to address the issue of a loss of knowledge with an aging maintenance and engineering workforce.
Leave a Reply