From asset record to Health score: setting up Health and Criticality scoring in Maximo Application Suite
Maximo Health is the part of the suite that answers a deceptively simple question: of all the assets I look after, which ones should I be worried about? It answers it with three numbers on a 0 to 100 scale. Health is the asset's condition, Criticality is how much it matters if that asset fails, and Risk is the two combined. Plot Criticality against Health on a matrix and the red corner is your worry list.
What surprises people is where those numbers come from. Health does not have a data model of its own. It is installed as a component of Maximo Manage, reads the asset records, meters and work orders that are already there, and scores them using rules you define. The score even lands back on the Manage asset record, in a Health field under Details. In 9.2 the scoring engine was reworked to calculate assets in batches rather than one at a time, which matters once you point it at tens of thousands of assets, but the formulas and contributors are the same. The hard part is not the engine. It is deciding what a score should mean, and then making sure Manage holds the data to calculate it.
The chain you are building has five links: the data in Manage, a saved query that picks the assets, a scoring group, one or more score types (Health, Criticality, Risk), and the contributors with their weights that actually produce the numbers. Each link depends on the one before it.
1. Start in Manage, not in Health
Before you open Health at all, look at the asset records you intend to score. The out-of-the-box contributors lean on a handful of fields that are optional in Manage and routinely empty: Priority in the Details section, and Installation Date, Expected life in years and Replacement Cost under Purchase Information. Add to that any meters you intend to score against, which must already exist on the asset and be receiving readings, and the work order and service request history if you want anything based on open work or failures.
Health knows this is where projects stall, which is why it ships with Work queues. Seven are delivered: Failing before PM, High probability of failure, Low health, and four Missing queues for criticality score, expected life, installation date and replacement cost, each with a count of remaining assets and an Open first asset link. Open them on day one. In our training environment the missing-data queues held between 450 and 1,437 assets each, which is the most honest view you will get of whether an asset register is ready to be scored.
2. Save a query that picks the assets
Scoring is done per group of assets, and a group is defined by a Manage query. In the Assets application, filter down to the set you want to score together, save the query, and tick Public so Health can see it. We used four centrifugal pumps at one site, which is small enough to reason about every number that comes out.
Choose the grouping with the scoring model in mind. A pump and a transformer should not share a Health formula, so they should not share a group.
3. Create the scoring group
In Health, open Score settings. It has five tabs: Groups, Score types, Contributors, Thresholds and Probability of failure analysis. On Groups click Create group. Give the group a name and description, choose Asset rather than Location, and under Configure scoring and DGA by choose Building scores. The other option, Default scores, applies the delivered calculations without letting you see inside them; for a first model you want to build the scores yourself.
Under Scored assets click Select, pick the saved query, and Create. Assets are added to and removed from the group automatically as they meet or stop meeting the query, which is worth remembering when an asset you expected to see is not there.
You will also notice a group called Default_scores already on the Groups tab, described as all assets of the system with activated health, risk and criticality. Since Health 9.0 it is activated in one action from the row's overflow menu, after which every existing asset and every asset added later is scored by it, on the schedule of a cron task in Manage. You cannot edit or delete it, though you can duplicate it and modify the copy. The practical effect is that assets you have never configured already have a number against them. More on what that number is made of in a moment.
4. Build the Health score
Open your new group. The Scores panel is empty, so click Add score. The Add a score dialog offers each of Health, Criticality and Risk twice: once with a Default badge and once without, alongside End of life and Effective age. The badge is the difference between a score that arrives with the delivered contributors already wired in and one that arrives empty. Choose the Default tiles for all three so that you can see what IBM thinks a score is made of, and the group page lists them with Active: N, Risk showing Criticality, Health under Score dependencies.
A score is nothing until it has contributors. The Contributors tab on Score settings is the catalogue, and the badges matter. Four carry a Default badge: RUL (remaining useful life, from installation date and expected life), OPENCMWO (the summed priorities of the asset's open corrective work orders), OPENSR (the same for open service requests) and PRIORITY (the asset's Priority field). Six older samples carry a Custom badge, FACAGESAMPLE, FACOPERATINGSAMPLE, FACMTBFSAMPLE, FACRULYEARSSAMPLE, FACYTDBGTCOSTSAMPLE and FACYTDREPCOSTSAMPLE, and their names tell you what they do.
This is also where you find out what the Default_scores group has been doing. Open any asset it has scored and the Score details panel shows the Health score built from OPENCMWO, OPENSR and RUL, weighted 33, 34 and 33 percent. So a default Health score is really saying: how much open work is there against this asset, and how far through its expected life is it. Useful, but it is not condition.
To get condition in, click Create contributor. The form offers Duplicate existing if you want to start from one of the catalogue entries. Otherwise give it a name, choose Asset, and pick a Data source of Meter or Formula. For a meter, Select the meter and Health tells you its type. We used a characteristic meter called Vibration, which has the interesting consequence that the formula is written for you as a skeleton: IF(METERVAL("VIBRATION")="Medium",?, IF(METERVAL("VIBRATION")="Low",?, IF(METERVAL("VIBRATION")="High",?,?))), with question marks to replace with the score each reading should produce. Under Value normalization give the best possible value and worst possible value, which is how a raw reading becomes a 0 to 100 contribution.
Back on the Health score page, the Contributors panel has four icons: a pencil to edit weights, a grouping icon to create a contributor group, fx to write a formula straight into the score, and + to add a contributor from the catalogue. Because you chose the Default tile, the Health score arrives with the same three contributor groups the default group uses, OPENSR, OPENCMWO and RUL at 34, 33 and 33 percent, each wrapping a copy of the default contributor so that you can change it without touching the original. That is a sensible starting point and an easy one to mistake for a finished model. Open a group, and the + inside it is where your vibration contributor goes; the pencil opens Edit contributor settings, where each contributor has a weight and a Zero override switch, which when on makes a contributor score of 0 drag the whole group to 0. Use it for the measurement that should trump everything else. Keep the weights totalling 100, then set the score Active.
5. Criticality, then Risk
Open the Criticality score and it holds one contributor group, PRIORITY, at 100 percent, wrapping a copy of the default Priority contributor. That is the whole of a first criticality model, and it is deliberate: it means the criticality conversation happens where it should, in Manage, in the Priority field, by people who know the plant. A more mature model scores criticality across safety, production, regulatory and financial consequence, each as its own contributor with its own weight. The mechanics are identical; the work is the workshop where the plant agrees the weights.
Risk is built from the other two, which is why the group page lists Criticality and Health as its dependencies. Open it and the contributors are not formulas at all but the CRITICALITY and HEALTH scores themselves, with no weights to set. There is nothing to configure beyond switching it to Active, and it will stay blank for any asset where either input is missing, which is the right behaviour.
6. Ranges, calculate, and look at the result
Each score type has ranges that turn a number into a colour, an icon and a label, and they are edited from the Score types tab. The delivered ones are worth knowing. Health is Good above 70, Fair above 40, Poor below. Criticality is lettered A to E in bands of 20, with A at the top. Risk is High above 40, Medium above 10, Low below, and End of life follows Health's 70 and 40. Each score type also has a Needs attention setting that says which end of the scale is the bad end. The thresholds are yours to change, and where you set them is a statement about how much you are prepared to act on.
The Thresholds tab next door is a different animal: it is where you set a mean-time-between-failures threshold in hours, days or similar, with Meets threshold, Doesn't meet threshold and No failure bands, and it feeds the MTBF chart on the Assets page rather than any score.
Now open Assets & Locations in Health. The list view shows Health, Criticality and Risk as coloured columns, and a Fix missing data link if the group has gaps. Once our group's scores were active the four pumps spread out: Health between 77 and 99, Criticality at 50 or 75 depending on their Priority, and Risk somewhere in between. Open an asset and the top row of cards shows Health, Criticality, Risk and Status, with Priority alongside. Below that, Score details lists each contributor's score, site and weight, and Health history plots the score and its contributors over time against the range lines. Change a reading or a field in Manage, then use Actions → Refresh scores on the asset to watch the number move.
Finally, switch to the Matrix view. Two delivered versions matter here, Criticality and Health and Criticality and Risk, each with criticality bands A to E down the side, the other score's bands across the bottom, and cells coloured by Need for action: High, Medium or Low. The gear icon opens Matrix settings, where you choose the axes, their direction and which cells belong to which category. Hover a cell for its definition and click it to list the assets inside. That grid, with your assets counted into it, is the thing the whole exercise was for.
The two things that catch people out
The first is editing a live score. You cannot add or change contributors while a score is active; the moment you try, a Deactivate score dialog tells you that before a score can be edited it must be deactivated, and warns that while it is inactive temporary errors might appear on the scored records, to be resolved when you activate it again. So on a production system the edit is a small change window, not a tweak: deactivate, change, reactivate, refresh.
The second is the confident wrong number. A score is always calculated, even when the inputs are poor; IBM's own documentation says that if contributor data is missing the score is calculated from whatever partial data is available, and that an asset with no Priority in Manage simply gets the lowest criticality. The default group gave every one of our four pumps a Health score in the high nineties, because none of them had open work orders and all of them had a recent installation date. That is not a statement about their condition; it is a statement about the data. The Work queues will tell you where the gaps are, and so will the contributor breakdown on the asset, which is why the first thing to do after the first calculation is to pick three assets you know well and ask whether you believe the number.
Why bother?
Because a Health score is the first place where everything else in asset performance management lands in one number: meter readings from Monitor, open work and failure history from Manage, the plant's own view of what matters in the Priority field. In 9.2 Manage also gained an Alerts application, with condition monitoring points able to raise an alert automatically when a reading crosses a warning limit, so the same meters that feed a Health contributor can now interrupt someone's day as well as move a score. And because the configuration takes an afternoon. Deciding what a score should mean, which contributors to trust and where the red line sits takes the workshop, the reliability engineer and the planner. That part is not a Maximo setting. It is the job.