A-FMDS User Guide
Day-to-day use of the Daily Management module in ProAlert Core
Day-to-day use of the Daily Management module in ProAlert Core
This guide covers how to use the screens. For why A-FMDS exists and how the pieces fit together, read the A-FMDS System Overview first. It explains the Target → Gap → Problem → Action loop this guide assumes you know.
A-FMDS (the Automated Floor Management Development System) lives under the AFMDS menu in the main ProAlert Core navigation bar. It is available to every signed-in user.
| Menu item | Opens |
|---|---|
| Daily Management Board | The live board of recent gaps and problems. |
| Problems | The full problem list, with filtering and manual creation. |
| Actions | Every action across all problems. |
| Escalations | Escalation rules and the events they have fired. |
Every A-FMDS screen shows a small connection badge in the top-right (Connected). When it is green, the screen updates itself in real time as other people raise gaps, create problems, and complete actions: no need to refresh. All times are shown in your local time zone.
The Daily Management Board is the home screen of A-FMDS: the one you would put on a TV by the line. It has two live lists:
Every gap that has been detected, newest first: the metric, how far the actual missed the target, the severity, and a Promote button.
Every problem, newest first: number, title, category, severity, priority and status. Click a problem number to open it.
Severity is colour-coded everywhere in A-FMDS, so the board reads at a glance:
Critical Warning Info
Above the tiles is a window selector with two choices:
Each tile carries a large figure and a quieter one beneath it, and they answer different questions:
The window deliberately scopes activity only. A gap raised last week and still open is real outstanding work, so it keeps counting no matter which window you pick; a board that hid it because it fell outside "today" would be lying by omission. For the same reason an SQDCP cell stays red while it holds a critical item, even on a quiet day when nothing new was raised.
The production tiles read from the previous day's roll-up. When there is no figure to show, the board says why rather than showing a bare dash:
Note that production metrics are computed as a nightly roll-up, so running a work centre today will not populate those tiles until the next night's run.
The OEE (Overall Equipment Effectiveness), scrap, downtime and safety tiles are doors, not just readings. Click one and it opens that metric's trend, the first step of a five-level path that ends at the countermeasures being worked. The whole path is described in Drilling into a Metric.
The board answers "what is wrong right now." The SQDCP Grid button answers "how have we been": the shape of the paper wall board, with the SQDCP (Safety, Quality, Delivery, Cost, People) categories down the side and working days across the top, one coloured square per intersection. It opens on the last 14 working days, with 7, 14 and 30 day buttons.
The colours are the same everywhere in A-FMDS: Clear Info Warning Critical
On development and test servers the board shows a red Raise Test Gap button. It injects a sample scrap gap so you can watch the live flow end to end without waiting for a real event. It does not appear on production.
The Quality tile measures units that passed without quality scrap or rework:
FTT = (units made - first-pass failures) / units made x 100. A record
flagged as both counts once. With 1,000 units, 20 quality rejects and 30 separate
reworked units, FTT is 95%. OEE Quality can differ because it uses the
Affects Quality flag. If FTT says not configured, an
administrator must flag at least one rework type in Scrap Types.
The breakdown and Pareto explain all first-pass failures by asset, product, shift,
reject code and stage.
Record customer-found defects through Quality > Customer Defects. External PPM (parts per million) divides affected units by units produced in the attributed period, then multiplies by 1,000,000. It uses production output, not shipments. The production lot date takes precedence, then a stated production date, then receipt date as a labelled fallback. The trend is plant-wide; the breakdown groups by product or customer and drivers rank reject codes. Non-countable output uses a percentage.
When enabled, internal scrap trends, breakdowns and Paretos show value and recorded weight beside the rate. Cost uses product cost plus route steps through the scrap asset when that step can be identified. Product-only estimates and missing costs are identified; missing weight is not measured as zero. Public displays require a separate cost-display opt-in. These figures do not replace the rate or its target.
Some EdgeSense sensors report a person's action rather than a machine fault: a
touch pad pressed, a light gate broken. When a workflow rule for one of those
sensors is set to require disposition review, ProAlert still raises the gap, but
it also holds the gap for a human to look at before deciding what it becomes. The
queue for that decision is at /Afmds/Dispositions, titled
Device Event Review Queue.
Each row gives the gap number, which links to the gap itself, the asset, the rule that fired, when it was detected, and a small video clip if the system captured one. The clip is the point of the screen: you watch what actually happened before you judge it.
Two buttons sit on a row awaiting review.
If that inspection comes back with no fault found, the row returns to the queue with a No Fault Found, Awaiting Coachee badge and a single Assign Coachee button. Maintenance has ruled the machine out, so the matter becomes a coaching one and you name who gets coached.
What you see in the queue depends on the rules. A workflow rule can name a reviewer role, in which case only people holding that role see its events; a rule with no reviewer role is visible to everyone. Two people opening the queue can therefore see different rows, and that is intended. Those fields are set on the rule itself in EdgeSense device management.
There is no menu entry for this screen yet. Reach it by typing
/Afmds/Dispositions or by keeping a bookmark. A
Back to Board button returns you to the A-FMDS board.
The green cross calendar shows each day by its worst incident: green means no incident recorded, amber an incident without a recordable injury, red a recordable or lost-time injury, and future days stay blank. Select a day for its incidents or navigate to an earlier month. The calendar follows the Business Resource filter, although the safety trend remains plant-wide. Days since the last recordable injury appears only when one exists in the history.
Select the incident type and coded hazard when recording an incident.
Administrators maintain these at /Admin/Safety/Lookups. Type
describes the outcome; hazard supplies the safety Pareto's cause. Uncoded
incidents remain visible.
The People area links to Skills Matrix, also available from Workforce Administration. Compare people against roles, proficiency and certification. Fewer than two qualified people for a role is a coverage risk; lapsed certifications do not count. Use expiring and lapsed indicators to plan training and renewals.
The Kaizen tile counts ideas implemented in the selected period by completion date. Its plant-wide trend has no asset or driver breakdown. The Kaizen dashboard adds cumulative submitted, in-progress and implemented series with range controls; compare the curves to see whether implementation keeps pace with ideas.
The Kaizen Ideas list carries an Assigned To column showing the person's name. An idea nobody has picked up shows a greyed Unassigned. The same name appears under the progress stage on each card in My Submissions, and is omitted there when the idea is unassigned. Exporting the list to a spreadsheet carries the same names across, so a printed list matches the screen.
Every metric tile on the board opens the same path, five levels deep. Each level is one click on the level before it, and each page carries buttons back up the path so you can retrace your steps at the same range. This section walks the path for scrap; OEE, downtime and safety follow it the same way, with the differences noted where they occur.
Click a tile and the metric is plotted across the last twelve months, one point per month. Where supported, the range buttons include Hour, Shift, Day, Week, Month, Quarter and Year. Each bucket is a real calendar unit: a week starts on Monday and a month is the month, so the label under a point is the period it covers. Four cards above the chart give the Latest, Average, Best day and Worst day for the window.
A plant that reports to its customers in parts per million can have scrap shown that way everywhere: the tile, the trend, the breakdown and the drivers all read Internal PPM instead of Scrap Rate. That is an installation setting made when ProAlert is set up, not a button. PPM only means something when the output can be counted, so if anything produced in the window is measured in a unit that is not countable, the page falls back to a percentage for that window and says so above the chart.
The breakdown shows one bucket split by one dimension, worst contributor first, with each row's Share of the total and the quantity it was measured over (Units produced for scrap, planned hours for downtime). Bars are scaled against the largest row so the shape reads at a glance. The buttons above the table switch the dimension:
| Metric | Split by | Ranked by (Level 4) |
|---|---|---|
| Scrap and FTT | By asset, By product, By shift | By reject code, By stage |
| Downtime | By asset, By product, By shift | By call type, By failure mode |
| OEE | By asset, By product, By shift | Availability, performance and quality decomposition |
| Safety | By asset, By location | By hazard category; uncoded incidents remain visible |
Three buttons sit at the top of the page: Trend takes you back up at the same range, Drivers takes you down to Level 4, and Gaps and problems opens the category's list for that day. The last one appears only on a single-day bucket, because a month has no one day to open.
The safety trend does not follow the board's line filter, and the page says so. The nightly roll-up it reads counts incidents for the whole plant. One level down, the breakdown reads the incidents themselves, which do record an asset and a location, so a safety point splits by asset and by location and opens the incident list behind each row. Showing the plant figure under a line's name at the trend would be the more comfortable option and the wrong one.
The drivers page is a Pareto. Rows are ranked largest first, each with its share and the running Cumulative percentage, and the rows that make up the first 80 percent of the total are highlighted. Three cards above the table give the total, the number of Distinct drivers, and how many of them make up that 80 percent, which is the number that tells you whether one thing is hurting you or twelve. The buttons switch the driver taxonomy: for scrap, By reject code or By stage; for downtime, By call type or By failure mode. Safety ranks coded hazards and external defects rank reject codes. A bar chart with a cumulative percentage curve accompanies the table; select a bar or row to investigate its driver. Click a row and you are at Level 5 for that driver.
Levels 1 to 4 narrow down to a cause. Level 5 looks forward and asks whether the countermeasure worked, which is a different question and gets a different shape. The one driver you chose is plotted at Week, Month and Year side by side, because a defect that looks beaten over a week and flat over a year has not been beaten, and a single range would let you believe the first of those.
Beneath the charts is the Countermeasures table: problem number, countermeasure, status, due date and owner, read straight from the problems and countermeasures already recorded. A countermeasure nobody has been assigned to carry out is badged No one assigned, which is often the most useful thing on the page. The note above the table is honest about its scope: it lists what is being worked in the same category and area, not countermeasures tied to the one defect code above it, because ProAlert does not yet link a reject code to a problem.
A choice the engine cannot honour is not offered, and a note above the chart says why. These are the notes you will meet and what to do about each:
| What the page says | What it means |
|---|---|
| Plant-wide OEE has limited ranges | Plant-wide OEE is offered at Shift and Day. OEE cannot be averaged across days without the production time behind each one, which the plant-wide roll-up does not keep. Narrow the board to a Business Resource and Week, Month, Quarter and Year come back, computed from that line's own production time. |
| "This trend is plant-wide" | Safety, with a line filter in force. The line filter is ignored at the trend and honoured at the breakdown. |
| "Shown as a percentage because not everything produced in this period is measured in countable units" | PPM is configured for the plant but withheld for this window. Run of a product with no unit of measure, or one measured by length or weight, is enough to withhold it. |
| No Hour or Shift button | Choose a Business Resource for Hour. Plant-wide production metrics support Shift, but not Hour. Safety and external defects are day-based and do not offer Shift. |
| No Drivers button | OEE uses its factors page at Level 4. Kaizen has a plant-wide trend without a driver breakdown. Other unavailable levels state their prerequisite. |
The trend draws the target that governed each bucket, including effective-date changes. Read the explanation when a target cannot be drawn: a live per-event downtime limit is not a daily downtime budget, and an incompatible cadence is not converted. PPM displays convert the percentage target alongside the metric.
At Level 3, choose a second supported dimension for a stacked chart and intersection table, such as asset by shift. Each intersection retains its own numerator and denominator. Use this to distinguish a problem confined to one shift from one shared across an asset.
Open the OEE factors page from the breakdown to compare Availability, Performance and Quality for the bucket and their trends. Availability leads to downtime drivers, Quality to scrap drivers, and Performance to the product breakdown. This decomposition does not include a Six Big Losses waterfall.
On an A-FMDS problem, use the Standardised in document card to select an approved controlled instruction containing the fix. Resolving or closing a problem, or completing a corrective or preventive action, also offers this optional step. Level 5 shows the document beside the countermeasures. Public displays show its title without exposing the document download.
All five levels open from the public board without signing in, so anyone at the monitor can touch a tile and follow the path to the countermeasures. The Gaps and problems button opens the read-only category page rather than the working one, and the owner column on the countermeasures table is absent unless the installation has chosen to publish names on public screens.
A Gap is the system saying "an actual value missed its target." Most gaps are raised automatically by ProAlert's detectors: a scrap-rate breach, a stretch of unplanned downtime, an EdgeSense sensor alarm, an OEE drop, a missed schedule, a failed inspection, a skipped leader routine, a preventive maintenance task more than a day past due. They appear on the board on their own; you do not create them by hand, though you can raise one directly for something no detector would catch. The full list of sources is in the System Overview.
A gap on its own has no owner. To make it someone's responsibility you promote it to a Problem.
Critical gaps can be configured to promote themselves: the problem is created the instant the gap is detected, so the most serious issues never wait for a human to click. One problem can also gather several recurring gaps of the same kind, rather than spawning a new problem each time.
Some gaps are tied to a period rather than to a condition. A daily leader walk that nobody performed cannot be made up tomorrow, because tomorrow has its own walk. Left alone, one skipped routine per day would sit on the board for ever, and after a few months the count would say more about how long the system had been running than about the state of the plant.
So when the period a gap measured has passed, A-FMDS closes it as Closed, Not Completed. This is deliberately its own status, separate from the two it could be confused with:
| Status | What it means |
|---|---|
| Closed | The work was done and finished. |
| Ignored with reason | A person looked at it and decided to dismiss it, and recorded why. |
| Closed, Not Completed | The period ended with the work outstanding. Nobody dismissed it and nobody did it. |
Counting a skipped walk as "closed" would flatter your compliance figures; counting it as "ignored" would blame someone who never saw it. Keeping it as its own status means leader-standard-work compliance can be measured honestly, so you can ask how often routines are actually performed and get a true answer. Only routines with a fixed cadence close this way; on-demand and per-shift routines have no next period, so their gaps stay open for a person to deal with.
The shift schedule and the floor are compared continuously. When they disagree for more than half an hour, the mismatch is raised as a gap on the board rather than living only on the schedule screen:
The half-hour delay is deliberate: a machine a few minutes late to start is a blip, not a management problem. And when the mismatch clears, the line starts or the device reconnects, the gap resolves itself.
The Problems screen is the heart of A-FMDS. A Problem is an operational issue that someone is accountable for resolving.
AFMDS → Problems lists every problem, newest first, with a Status filter at the top. Choose a single lifecycle status, or leave it on "All statuses." Click any problem number to open its detail page.
Not every problem starts as a detected gap: sometimes a person simply spots one. Use the New Problem button on the Problems screen:
The detail page has six areas:
Follow-Up Sheet in the problem header downloads that one problem as a one-page PDF: the same facts the screen shows, laid out as a working document to carry into a tiered meeting and write on. It is the single-problem counterpart to exporting the problems list, and it is a different document from the Root Cause Analysis sheet. That one is the investigation record, this one is the page you stand at the board with.
Click Edit on the header to change the title, classification, due date, description, or the Owner (chosen from a list of users). Saving an owner change formally assigns the problem: an unassigned Open problem moves to Assigned automatically.
A problem follows a controlled lifecycle. A-FMDS only allows valid moves, so a problem can never skip a step. The lifecycle buttons on the detail page show only the moves that are valid right now.
| Status | Meaning |
|---|---|
| Open | Raised, not yet assigned. |
| Assigned | Has an owner; work not yet started. |
| InProgress | The owner is actively working it. |
| AwaitingSupport | Waiting on another person or team. |
| Blocked | Stopped by an obstacle. |
| Resolved | The owner believes it is fixed. |
| Verified | An independent person confirmed the fix. |
| Closed | Done. The terminal state. |
| Reopened | Came back: closed too soon, or recurred. |
Resolved means "I think it's fixed." Verified means "someone else checked." Use Request Verification to resolve a problem and flag that it must be independently verified. Once flagged, it cannot be closed until someone clicks Verify. This is the ISO 9001 (International Organization for Standardization) corrective-action discipline built into the workflow.
If two problems are really the same issue, open the duplicate and click Merge. Enter the number of the problem to keep; the duplicate's linked gaps move onto the surviving problem and the duplicate is closed. Merges are recorded in both problems' timelines.
Every problem keeps a permanent Timeline: created, assigned, edited, status changes, verification, merges. Each entry is stamped with who did it and when. Expand Payload on any entry to see exactly what changed. The timeline is append-only: it is the audit trail and it cannot be edited.
A Problem says "this is mine to fix." The Actions are the concrete, individually-tracked steps that actually fix it. A problem usually needs several: contain it now, find the cause, apply the permanent fix, verify it held. Each has its own owner and due date.
On a problem's detail page, in the Actions panel, click + Add Action. Give it a title, choose an Action Type, a priority, an owner and a due date. Tick Verification Required if the action must be independently checked before it counts as done.
| Investigate | Dig into what is happening. |
| Containment | Stop the bleeding now, a temporary measure. |
| Correction | Fix this specific occurrence. |
| Corrective | Fix the root cause so it does not recur. |
| Preventive | Stop it happening elsewhere. |
| Verification | Confirm a fix actually worked. |
| Coaching / Followup / Custom | People follow-ups and anything else. |
Each action row shows the buttons valid for its current status:
NotStarted InProgress Complete Verified
(with Blocked and Cancelled as optional side exits)
| Button | What it does |
|---|---|
| Start | Begin work: NotStarted → InProgress. |
| Block / Resume | Flag an obstacle, or clear it. |
| Complete | Mark the work done (with optional notes). |
| Verify | Independently confirm a completed action. |
| Assign | Set or change the owner. |
| Cancel | Abandon an action that is no longer needed. |
AFMDS → Actions shows every action across all problems, with a status filter. It answers "what is outstanding everywhere?" and each row links back to its parent problem. Use the per-problem panel to work an action; use this list to survey them.
Escalation is A-FMDS's safety net: if a problem or action sits too long, the system raises it up the chain automatically. Nobody has to remember to chase it.
The AFMDS → Escalations screen has two parts:
Every escalation the engine has fired: which rule, which target, which level, and when. New ones appear live.
The active rules and their level ladders: "after N minutes, notify X." Add, edit, switch off or delete a rule from here.
A rule is a ladder of levels. Each level says "once the target has been open this many minutes, notify this person or role." A level can repeat. The background engine runs once a minute and fires any level whose time has come, once each, so you are never spammed.
A rule's Applies to setting decides what it chases. Every one of these is evaluated:
| Applies to | What gets chased |
|---|---|
| Gap | A detected gap left sitting without being promoted or dealt with. |
| Problem | An open problem past its due date, or simply open too long. |
| Action | An action nobody has completed. |
| Work Order | A corrective or preventive work order running late. |
| Schedule Discrepancy | The floor not matching what the schedule says should be running. |
| LSW Execution | A leader routine still outstanding past its grace window. Once that routine is written off as missed it raises a gap instead, so one lapse never chases the same person twice. |
| Coaching Requirement | A coaching note that is owed and has not been given. |
| Tiered Meeting Action | Something committed to in front of the board. These are ordinary actions as well, so this target exists to let a plant chase what was promised at the meeting on its own cadence without escalating every action in the plant. |
Example ladder for a critical problem:
When you have seen an escalation and taken it on, click Acknowledge on its row. The badge turns Acknowledged and records who acknowledged it. Acknowledging is the signal that a human now owns the escalation. It does not close the underlying problem; you still resolve that through its own lifecycle.
An escalation firing is not a reprimand. It is the system protecting the problem from being forgotten. The fastest way to stop escalations is to keep problems moving through their lifecycle.
Leader Standard Work routines can name their governing controlled documents. Attach them on the saved routine editor; the execution page shows those documents read-only. Tiered meetings have their own governing-document card.
A-FMDS does not enforce job titles: everyone can see everything. It works best when each role uses it for a clear purpose.
| Role | How they use A-FMDS |
|---|---|
| Operator / Line | Spots an issue the detectors miss and raises it with New Problem. Works the actions assigned to them. |
| Supervisor / Lead | Owns the daily board. Promotes gaps, takes ownership of problems, creates and assigns actions, drives the lifecycle, acknowledges escalations. |
| Maintenance / Engineer | Carries the corrective and preventive actions; completes them and records what was done. |
| Manager | Verifies resolved problems, watches the Escalations screen, uses the Problems list to see what is open and ageing. |
A routine on AFMDS → LSW Routines carries a cadence, and the cadence decides when the system creates the next one for you to perform:
| Cadence | When it appears |
|---|---|
| Per Shift | Once per shift, taking its start from your shift schedule, so the routine editor does not ask you for a time of day, and a site that changes its shift pattern does not have to restate it here. |
| Daily / Weekly / Monthly | Once per period, from the start time of day you set on the routine. |
| On Demand | Never generated. You start it yourself. The name is the promise, and it is kept. |
A Daily, Weekly or Monthly routine that was never performed is closed off when its period ends, because yesterday's routine cannot be performed today and leaving it open would build a backlog nobody can ever clear. A Per Shift or On Demand routine has no next period to roll into, so its gap stays open for a person to deal with rather than being written off by the calendar.
Everything described in this guide is available to you today: the SQDCP board and its TV display, the SQDCP grid by day, click-to-drill cells, the metric-specific drill-down behind the board tiles, the printable follow-up sheet, coaching notes, Leader Standard Work, Tiered Meetings, the knowledge base of past resolutions and mobile push. The rollout phases in the System Overview show where each of those landed.
A-FMDS continues to develop. This is what is not in your hands yet, described the way you would notice it on the floor:
| Capability | What you will be able to do |
|---|---|
| Targets scored per shift | Set a target that is judged over a single shift, or live as the shift runs, and have gaps raised against that window. Today a target is scored over a day, a week, a month, a quarter or a year; the shift and real-time choices appear in the target editor greyed out, so you can see them coming without setting one that would quietly be scored over the wrong window. |
| Safety narrowed to a line | Filter the safety trend to one Business Resource the way OEE, scrap and downtime already filter. Today the trend reads a plant-wide roll-up and says so; one level down, the breakdown already splits incidents by asset and by location. |
A-FMDS is built against real use. If a screen does not match how your floor actually runs its day, that feedback is wanted. It is far cheaper to adjust now than after the fact.
Check which of the two numbers you are reading. The large figure is activity in the selected window; the quieter "still open" figure is the outstanding backlog across all time.
A large backlog usually means routines or gaps are being raised but never acted on. Open the SQDCP cell to see what they are. If they are Leader Standard Work misses, the routines are being scheduled but not performed, and that is a real finding, not a display fault.
Look at the label beside the dash. "No production scheduled" means nothing was planned to run, and blank is the correct answer. "Scheduled, no cycles recorded" means production was expected and nothing was captured, which is worth investigating.
Remember these figures come from a nightly roll-up, so a work centre run today will not appear until tomorrow.
Gaps are matched to a line through the asset they were raised on. A gap with no asset, for example a Leader Standard Work routine that has no asset or business resource set in its configuration, cannot be attributed to a line, so it only appears on the plant-wide board. If a routine should show on a line board, set its business resource in the routine configuration.
Two things close on their own. A schedule adherence gap resolves when the mismatch clears, when the line started or the capture device reconnected.
A period-bound routine gap closes as Closed, Not Completed once its period ends. That is not the same as "resolved": it is a permanent record that the work was never done, and it is what compliance reporting counts. The gap's history shows exactly when and why it closed.
The drill-down pages 50 rows per list, with a pager at the bottom and the full total shown in each panel header. If you are routinely paging through hundreds of rows in one category, that is worth raising, because it usually means a detector is producing items nobody is acting on.
The TV board shows a connection indicator beside the clock. When it reads connected, gaps and problems arrive live; the board also refreshes itself on a timer as a safety net, and reloads periodically so a display left running for days stays healthy. If the indicator shows a lost connection, the display has lost its network path to the server rather than the floor having gone quiet.
That is OEE with no Business Resource selected. A month's OEE is not the average of its days; it needs the production time behind each day, and the plant-wide roll-up does not keep it. Pick a line on the board and open the tile again: the longer ranges come back, computed from that line's own production time. The note above the chart says the same thing.
PPM is parts per million, so it is only offered when everything produced in the window is counted in units. A month in which a product measured by length or weight, or a product with no unit of measure at all, was run falls back to a percentage for that window, and the note above the chart tells you so. The underlying number has not changed; a PPM figure is the percentage times ten thousand.
On the safety trend a single-day point opens that day's breakdown by asset or location. A point covering a week, month, quarter or year does not open at that range: the incident roll-up has no one day to show, and opening the first day of a month under a month's label would show one day of data under a month's name. Switch the range to Day and the points open.