RPD Vaktari Machine

01What it is and what it does

RPD Vaktari Machine reads the files the machine's SCADA writes during the day and gets three things out of them: what state the machine is in now, how long it has been running and stopped, and why it stopped. Nothing has to be written down by hand: every change of state and every error is recorded with its time.

You look at it with a browser, from the same computer or from any other on the network. It has five tabs:

The states of the machine

The whole module is built on splitting time into states. Three of them count as production time, and three do not:

02Starting and access

On the computer where it is installed

• Double-click start_vaktari_machine.bat, or the shortcut that points to it. A black window with text opens: it is the program that reads the files. Do not close it; you can minimise it.
• Open the browser and go to http://localhost:8761.

From another computer on the network

• Type http://PC-IP:8761 in the browser, where PC-IP is the address of the computer running Vaktari. Whoever did the installation will give it to you.
• It is worth saving as a bookmark: it is the only thing you need to remember.

No need to refresh

The Live tab updates by itself every 5 seconds, and the state clock ticks every second. Accumulated is recalculated every minute when the period includes today.

Each installation of Vaktari Machine follows one machine. If there is more than one, each has its own address.

03The header

1

Brand and module.

2

The machine and, below it, the reference it is making according to the last reading.

3

Help Opens this manual in a new tab, on the RPD website and in the language of the page.

4

Language Catalan, Spanish, English, Basque, French and Portuguese. The browser remembers the choice.

5

Theme Dark or Light. Also remembered. The state colours are the same in both.

6

Tabs. The browser remembers the last one you had open.

Error texts are not translated: they are shown as the machine's SCADA writes them.

04“Live” tab

This is the screen to keep open on the shop floor. It always refers to today.

4.1 · The state

The band at the top says in large letters what the machine is doing and changes colour with the state: green when running, red when stopped by an error, amber when stopped without an error, and grey when it is not production time. On the right, a clock counts how long it has been in that state.

When the stop is due to an error, the list of active errors appears under the state. The first one carries the first error label: it is the one that stopped the machine, and the others are usually a consequence of it.

4.2 · The counters of the day

If the machine provides general counters, a seventh card appears, General counters, with OK, NOK and NV (parts not seen). And if the machine does not classify the parts — it counts the total but not OK or NOK — the NOK lines are not shown.

4.3 · Day timeline

1

The black line is now. To its left, what has already happened.

2

Legend. Only the grey states the day actually has are shown.

Each coloured stretch is a state. Hover over it to see the state, the start and end times, the duration and, if any, the errors. A day with activity looks like fig. 12.

4.4 · Time breakdown

A bar with the three parts of production time — running, stop with error, stop without error — and a table with, for each one, how many stops, how much time and what percentage. The last row, Availability, is the total number of stops, the production time and the percentage running.

4.5 · Cumulative output

The curve of parts made through the day, with the stops marked in the background in their colour. While the machine is running the line rises; at each stop it stays flat. The slope is the rate: the steeper, the faster. It is the same chart as in fig. 12.

4.6 · Errors and stops of the day

Below are the two error tables — Errors by frequency and Errors by downtime — and the Today's stops list, with the most recent at the top; the one still going on says Ongoing. They are the same tables as in “Accumulated”, where they are explained (fig. 14 and 15), with today's data.

05“Accumulated” tab

Here you choose a day or a period and see how it went. The selection you make is the one used by Report and Export.

5.1 · The selection

1

From and To The period. With the same date in both, a single day is shown.

2

Reference All, or one of those made in the period. With a reference chosen, NOK cannot be given, because the machine counts it per lot and not per reference.

3

Shortcuts. Last day is the last day with data; 7 days and 30 days end on that same day; All goes back to the first file.

5.2 · The counters of the period

Output with the NOK, Availability with the running hours, Stops with and without an error, Downtime and Average rate while running: the parts made divided by the running hours. Under the last one, how many days of the period have data.

5.3 · A period: day by day

1

Timeline / Breakdown The two ways of drawing each day. The browser remembers the one you choose.

2

The hour axis is the same for every row: it runs from the earliest start to the latest finish in the period. So you can see at a glance which day started later or finished earlier.

3

Pause between lots, in light yellow: a stop longer than the limit set at installation. It is usually a wait for material between one lot and the next. It does not count for availability.

4

No activity The day has a file but the machine was not started.

In Timeline, each day is drawn as it happened, from left to right, with the same colours as the timeline of a single day. You can see the lots that were made, where the long stops were and whether the machine was left stopped with an error. Short stops have a minimum width so they do not vanish: no error is hidden, however short. Hovering over a stretch shows the state and the times.

In Breakdown, each day's bar groups the time by state: running, stop with error, stop without error and pause between lots. It is useful for comparing proportions, but it does not say when each thing happened. On the right, in both views, the day's availability and output.

5.4 · A single day: timeline and output

1

Timeline. Each stretch is a state, in its colour. Hover over it to see the times, the duration and the errors. Here the day starts with a long breakdown, in red, and around midday there is a stop without an error of almost an hour.

2

Cumulative output. The same story in parts: the line is flat during the stops, which are marked in the background, and rises while the machine is running.

3

Legend. The states that do not count for availability (grey and light yellow) only appear if the day has them.

5.5 · Errors

1

Errors by frequency How many times each error came up. If a stop has three errors, all three count.

2

Errors by downtime How much time each error cost. The whole time of a stop is assigned to its first error, which is the one that caused it.

The two lists rarely agree, and both are useful. An error that comes up often but is short annoys the operator; one that comes up once and keeps the machine stopped for an hour is the one that brings availability down. In the example, the most frequent error came up twice, but 96 % of the time lost belongs to another that came up only once.

The Cumulative column is the percentage added up to that row. The top ten are shown and the rest are grouped under Others.

5.6 · Stops

1

Show Filters the list: all stops, only those with an error or only those without. The number next to the title is how many there are.

For each stop: start, end, duration, type, the reference being made and the errors, with the first at the top. In a period of more than one day there is also a date column. If there are more than a hundred, the last hundred are shown with a button to see them all.

Only stops that count towards availability are listed: the wait to start, the end of the day and the pauses are not in the list.

06“Errors” tab

This is where you see how each error evolves over the days. It is the maintenance screen: an error that starts to appear more often usually points to a dirty sensor, a part wearing out or an adjustment drifting, and it is better to see it before it turns into a breakdown.

At the very top, Vaktari already says what it finds unusual. Below are the chart and the table to look at it in detail and, at the end, the e-mail alerts, so you do not need the screen open to find out.

6.1 · What to watch

Vaktari compares the errors of the last 3 production days with those of the previous 20, and warns when an error appears more often than chance can explain. Everything is measured per hour of running time, so a short day and a long one compare fairly, and days without production do not count.

1

Sudden increase In the last few days the error has appeared far more often than usual. The big figure (×8.8 in the example) is how many times the usual rate. It may be a fault starting, a lost adjustment or a bad batch of material.

2

Rising trend The error has not jumped, but it has been rising little by little for days; the card says since when. This is the typical sign of wear: a gripper, a guide, a sensor getting dirty. It is the one most worth dealing with before the machine stops.

3

New error It had not appeared in any of the reference days and now it repeats. Something has changed in the machine.

4

Chronic, unchanged Errors that appear almost every day, always the same. They are not an alarm: Vaktari keeps them apart so they do not hide the alerts that really are new. If they are a nuisance, they have to be solved separately.

5

Muted Errors with an alert that have been muted for e-mail (see 6.4). They are not shown as a card, but they stay here, in grey, so they are not lost from sight.

Clicking a card puts that error on the chart, scrolls down to it and highlights its line. If nothing is out of the ordinary, the panel says No error is occurring more often than usual and how many days it has analysed. The alerts are recalculated every ten minutes and always look at the latest days with data, whatever period is chosen for the chart.

6.2 · Error trend

1

The period. From and To, or the shortcuts 7 days, 30 days, 90 days and All, which end on the last day with data. When the tab opens it shows the last 30 days. This period is independent of the Accumulated one.

2

Days without production, shaded. The line passes over them, from one production day to the next, so the trend reads continuously and a weekend does not look like a day without errors. The last point of each line is always marked, slightly larger.

3

A day's detail. Hovering over a day shows every error on the chart with its count for that day, highest first.

4

Legend. The errors on the chart, each with its colour.

The chart draws up to eight errors at a time. By default, the eight most frequent in the period. With so many lines it gets busy: hovering over an error in the legend or in the table below highlights its line and fades the others.

6.3 · Errors in the period

1

The tick boxes decide which errors go on the chart, up to eight. With eight ticked, untick one to tick another. Each error keeps its colour while it is ticked.

2

Alert How this error is reported by e-mail. Each click moves it to the next state: normal, critical (red !) and muted (crossed-out bell). Section 6.4 explains it.

3

Times and Days with the error How many times it appeared in the period and on how many different days.

4

Per day One bar per day. A day without production stays empty; a production day without the error shows a grey line at the bottom. Hovering over a bar shows the day and the count.

5

Own scale / Common scale With own scale, each row is drawn on its own maximum, which shows the shape of each error's trend well. With common scale, every row is drawn on the same maximum, that of the error's busiest day, and heights can be compared between errors. The browser remembers the choice.

Here every time an error appears is counted, even if the stop ended up as a pause between lots or the end of the day: for maintenance what matters is how many times it failed, not how the stop was counted. That is why the totals can be somewhat higher than in Errors by frequency on Accumulated.

6.4 · E-mail alerts

Vaktari Machine can send e-mails about the errors, so you do not need the screen open to know what is happening on the machine. Which addresses they go to and which server sends them is set by whoever installs it; the page decides which errors raise alerts.

1

Recipients Who receives the e-mails.

2

What is sent The three kinds of e-mail, with the time of the daily summary and the minutes over which critical errors are grouped.

3

Critical errors Those that send an e-mail every time they occur. The × sets it back to normal.

4

Muted errors Those that never send any e-mail. The × unmutes it.

5

Last e-mail sent When the last one went out. If a send fails, the error is shown next to it in red, and it is retried after 5 minutes.

6

Send a test e-mail To check that the e-mails arrive. The result is shown next to the button.

Each error in the table can be in one of these three states, changed with the button in the Alert column:

Muting an error only takes it out of the e-mails: on the page it still counts the same, in the charts, the table and the Paretos.

There are three kinds of e-mail:

• Daily summary, every morning at the set time (usually 06:00): the previous day's output, availability and stops; the day's errors, with times and downtime, critical ones in red; the current What to watch; the chart of the last 30 days with up to five errors, first those with an alert and then the most frequent; and, at the very end, the muted errors with their count for the day, in case one needs unmuting. Not sent if there was no production the day before.
• Errors with an unusual frequency, when a new alert appears in What to watch. The same error is not sent again while the alert stays active, nor if it was already in the morning summary.
• Critical error, every time a critical error stops the machine. Those within a few minutes (usually 5) go together in one e-mail, with the time of each and how many times it has occurred today.

07“Report” tab

The period chosen in Accumulated, on a single sheet to print or save.

The summary carries the output, the NOK, the running time, the time and number of stops with and without an error, the availability and the average rate while running.

• If the period is a single day, there is the day timeline.
• If it is several days, instead of the timeline there is the By day table: minutes running, minutes stopped with and without an error, stops, availability and output for each day.
• The errors, by frequency and by downtime. Up to fifty are shown here, not just ten.
• The stops, all of them.

08“Export” tab

1

What will be exported. Always the Accumulated period. To change it, go to that tab.

2

Stops One row per stop: date, start, end, duration in seconds, type, reference and errors.

3

Errors One row per error, with the duration of the stop and whether it was the first. It is the file for building your own Pareto.

4

Daily summary One row per day: minutes running and stopped, stops, availability, output and NOK.

5

PDF report Create PDF jumps to the report and opens the print dialog. Choose “Save as PDF”.

The CSV files use a semicolon as separator. Decimals follow the interface language: a point in English, and a comma in Catalan, Spanish, Basque, French and Portuguese. The file name includes the machine, the period and the reference.

09The footer

• The machine and the last data: the time of the last reading received from the SCADA. If it is old while the machine is working, something is not getting through.
• The version, on the last line, in bold. It is the first thing you will be asked for if you have to report an issue.
• If server v… appears next to it, the program and the page are not the same version. Reload the page with Ctrl + F5; if it is still there, tell whoever did the installation.

10How to read the numbers

Availability

Running time divided by running time plus stop time. 90 % means that, out of every hour in which production was due, six minutes were stops. The time before starting the day, the time after finishing and long pauses, if configured, are left out of the calculation.

Availability says nothing about speed or quality: a machine can be running 100 % of the time and be slow, or make bad parts. That is what the rate and the NOK are for.

Rate

• The one on Live is that of the last few minutes and is only shown with the machine running. It tells how fast it is going now.
• The average is the parts of the day, or of the period, divided by the running hours. Stops do not lower it: it is the pace of the machine when it works.

Output

Vaktari adds up how much the machine's lot counter advances through the day. The first reading of the day is the starting point and does not count, because the lot may come from the day before. If the counter goes down, it takes the lot to have changed and carries on adding from zero. A brief zero caused by a SCADA restart is ignored, so that parts are not counted twice.

Errors: frequency and time

A stop can have several errors. By frequency, each error counts once. By time, the whole stop goes to the first error. That is why the counts do not add up to the number of stops, whereas the durations do add up to the stop time with an error.

Errors over time

An error is not more serious because it appears a lot on a single day: it may be a bad batch of material. What warns of a fault is the trend: an error that appears more day after day, or a new one that repeats. The What to watch panel in the Errors tab works it out per hour of running time, so a day with more output does not look worse; to look at it yourself, use a period of 30 or 90 days.

Stop without error

It is a stop for which the SCADA recorded no error. Vaktari does not know the reason. If they weigh heavily on a day, it is worth asking what is being done during those stops: it is lost time that no alarm explains.

NOK

• With a reference chosen it is not given, because the machine counts NOK per lot.
• If the machine does not classify the parts, neither: the total rises but OK and NOK stay at zero, and showing 0 % NOK would be misleading.

11Common problems

If you have to report an issue, note the version shown in the page footer and take a screenshot of the whole screen.