Skip to main content

Fillers. Vessels. Lab robots. Controllers. Seal Automation. Your plant, driven from the record.

Seal Automation sends each step's set points to your equipment once the step is signed, and keeps what the machine took on the lot.

Step 4.1 of lot 26-0921's batch record, Start of fill, at 00:21 on Monday 16 November. Its three set points, fill program FP-24 v3, a fill target of 10.20 g and a line speed of 75 vials a minute, are each inside the recipe's window. A. Okafor signed it at 00:17 and J. Lindqvist verified it at 00:18; Seal Automation then sent the three to filler F-02 from EDGE-L2, and F-02 read all three back a second later. A. Okafor started F-02 at its own panel at 00:20, and Seal Edge reads it back running at 75 vials a minute
Automation › Line 2 › Lot 26-0921
F-02 running
Line 2, F-02Mon 16 Nov, 00:21:00

4.1 Start of fill

Set pointSentWindow
Fill programFP-24 v3FP-24 v3 only
Fill target10.20 g10.10 to 10.30 g
Line speed75 a minute70 to 80 a minute
75a minuteRead by Edge at 00:21:00

Lot 26-0921

Step 4.1 released
Step 4.1's set points are MB-⁠24's, each inside the window the recipe gives itSigned by the operator and verified by a second person: nothing is sent before bothSent from EDGE-⁠L2 at 00:18:04; F-⁠02 read back all three a second laterStarted at F-⁠02's own panel. Edge reads the line back onto the same stepWrites only what a signed step releasesThe machine's own safety stays the machine'sA. Okafor, trained on SOP-⁠FIL-⁠007 v13. Password and authenticator code, 00:17:42J. Lindqvist checked the three set points against MB-⁠24 and signed at 00:18:03From EDGE-⁠L2 in Line 2's cabinet, to F-⁠02's controller, at 00:18:04FP-⁠24 v3, 10.20 g for each of four pumps and 75 a minute, read back at 00:18:05F-⁠02 starts only from its own panel, with its own interlocks, as it always hasLine speed, read by Seal Edge from F-⁠02 every second, read-⁠only

Nothing reaches a machine unsigned.

Today a step's set points are typed in at the panel, and the record holds what someone says the machine took.

Compared onTodaySeal Automation
A step's set pointsTyped in at the panel from the recordSent from the step once it is signed
The second personWatches the typingVerifies the step before anything is sent
A value typed wrongFound at the next check, or not at allRefused before it reaches the machine
What the machine tookThe operator's word for itIts read-back, an entry on the lot
A plate in the labIts map keyed into the handlerLaid out by the method, results on the samples

Every write is an entry.

Each value sent to a machine, when, on whose release and what the machine read back, kept on the lot and its step.

Every write lot 26-0921's batch record made to a machine, at 04:00 on Monday 16 November. On Sunday at 12:30 step 2.4 sent vessel V-201 an agitation of 120 rpm and a mixing time of 35 minutes, released by B. Adeyemi and L. Ferreira. At 00:18 step 4.1 sent filler F-02 fill program FP-24 v3, a fill target of 10.20 g and a line speed of 75 a minute, released by A. Okafor and J. Lindqvist. At 03:52 a pump 3 target of 10.42 g, signed by D. Osei, was refused as outside 10.10 to 10.30 g and never sent; at 03:54 10.24 g was released by D. Osei and A. Okafor and read back by F-02. Every write sent was read back within a second
Automation › Lot 26-0921 › Writes
6 sent, 1 refused
Mon 16 Nov, 04:00
Step 2.4 sent V-⁠201 its speed and its time on Sunday; the vessel read both backStep 4.1 loaded F-⁠02's program and set points in one send as the night shift signed10.42 g was outside step 4.2's window, so it was refused on the record and never sent10.24 g was signed again, sent and read back by F-⁠02 within a secondWrites only what a signed step releasesThe machine's own safety stays the machine'sSun 15 Nov, 12:30:14. V-⁠201's controller read back 120 rpm. Window 100 to 140 rpmSun 15 Nov, 12:30:14. V-⁠201's controller read back 35 min. Window 30 to 45 minMon 16 Nov, 00:18:04. F-⁠02 loaded FP-⁠24 v3 and read back its name and versionMon 16 Nov, 00:18:04. F-⁠02 read back 10.20 g for each of its four pumpsMon 16 Nov, 00:18:04. F-⁠02 read back 75 a minute. Window 70 to 80 a minuteMon 16 Nov, 03:52:40. Signed by D. Osei and refused, outside 10.10 to 10.30 g: never verified, never sentMon 16 Nov, 03:54:12. F-⁠02 read back 10.24 g for pump 3; pumps 1, 2 and 4 kept 10.20 g

Outside its window, it never leaves.

A set point outside the recipe's window is refused on the record before it reaches the machine, and Neil drafts the deviation.

Step 4.2 of lot 26-0921 at 03:58 on Monday 16 November. At the 03:50 fill weight check pump 3's mean was 10.16 g. At 03:52 D. Osei signed a pump 3 target of 10.42 g; it was outside the window of 10.10 to 10.30 g, so the record refused it and nothing was sent to F-02, which kept 10.20 g. At 03:54 he entered 10.24 g, A. Okafor verified it, and F-02 read it back. Beside it Neil's draft of a deviation from the refusal: what happened, no effect found on the lot, the cause, and a proposal of minor, for D. Osei to accept before it goes to QA
Automation › Lot 26-0921 › Step 4.2
1 refused
Line 2, F-02Mon 16 Nov, 03:58

4.2 Fill weight check

  • Pump 110.21 g
  • Pump 210.19 g
  • Pump 310.16 g
  • Pump 410.20 g
03:52Pump 3 target 10.42 gRefused, not sentOutside 10.10 to 10.30 g
03:54Pump 3 target 10.24 gRead back by F-02D. Osei, A. Okafor

Deviation, draft

For D. Osei
Accept the draft
Pump 3's mean at the 03:50 check: 10.16 g, low but inside the lot's limits10.42 g was outside 10.10 to 10.30 g: refused as D. Osei signed it, and never sent10.24 g, verified by A. Okafor, sent at 03:54:12 and read back by F-⁠02Neil drafted the deviation from the refusal. It goes nowhere until D. Osei accepts itWrites only what a signed step releasesThe machine's own safety stays the machine'sThe refusal, an entry on step 4.2 at 03:52:40, signed by D. OseiF-⁠02's read-⁠backs at 00:18:05 and 03:54:13, and Edge's fill weightsD. Osei's 03:54 entry on step 4.2, verified by A. OkaforNeil proposes; S. Amato in QA classifies once D. Osei accepts

The lab, run the same way.

The method lays out the plate, the liquid handler takes it, and the reader's results come back onto the samples.

Lot 26-0918's binding ELISA by method QCM-038 v4 in the QC lab, at 15:40 on Monday 16 November. The method lays out a 96-well plate: the reference standard in columns 1 and 2, the control in 3 and 4, three preparations of the lot in 5 to 10 and blanks in 11 and 12. N. Byrne signed the run at 09:42; Seal Automation sent the map to liquid handler LH-01, which read it back, and she started it at 09:44. Plate reader PL-02's read came back through Seal Edge at 15:31, and each well's result is on its sample: binding 101%, inside 70 to 130%, for her to review
Automation › QC lab › R-1031-01
Read, for review
LH-01 and PL-02Mon 16 Nov, 15:40

Binding, ELISA

  • Standard
  • Control
  • Lot 26-0918, 3 preparations
  • Blank
101%Lot 26-0918, mean of 3. Limits 70 to 130%

R-1031-01

QCM-038 v4
QCM-⁠038 v4 lays out the plate: standard, control, three preparations of the lot, blanksN. Byrne signed the run and LH-⁠01 read back all 96 wells before she started itPL-⁠02's read came back through Edge, each well onto its own sampleBinding 101%, inside 70 to 130%, for N. Byrne to reviewWrites only what a signed step releasesThe machine's own safety stays the machine'sQCM-⁠038 v4, effective, with the three samples registered for lot 26-⁠0918N. Byrne, trained on QCM-⁠038 v4. Password and authenticator codeFrom EDGE-⁠QC1 in the QC lab, through LH-⁠01's own interfaceLH-⁠01 read back the map well by well; it matched all 96LH-⁠01 starts from its own screen, with its own door and deck checksPL-⁠02's file, collected by Seal Edge on EDGE-⁠QC1, read-⁠onlyEach well's reading on its sample; the curve and its fit kept with them

Questions about Seal Automation.

Can we buy Seal Automation today?

Not yet: it is a proposal, and nothing on this page is shipped. Seal Edge, which reads your equipment from the same computer, is installed at customers today.

What can it send to a machine?

Only what a signed step releases: the step's program and set points, each inside the window the recipe gives it. It never starts or stops a machine. Your people do that at the machine's own panel.

Does it replace the machine's own safety system?

No. Interlocks, alarms and emergency stops stay the machine's own and act as they do today. Its controller still checks every value it is sent, as it checks one typed at its panel.

Does it replace our DCS or SCADA?

No. Your control system keeps running the plant. Automation sends a step's set points to it, or to a machine's own controller, through the interface each already offers.

Can Neil send a value to a machine?

No. Neil drafts and checks: the deviation from a refusal, the review of a run. Only a step your people sign releases a write.

What happens when the network drops?

A machine keeps what it was sent and runs as it would from its own panel. Nothing new is sent until the link is back, and what was sent stays on the record. Your people can always run a machine from its panel, recorded as they are today.

What does IT have to open?

Nothing inbound. It runs on the Seal Edge computer on your network and calls out over HTTPS. Each machine it may write to is granted by name, apart from what Edge may read.

Who approves a recipe's windows?

Your QA. A window, and each machine's connection, is a change in Seal: tested, with its evidence, and in effect only once QA approves it.

Which equipment can it drive?

Equipment whose controller takes set points through an interface it already has, and lab robots through their standard interface. Each is connected and tested on your site before a step may write to it.

See your process in Seal.

Working 48 hours after we agree its scope: tested, with evidence your QA approves.