Packaging Hall

A soft drink packaging hall with three identical 500 mL PET lines at 400 bottles a minute, all running from one syrup room, one compressor set and one CIP skid. 18 devices and 293 tags over Modbus TCP.

A single line cannot show what goes wrong when three lines share one set of utilities. Each line here is filler, capper, labeler, checkweigher and case packer, and every line runs the same SKU at the same nominal rate. The three shared assets are their own ISA-95 area, separate from the lines:

Three of the five scenarios start in a shared asset, so they arrive on every line at once and cannot be traced from any single line's data. The other two are confined to one line, with two healthy lines beside it for comparison.

Shared-utility contention. The product header pressure depends on how many lines are drawing, and no line publishes it. The fillers need 2.60 bar. With a worn pump impeller the header holds 2.80 bar at two lines and falls to about 2.18 bar at three, so the fault costs nothing until the third line starts. In six minutes that is 166 rejected bottles against 17 on a healthy pump. Each filler then lengthens its fill time to hold its mean, and when the sag clears, or a line is simply stopped, that trim is left standing: giveaway rises on every line, including the two that were never affected, a minute after the cause has gone.

Loose caps on every line at once. The compressed air plant carries three lines and the CIP skid with a small margin. Trip one compressor and two lines still fit. Run three and the header falls below 6.20 barg, every pneumatic capping head delivers under its setpoint, and 37 percent of caps go out below the 1.35 Nm seal minimum on all three lines, with 7 rejects. A loose cap weighs the same as a good one, so it passes the checkweigher. The only tag that reports it is the capper's LowTorqueCount, and the cause is on a different device.

CIP scheduling and verification. One skid serves three lines, so a clean is a resource a line waits for, and a wait past the clean window is a compliance event rather than a rate loss. When the caustic wash runs cold, the cycle completes normally and fails verification, and the line rejoins the queue behind whoever is already waiting. The overdue time grows with each failed cycle until the line misses its window, without any schedule causing it.

Which line to walk to first. With one SKU and one nominal rate, the difference between the three lines is the plant rather than the product. The operator view ranks them worst first by OEE and names the dominant loss. Labeler wear on line 3 widens the OEE spread from about 1.5 to about 10.3 points, and the ranking says "line 3, availability" rather than only "line 3".

Every device carries the same PackML block: UnitMode, State, StopReasonCode, EquipmentBlocked, EquipmentStarved, ProdProcessedCount, ProdDefectiveCount, CurMachSpeed, DesMachSpeed and AccStoppedTime. Each machine adds its own tags, for example FillPressure and FillTimeMs on the filler, TorqueApplied and AirPressure on the capper, MeanNetWeight and RejectCount on the checkweigher.

Device names are the line and the machine: PL2_Capper is the capper on line 2, and PU1_SyrupRoom, PU1_AirHeader and PU1_CipSkid are the shared assets. The shared assets carry the plant numbers no line publishes, such as LinesDrawing and HeaderPressure on the syrup room, TotalDemand and CompressorsRunning on the air plant, and ServingLine and QueueDepth on the CIP skid.

Five machine models are instantiated once per line, the three shared assets have one model each, and a PackagingLine model has one instance per line whose attributes come from five different devices. A fourth line would add six instances and no new models.

Cap Integrity, Plant Giveaway and Line Availability each run one processor chain across all three lines, scoped by the tag name, so they do not grow when a line is added.

The solution ships two ways. Use the registry if the host can reach Google Artifact Registry, and the downloaded archive if it cannot.

Requires a read-only registry credential, supplied by Litmus separately.

Save this as docker-compose.yml:

Put the three values in a .env file beside it:

Download food-beverage-multiline-demo-0.5.0-amd64.tar.gz from the solution's page on portal.litmus.io. The archive is the container image, not a source bundle, and needs no registry access.

docker load prints the image it added:

That reference is the one the compose file above already names, so the same docker-compose.yml and .env now work offline with no docker login:

Open http://localhost:4021. Within about a minute the Edge is configured and the page fills in.

The page stays on its waiting panel until the Edge carries this solution's devices. That is deliberate: the simulation does not tick and no value is shown until Litmus Edge is verified to hold the devices and poll them, so nothing on screen can be mistaken for data that never made the trip. With APPLY_ON_START at its default of 1 the solution configures the Edge itself on startup, which takes 30 to 60 seconds. If you set it to 0, or the apply failed, press Apply to Edge in the page header.

An incorrect SIM_HOST fails quietly: the devices are created, they report as online, and no tag ever reads a value, because the address written into the template is one the Edge cannot reach. The solution refuses to apply a template whose address is obviously unreachable and reports why.

The header shows the worst line with its dominant loss, the product ring main pressure against its specification, the compressed air header against demand and running capacity, giveaway per day, the share of caps under torque, and what the CIP skid is doing and how many lines are waiting for it.

Below that are the three lines and the shared utilities, with the product and air mains drawn from the utilities to every line. Below them, the lines are ranked worst first. Each line can be started and stopped on its own, because how many lines run changes what each line sees.

Below the lines are the faults. Three are marked as shared utilities, affecting all three lines:

Each fault has a guide naming the DataHub topics, the Analytics group and the Digital Twin topic to watch while it runs.

The header carries both controls. Apply to Edge pushes the devices, twins and analytics, and greys out once the solution is on the Edge. Remove from Edge previews first, listing what would be removed and from which Edge, then offers to remove it.

The solution removes only what it created, and records which Edge its apply landed on, so repointing at a different Edge cannot leave a solution stranded on the first one.

Preview report and Export report produce one self-contained HTML file that opens offline, emails and prints. It carries a summary, the lines ranked by OEE with the dominant loss for each, the shared utilities with header pressure, giveaway by line and the weight distribution, a downtime Pareto, every CIP cycle with its result and how far past its clean window it ran, the faults injected, and the Litmus Edge configuration that produced it.

[email protected]