Using the Open Source OPC-UA Client and Server for Your IIoT Solutions

While currently a lot of the focus of time-series databases is on DevOps use-cases, they can also have a major impact for industrial applications. This talk focusses on the industrial application of InfluxDB, enabled bij the OPCUA protocol. It shows how we built an open-source OPC-UA client and server to bring data from and to your process control systems. We also zoom in on how Flux fits into the picture.

(Talk originally presented at InfluxDays London 2019

Transcript

1. Using the Open Source OPC-UA Client and Server for Your IIoT Solutions JEROEN COUSSEMENT - INFLUXDAYS 2019 - LONDON
2. ABOUT FACTRY ▸ Based in Ghent, Belgium ▸ Industrial software solutions ▸ Tools for data-driven operational improvement ▸ Factry Historian ▸ Lots of open-source
3. TIME-SERIES MEETUP https://www.meetup.com/Time- Series-Belgium/
4. INDUSTRY
5. INDUSTRY 1.0 STEAM POWER / MECHANISATION
6. INDUSTRY 2.0 MASS ASSEMBLY
7. INDUSTRY 3.0 COMPUTERS & AUTOMATION https://www.flickr.com/photos/greenmambagreenmamba/5307396963/in/photostream/, (CC BY-ND 2.0))
8. SUPERVISORY CONTROL SYSTEM (SCADA/DCS) PROGRAMMABLE LOGIC CONTROLLER (PLC) PROCESS SENSORS AND ACTUATORS PROGRAMMABLE LOGIC CONTROLLER (PLC)
9. INDUSTRY 4.0 CYBER PHYSICAL SYSTEMS / AI / ML
10. A GAP BETWEEN OT AND IT. OT IT ?
11. SOMETIMES WE HAVE TO GO FULL RETRO VERY LONG LIFECYCLES
12. “WE’VE BEEN DOING IT FOR YEARS AND IT WORKS” SLOW ADOPTION
13. EVERYONE HAS THEIR OWN STANDARD VENDOR LOCK-IN
14. SUPERVISORY CONTROL SYSTEM (SCADA/DCS) PROGRAMMABLE LOGIC CONTROLLER (PLC) PROCESS SENSORS AND ACTUATORS EXTERNAL APPLICATIONS ( Artificial Intelligence / Machine Learning / … ) PROGRAMMABLE LOGIC CONTROLLER (PLC)
15. WHAT KIND OF TIME-SERIES DATA DO WE HAVE IN INDUSTRY ?
16. DEVOPS DATA vs INDUSTRIAL DATA DevOps Industry Metrics cpu load, disk I/O, database stats, ... temperature, pressure, flow, valveState ... Resolutions seconds to minutes (sub)seconds Retention weeks, then downsampling 5 to 10 years, no downsampling Main goals incident detection, performance monitoring quality guarantee, predictive maintenance
17. THE FIRST TIME-SERIES DATABASES
18. PROCESS HISTORIAN
19. Process historian refers to a database software application that logs or historizes time-based process data.
20. SUPERVISORY CONTROL SYSTEM (SCADA/DCS) PROGRAMMABLE LOGIC CONTROLLER (PLC) PROCESS SENSORS AND ACTUATORS InfluxDB EXTERNAL APPLICATIONS ( Artificial Intelligence / Machine Learning / … ) PROGRAMMABLE LOGIC CONTROLLER (PLC)
22. OPC-UA DESCRIPTION ▸ Industrial communication protocol ▸ Successor of OPC - OLE for Process Control ▸ Vendor independent / Platform independent ▸ Released in 2006, but became adopted only recently ▸ Integrated on newest controllers, but retrofit on older ones possible.
23. OPC-UA DATA STRUCTURE ▸ Hierarchical data structure ▸ Everything is represented as a data node. ▹ Each node has a unique nodeID ▹ Consists of namespace + identifier ▹ ns=2;s=S7:12_03_12.DB100.4,r
24. SUPERVISORY CONTROL SYSTEM (SCADA/DCS) PROGRAMMABLE LOGIC CONTROLLER (PLC) PROCESS SENSORS AND ACTUATORS InfluxDB EXTERNAL APPLICATIONS ( Artificial Intelligence / Machine Learning / … ) PROGRAMMABLE LOGIC CONTROLLER (PLC)
25. OPC-UA LOGGER OPEN SOURCE ▸ Reads data from OPCUA sources ▸ Local buffering for no data loss ▸ github.com/coussej/ node-opcua-logger
26. OPC-UA LOGGER DEMO ▸ Configure some nodes to be collected from simulation server
27. OPC-UA LOGGER OPEN SOURCE ▸ V2 coming soon ▹ Cleaner code (async/await!) ▹ Better buffering ▹ Deploy as a single binary
28. USE-CASE: BIOREACTOR
29. USE-CASE: ENERGY MONITORING
30. USE-CASE: WIND TURBINE OPERATIONS
31. SUPERVISORY CONTROL SYSTEM (SCADA/DCS) PROGRAMMABLE LOGIC CONTROLLER (PLC) PROCESS SENSORS AND ACTUATORS InfluxDB EXTERNAL APPLICATIONS ( Artificial Intelligence / Machine Learning / … ) PROGRAMMABLE LOGIC CONTROLLER (PLC)
32. OPC-UA SERVER OPEN SOURCE ▸ Return algorithm results back to the process control ▸ Exposing last() value of specified series over OPCUA ▸ github.com/factrylabs/ influx-opcua-server
33. OPC-UA SERVER OPEN SOURCE ▸ DEMO
34. HOW DOES FLUX FIT INTO THE PICTURE?
35. FLUX CUSTOMER QUESTION 1 ▸ Input Pressure ▸ Output Pressure ▸ Can we calculate Pressure drop? ▸ Math across measurements!
37. DEMO FLUX - Pressure drop (1/2) UF1PT01 = from(bucket: "historian") |> range(start: 2019-05-01, stop: 2019-05-25) |> filter(fn: (r) => r._measurement == "CV-UF1PT01") |> aggregateWindow(every: 3m, fn: mean, createEmpty: false) UF1PT02 = from(bucket: "historian") |> range(start: 2019-05-01, stop: 2019-05-25) |> filter(fn: (r) => r._measurement == "CV-UF1PT02") |> aggregateWindow(every: 3m, fn: mean, createEmpty: false)
38. DEMO FLUX - Pressure drop (2/2) join( tables: {pt01:UF1PT01, pt02:UF1PT02}, on: ["_time"] ) |> map(fn: (r) => ({ _time: r._time, _pressureDrop: r._value_pt01 - r._value_pt02 })) |> filter(fn: (r) => r._pressureDrop > 0) |> aggregateWindow(columns: ["_pressureDrop"], every: 1h, fn: mean, createEmpty: false)
39. FLUX CUSTOMER QUESTION 2 ▸ NTU (Nephelometric Turbidity Unit) ▸ Machine Status ▸ Can we see the NTU only when the machine is in production?
41. DEMO FLUX - Filter on status (1/2) NTU = from(bucket: "historian") |> range(start: 2019-05-01, stop: 2019-05-02) |> filter(fn: (r) => r._measurement == "CV-UF1NTU01" and r.status == "Good") |> aggregateWindow(every: 30s, fn: mean) |> keep(columns: ["_value", "_time"]) Status = from(bucket: "historian") |> range(start: 2019-05-01, stop: 2019-05-02) |> filter(fn: (r) => r._measurement == "Status" and r.status == "Good") |> map(fn: (r) => ({ _time: r._time, _status: float(v: contains(value: int(v: r._value), set: [11,21]) }), mergeKey: false)
42. DEMO FLUX - Filter on status (2/2) union(tables: [NTU, Status]) |> sort(columns: ["_time"], desc: false) |> fill(column: "_status", usePrevious: true) |> fill(column: "_status", value: 0.0) |> fill(column: "_value", usePrevious: true) // same for the values themselves |> fill(column: "_value", value: 0.0) |> map(fn: (r) => ({_time: r._time, _filteredValue: r._value * r._status}), mergeKey: false) |> yield()
43. THANK YOU! ANY QUESTIONS?
44. JEROEN COUSSEMENT jeroen.coussement@factry.io FACTRY OPEN OPERATIONAL INTELLIGENCE