This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Virtual Commissioning

Former Community Member
Former Community Member
Hello,


​I would be interested to hear the experiences from those who uses virtual commissioning to test PLC code prior to actual commissioning.


​We use a system called WinMOD, where the machine / line is modeled in 3D and to scale. Sensors, safety devices, etc. are added to the model as they would be on the actual line. As the system is to test the completeness and quality of the PLC code, hardware such as VFD's are substituted using macros that replicate the actual functions and connections they are emulating. The PLC I/O is assigned to the emulated hardware and line sensors. Once completed, the PLC is connected to the PC running the 3D model and associated software. An emulated control station & HMI can then be used to run the line, as if it were the actual line.

The PLC / HMI software is tested through normal and abnormal conditions, faults simulated and recovered. Software issues can be resolved and fault recovery simplified. The line can be left running for a number of days to test the robustness of the PLC code. All this prior to the actual commissioning. 

Whilst the VC process can appear to add significant cost to the project, actual commissioning time will be reduced and subsequent downtime from software bugs should be eliminated. If done correctly, a digital start up should be possible, hence achieving an overall cost reduction. VC is not for every automation and controls project, cost analysis is needed to ensure that it is cost effective for the project.

4375c582ab45724a99c974f00e01dd7a-huge-winmod.png

3D Model of a production line.



Kind regards,


Jon
Parents
  • I've had some experience of using models to help commissioning, but more for large bespoke machinery control software (e.g. hydraulic marine equipment, position and force control) rather than manufacturing lines (mainly fast logic based automation). The main thing to consider what aspect is it you are not sure about that would benefit from model based commissioning, it could be related to complexity, uncertainty in solution, penalty with getting design wrong and it not working or low performance (cost, delays), inacessibility to equipment for testing sub-systems prior to commissioning, normal operation or checking response to fault conditions (important in safety critical) etc. I know those things are obvious, but it is worth identifying what the main reason for doing it, as that will help define what aspects you need to use models to support commissioning.


    There are many levels of doing virtual commissioning - MIL (model in loop - i.e. just simulations), SIL (software in loop - i.e. real control code but connected to simulated plant), HIL (hardware in the loop, i.e. real control code running on real hardware connecte via real IO to a separate unit that simulates all of the physical system). You'll probably find your PLC type will have some sort of add-on to do these sort of things - though with varying degrees of capability (e.g. Siemens have SIMIT). I've also come across Emulate 3D more for commissioning physical layout of conveyor systems.


    The virtual commissioning we have done used NI LabVIEW, and it was built specifically for a a given application including emulation of both dynamic elements and logic based faults so we could work through an entire FAT in the office a well before the final equipment was built. LabVIEW gave us a lot of flexibility to build a test harness / simulation for the specific things we needed to explore. We could have certain parts of the system simulated, and other parts connected to the real system - and this helped a lot.


    Our company has also built dynamic nonlinear math models in oil/gas applications where control settings obtained in early simulation studies have been directly used in the commissioning of the equipment, saving a lot of time setting up the control. Leading on from William's comment, this is only possible with a very well validated model - which for the type of work we do is pretty essential / and is largely down to the experience of people building the model. For our type of dynamic models we often use grey-box parameter identification to fit data to uncertain parameters in the models, but there is a real skill to defining parts of the model that are not well known. However, I don't think that type of modelling / validation is so relevant to manufacturing lines.


    I hope this is helpful, even if not totally like you are looking to do.

    Andy Clegg

    Industrial Systems and Control

Reply
  • I've had some experience of using models to help commissioning, but more for large bespoke machinery control software (e.g. hydraulic marine equipment, position and force control) rather than manufacturing lines (mainly fast logic based automation). The main thing to consider what aspect is it you are not sure about that would benefit from model based commissioning, it could be related to complexity, uncertainty in solution, penalty with getting design wrong and it not working or low performance (cost, delays), inacessibility to equipment for testing sub-systems prior to commissioning, normal operation or checking response to fault conditions (important in safety critical) etc. I know those things are obvious, but it is worth identifying what the main reason for doing it, as that will help define what aspects you need to use models to support commissioning.


    There are many levels of doing virtual commissioning - MIL (model in loop - i.e. just simulations), SIL (software in loop - i.e. real control code but connected to simulated plant), HIL (hardware in the loop, i.e. real control code running on real hardware connecte via real IO to a separate unit that simulates all of the physical system). You'll probably find your PLC type will have some sort of add-on to do these sort of things - though with varying degrees of capability (e.g. Siemens have SIMIT). I've also come across Emulate 3D more for commissioning physical layout of conveyor systems.


    The virtual commissioning we have done used NI LabVIEW, and it was built specifically for a a given application including emulation of both dynamic elements and logic based faults so we could work through an entire FAT in the office a well before the final equipment was built. LabVIEW gave us a lot of flexibility to build a test harness / simulation for the specific things we needed to explore. We could have certain parts of the system simulated, and other parts connected to the real system - and this helped a lot.


    Our company has also built dynamic nonlinear math models in oil/gas applications where control settings obtained in early simulation studies have been directly used in the commissioning of the equipment, saving a lot of time setting up the control. Leading on from William's comment, this is only possible with a very well validated model - which for the type of work we do is pretty essential / and is largely down to the experience of people building the model. For our type of dynamic models we often use grey-box parameter identification to fit data to uncertain parameters in the models, but there is a real skill to defining parts of the model that are not well known. However, I don't think that type of modelling / validation is so relevant to manufacturing lines.


    I hope this is helpful, even if not totally like you are looking to do.

    Andy Clegg

    Industrial Systems and Control

Children
No Data