Skip to content

2019 10 23 OBC Verification

Jump to bottom
Antoine Gautier edited this page Oct 23, 2019 · 2 revisions

[Hwakong Cheng (Taylor Eng.), Antoine Gautier (LBL, writing), Dave Guerrant (Integral), Philip Haves (LBL), Jim Kelsey (KW Eng.), William Casper-Ortiz (Integral), Jay Santos (Facility Dynamics), Paul Switenki (Arup), David Pritchard (Arup)]

Need for a "sandbox" testing of control sequences?

  • Typically how to guard against a bug in CDL translation leading to equipment damage?

  • Run the controller in a functional testing framework (test bed) with BACnet override before mounting the sequence.

What time series should be used to emulate the plant? Who would specify them?

  • Currently possible with the main graphical control programming frameworks: ALC, Trane ("simulation mode"), Siemens? (Text programming frameworks are unlikely to support that functionality.)

  • However seems the exception rather than the rule in current practice.

  • Could potentially alleviate the foreseen problems related to the use of a "digital twin" e.g. controller tuning parameters discrepancies, mechanical hardware effects, etc.

Feedback on LBL paper about verification & validation

Two concepts:

  1. Test that the sequence is correctly implemented = verification (can be done offline, see "sandbox" testing)

  2. Test that the system works correctly = validation (must be done on the installed system)

Example: Actuator rotating clockwise = assumption for simulation testing and installed differently

  • Point to point validation step is key.

Modifications during construction project

How to keep the upstream specification synchronized?

Need for reverse translator from vendor to CDL?

  • What do we verify in the end with that reverse translated CDL implementation? Not the original control intent.

The need for specifying how to take into account the changes that occur in the building process is critical.

OBC workflow must accommodate for that.

  • Is it a requirement for OBC and realistic regarding current practice? Alternative: specification of boundaries (expressed in terms of effects on trended variables) to what is acceptable: depending on the changes they can be significant...

How CDL gets chunked?

That topic appears more as a part of the translation process development: what is the exact impact on the verification process?

  • Isolate the different control functions that should be tested individually.

  • ALC e.g. economizer I/O on the same "expansion" module.

  • Feedback needed to Dave Robin.

  • Engage additional industry partners to benchmark practices.

Next steps

  • Develop three short form, high level use cases

    Describe current practice and what changes would be needed for CDL adoption and the added value of CDL. Are those changes realistic? etc. (Could evolve into a user manual of the verification tool in the end.)

    • Designer & consultant: Paul, Dave

    • Installer: Paul, Jay, William

    • Commissioning agent: (William), David

Clone this wiki locally