-
Notifications
You must be signed in to change notification settings - Fork 14
2017 08 29 review cdl
Date: August 29, 2017, 10am-12pm PDT.
Slides are posted at https://github.com/lbl-srg/obc/blob/master/meetings/2017-08-29-cdl-review/obc_cdl_review.pdf
- Project overview and CDL discussion. See slides.
- Questions and feedback from audience.
See meeting invitation.
- Jan Baker, Siemens Building Technologies
- Greg Bates, Reliable Controls
- Larry Bacher, Automated Logic
- Scott Boehm, Tridium (for the first part)
- Tony Bruno, Trane
- Chris Vause, Trane
- Jay Santos, Facility Dynamics
- Charlie (?), Delta Controls (we had invited David Amos)
- Paul Ehrlich, PNNL
- Philip Haves, LBNL
- Michael Wetter, LBNL
- Jianjun Hu, LBNL
- Milica Grahovac, LBNL
Michael initiates a round of introductions and the purpose of the meeting, which is to obtain vendor feedback on CDL basic blocks.
Michael presents OBC goals, challenges, provides implementation example and contents of the CDL library - Please refer to the slides for more detail. Michael and Phil asks for feedback.
Phil seeks feedback on the missing pieces and barriers. Jan references CFC and inquires about deadband, arrays, bit mapping, character mapping, etc. Michael suggests that Jan provides a list in written form. Jan says that they are developing the tool which creates control strategies and sequences based on written specification and answers to a set of questions, accompanied by a patent (he’s provided a patent number earlier).
Jan asks about any new NDA. Michael answers that OBC is a vendor neutral, open standard that does not favor any particular company. To keep the project vendor-neutral, transparent and open, we prefer not to enter individual NDAs for this work.
The discussion continues on the topic of language translation. Michael says that we intend to provide a proof of concept using one vendor sequence and replicate the results using identical inputs.
Jan asks about the following specific features: alarms, optimization, scheduling, peak demand limiting. Michael informs that some of start-up sequences are proprietary, in which case CDL could use the proprietary vendor optimal start-up block in the actual control implementation, but a generic implementation for the testing. The discussion continues about optimal start-up algorithms, Jan says that vendors might want to adopt a CDL solution if OBC provides it. Michael repeats that CDL intends to insure input/output connectivity that would allow the usage of proprietary start-up blocks.
In that context, Michael mentions convenience blocks, like those available in Niagara, and we'd need to pick what we want to have implemented in the first version. Jan expresses interest in calendar time, timetable, emphasizes the importance of Canadian market. Phil replies that OBC targets global market to maximize impact. Michael mentions that CDL calculates in SI units, but display can be in any other specified unit standard. Jan turns the discussion towards syntax, talks of PPCL, he would like to see as few statements as possible and elaborates on how number formats should not be a concern to the user.
Phil seeks feedback on compatibility issues. Discussion starts on sequence of execution, JSON gets mentioned. Michael explains that CDL uses sequential data flow. He further explains why change of value is not suitable - non-determinism if the order of assigning inputs is not deterministic. For that reason, in CDL even if the inputs do not change, blocks are being invoked and provide immediate response. Michael states that change of value also would not be suitable for simulation, which we would use to verify the energy use. He raises an issue of how to wrap around any alternative methods of calculation in terms of communication with vendor tools. Phil asks for feedback on common computation methods from vendors.
Phil discusses CDL as used for comparing the designs, prior to selecting the vendor, and implementation into hardware.
Paul thanks everyone and opens final questions, Jay joins. Michael expresses hope that we can keep developing the tool that will be useful to the building industry even after the initial 3 years of project duration expire.
A question about implementing fault diagnostics - not planned for the first version of the library.
Phil closes the meeting.
Jan to provide list of features that would be desired from Siemens perspective. Phil asks for feedback on common computation methods from vendors. Michael asks to indicate whether the feedback is shareable or not.
Next steps: library release, end of September, edits can be added afterwards as well, G36 seq implementation - demo with a close loop response, in about 3 months - sequence in an electronic format for downstream processing.
Library release.
Closed loop implementation.