-
Notifications
You must be signed in to change notification settings - Fork 14
2018 06 11 charrette agenda
Participation at this meeting is by invitation only.
Date: June 11, 2018
Location: Arup SF Office - 560 Mission, Suite 700, San Francisco, CA 94105, Yosemite Conference Room
Slides are posted at https://github.com/lbl-srg/obc/blob/master/meetings/2018-06-11-charrette/obc_overview.pdf
See below agenda for dial-in information.
| Time | Whole Group | Performance Group | Additional Functionality Group |
|---|---|---|---|
| 8:45 | Coffee and Pastries | ||
| 9:00 | Introductions/Welcome | ||
| 9:10 | Overview of OBC and challenges | ||
| 9:55 | Breakout to performance analysis and specification groups | ||
| 10:00 | Detailed problem statement | Detailed problem statement | |
| 10:10 | Current analysis practice | Content requirements | |
| 10:20 | LBNL team 'straw man' | Sources for content | |
| 10:30 | Input from practitioners | Can content be separated from systems/sequences? | |
| 11:00 | Break | ||
| 11:10 | Report back from specifications group | ||
| 11:30 | Report back from performance analysis group | ||
| 11:50 | Next steps | ||
| 12:00 | End |
https://lbnl.zoom.us/j/7506939211 669 900 6833 or +1 646 558 8656 Meeting ID: 750 693 921 International numbers available: https://zoom.us/u/oRJSpnOD
https://lbnl.zoom.us/j/6614042296Or iPhone one-tap : US: +16465588656,,6614042296# or +16699006833,,6614042296# Or Telephone: Dial(for higher quality, dial a number based on your current location): US: +1 646 558 8656 or +1 669 900 6833 Meeting ID: 661 404 2296 International numbers available: https://zoom.us/u/w8m8F
- David Pritchard, Arup
- Rick Stehmeyer, CxAssociates
- Karen Perrin, CEC
- Marina Sofos, DOE
- Amir Roth, DOE
- Brent Eubanks , Carbon Lighthouse
- Allan Daly, Brightbox
- Erik Dyrr, Guttmann & Blaevoet
- Erik Kolderup, Kolderup Consulting
- Jason Kirkpatrick, Interface
- Mark Hydeman, Google
- Neil Bulger, Integral
- Steve Taylor, Taylor Engineering
- Jianjun Hu, LBNL
- Milica Grahovac, LBNL
- Brian Turner, Operational Technology Integrators
- Jay Santos, Facility Dynamics Engineering
- Paul Erlich, PNNL
- Paul Switenki , Arup
- Annie Mroz, Integral
- John McDonald, Integral
- Jim Kelsey, KW
- Kevin Li, Sunbelt Controls
- Santosh Philip, Loisos+Ubbelohde
- Philip Haves, LBNL
- Michael Wetter, LBNL
Michael introduces the content and the intent of the meeting, which is to present current status of the OBC project and obtain feedback.
Philip and Paul introduce the intended breakout into two groups: performance and design group.
Philip presents goals, objectives, framework, challenges and the stakeholder structure of OBC.
Michael presents the envisioned workflow for controls design, implementation, testing and commissioning. Describes CDL as a vendor independent open source language for expressing, rendering and documenting control sequences. Allan inquires about aggregating blocks into complex blocks and additional subroutines. Michael explains that the latter currently is not allowed, whereas complex blocks are a suggested practice.
Michael shows an example G36 sequence implemented and documented using CDL. He also explains the position of OBC in DOE’s strategy to rearchitect building simulation tools - presents SOEP. The next example is last year’s multizone VAV G36 control sequence case study performed on a simulated multizone building, with the major finding that an upgrade of the control strategy may, with no additional equipment cost, significantly reduce the energy use. Michael also explains the benefits of having such complex sequences (as G36 sequences tend to be), which allow for energy savings, prepackaged as blocks. As a conclusion and a transition to work in groups, Michael presents a high level process flow diagram (mechanical designer, control provider, commissioning agent).
Phil introduces the open questions in the field of building energy/controls modeling and simulation:
- Who uses energy models? Which software is being used, at which stage (design, commissioning, retrofits, operation and maintenance)? Are control sequences being modeled?
The discussion touched upon the following items (see below for details):
- conventional buildings (bulk of new construction)
- lots of potential lies in retrofits
- model complexity
- simulation time
- level of detail (utilization stage)
- modeling for loop control?
- current practice includes 3 models - load, system design and a separate code compliance model
- approach to load modeling
- self-fixing logic
- scenario runs and presenting optimal solutions to customers
- energy model validation
- control sequence model validation - current practice
- new interest in energy models due to metering based incentives
- where do the savings come from?
Allan emphasizes that most buildings are conventional/average (refers to equipment and sequences implemented).
Erik K talks about retrofits - most of the savings are related to sequence and equipment savings (e.g. different fan curve). Allan talks about increasingly complex models during design. E.g. how does a piece of equipment improve the performance in reference to the baseline equipment, under the same building load. Allan points out the simulation time. Erik brings up the issue with the lack of pressure variable in currently available models. Steve's view is that modeling is not needed for loop control. He explains that 90% of savings potential can be identified with methods more rudimentary and much faster than simulation, while the time spent on the last 10% (simulation level of detail) can be costly. Steve talks about a need to create 3 models during design: a load model, a separate model to satisfy Title 24, and the energy model for the equipment design. He mentions how user behavior models tend to be too complex. In terms of design, the practitioners tend to use the 60.1 floor plan, rather than the baseline load shapes. He adds that it would be great to have logic able to fix itself. Neil and Erik bring up the ability to run scenarios using simulation. In response, Steve talks about optimization. Discussion continues on sizing and energy model validation.
Philip H talks about “the rest of the industry” sequences that are not out of a box. Philip S also states that there is no need for detailed loop simulation and Steve adds that the validation tests need to capture only "a gist of the sequence". Allan brings up some kind of specs model - to help validate the intuitive design, especially in the new territory.
Neil mentions future need for models due to meter based incentives - maybe regression models.
The key feedback from the content group is summarized as follows:
CDL need to allow for "hooks" that enable obtaining external data, such as price signals from a web server.
There should be training to allow others to develop control libraries.
There should be a site that hosts control sequence libraries.
Many mechanical engineers cannot edit or create control block diagrams, or implement sequences in general. Therefore, CDL should allow a mechanical engineer to overwrite the English language documentation, and then CDL should flag that the corresponding block diagram is no longer consistent with the English language. A control provider can then implement the block diagram for this part of the sequence.
English language is critical for contractual purposes, at least during a transition period. As insurance companies need to sign of on the work, and hence it cannot be experimental.
CDL need to specify what happens in non-normal events, e.g., no power, fire alarm, and who is notified (PI, building engineer, ...). Some sequences for non-normal events are hardwired, some software, but they need to be documented. Examples include
- Alarm processing
- Optimum start
- Time schedules.
For hardwired sequences, may be able to flag which blocks have a mechanical equivalent.
Provide standard blocks for special functions such as optimal start up. Provide blocks with math implementation so vendors can implement it.
Control spec builder and Eikon Builder is often not used by control engineer if changes are needed. Control implementer often starts with implementation from other projects as this is faster.
Instantiating a CDL sequence should be done through a check-box driven sequence selector.
Mark updates on ASHRAE G36: dry side sequences 3rd public review got voted out for publication.
Design group: Paul talks about steps needed on the design side: develop and manage a library of sequences, G36 and non-G36 sequences, and an approach to black box sequences (e.g. optimal start).
Topics:
- improved performance
- risk management
- English language description
- approach to sequence modification
- exposing parameters to users
- black box algorithms (demand response, optimal start, trend logging)
- workflow with CDL preconfigured sequences
Phil reports from the performance group:
- opinions on local loop simulation
- conventional buildings - decrease design and commissioning effort
- simulation as a tool to inform the design - the role of control simulation: ability to compare performance of different sequences
Comments: Brent - make the controls tool such that with a few multiple choice selections the design engineer can put together the control program. Paul - on the issue of how bits and pieces of the tool get adopted over time Jim - suggests that the sequences should be tested on real buildings Allan - Unit (before deployment) and integration (in real physical systems) testing
Participant feedback
Case study (Oracle)
TAG in a few months