WP1.2 Coordination Meeting January 20, 2021

Meeting Report WP1.2 ‘Modelica library for MPC’

1. MEETING SUBJECT, DATE

Subject: WP1.2

Date: 20-01-2021

Location: Skype for Business

Minutes taken by: Lieve Helsen (KU Leuven)

2. PARTICIPANTS

3. AGENDA and REPORT

JANUARY 20

1. Reporting results from testing & sign-up spreadsheet (all & Dave)
2. Weather uncertainty: analysis for Denver and Norway locations (Laura) - postponed
3. SDU emulator: ready for testing advanced controllers? (Konstantin)
4. BOPTEST methodology journal paper: update status (Dave)
5. BS2021 Workshop (Dave & Javier)
6. Outline BOPTEST methodology journal paper: update status (Dave)
7. Update emulators development (Developers)

MARCH 1

1. Suggestions for next on-line IBPSA Project 1 Expert Meeting
2. SDU emulator: ready for testing advanced controllers?

Reporting results from testing & sign-up spreadsheet

BOPTEST testing results:

• Javier: good experience with ML approaches in BOPTEST
• Iago: first test with PI controller – works great, easy to use
• Harald: first trial with MPC, easy to use. Issues:
• Delay in pump operation was observed (cf event handling when hysteresis was applied). Dave & Harald look in the notes again
  • How is energy measured? Do we need energy meters in the emulators using hydronic systems (such that we do not have to do the integration at each step ourselves, and have the accumulated value)? Now: integrate power from results file or even better, use the KPI calculator, but this integrates the energy use of all components together.
    • Links to the discussion whether to calculate the energy use in- (sampling at a lower frequency) or outside the FMU.
    • Are energy meters available in practice? If not, should we then add these to the emulators? Do we look at existing (imitate real situations) or future systems (access to smart meters)? The same holds for zone temperatures. Sensor costs should then be added to the installation cost, though this is very hard to maintain and very location-dependent. We keep it on checking boxes to track the sensors that are used  can we implement this already? This should be fairly easy by adding flags.

Spreadsheet to keep track of tests that have been done

How should we define the testing scenarios, and more specifically the time periods used (Dave)? Proposal to use peak heating/cooling or typical (median) heating/cooling periods, instead of random periods. Testing for 2 weeks, where we use 1 week (with baseline controller) prior to these 2 weeks to eliminate the influence of initial conditions. This 1 week simulation can be done beforehand and initialize the states in the FMU. The sensor data from that 1st week need to be accessible for e.g. state estimation, so needs to be put in history of data. https://github.com/ibpsa/project1-boptest/issues/270.

Reflections:

• Good to limit to specific periods to do the benchmarking on
• Some emulators might need more:
  • Longer periods might be needed for cases with borefields or other long-term effects (seasonal energy storage), e.g. Iago’s emulator with properly-sized borefield
  • Mid-season challenges (warm or sunny days combined with cold nights) that encounter switching between heating and cooling, e.g. Spring in Belgium

Improved spreadsheet to sign up for the tests (Dave)

A bit more structured, see https://docs.google.com/spreadsheets/d/1E-5wR7nasW8h6kEtrXnUzcRrEA6f7TIxjbtSdB1Cz3w/edit#gid=1356294432 Shouldn’t we make the developers of the emulators more visible? Acknowledge them in the Readme file of the Test Cases repository for example, possibly combined with acknowledgment to the funding agencies.

SDU emulator: ready for testing advanced controllers?

Update by Tao: emulator is not yet ready, but aim is to deliver by 1st of March (agenda next meeting).

Software update BOPTEST

Change Results API (Dave): no longer all historic data, you need to specify which point name you want data for, with start and end times, for multiple points a for-loop needs to be implemented. BOPTEST for RL is advancing (Javier): see BOPTEST-Gym track in repository

BS2021 Workshop

Proposal sent around: approved! Thanks to Engie for sponsoring ;-) Please submit (Dave & Javier) https://docs.google.com/document/d/1tP5CvX49NiETf6HP7c-gK2NmIxOAUc-TGDlxx-7IuqI/edit?usp=sharing

Outline BOPTEST journal paper focusing on methods

Authors have been contacted by Dave and outline can be found: https://docs.google.com/document/d/1a-faviZXIHm-uVkIzucHCD3GP3NuDvApw4SW81cR-MA/edit#

Focus on architecture and methodology. Showcase on BESTEST Hydronic Heat Pump with variable electricity price. 2nd paper is more application-oriented: showcase the benchmarking aspect for selected test cases – Journal Applied Energy

Further issues

1. Call to work on emulators or testing of advanced controllers on emulators that have been finalized
2. Updates on Emulators: a. Single-zone commercial hydronic: to be finalized by 1st of March (Konstantin/Tao) b. Multi-zone (8z) residential hydronic heating (gas boiler): almost finalized (Javier) c. Multi-zone commercial air-based: preliminary model + documentation ready by 1st of March (Yeonjin) d. Multi-zone (5z) commercial air-based: development started (Dave) e. Multi-zone office hybrid (simple): restart working on it, finalized in 2-3 months (Iago) f. Multi-zone office hybrid (complex): progress in March (Filip)
3. Status: