WP1.2 Coordination Meeting July 15, 2020

Meeting Report WP1.2 ‘Modelica library for MPC’

1. MEETING SUBJECT, DATE

Subject: WP1.2

Date: 15-07-2020

Location: Skype for Business

Minutes taken by: Lieve Helsen (KU Leuven)

2. PARTICIPANTS

3. AGENDA and REPORT

3.1. Approval/Modification Agenda

Approved

3.2. Status & further discussion where needed

a. Emulators:

4 models selected to start MPC testing

i. BESTEST Air (LBNL/KU Leuven):

merged to the master – READY! (Dave)

ii. BESTEST Hydronic (KU Leuven /LBNL):

merged to the master in IDEAS – READY! (Filip & Javier), BOPTEST case almost ready.

iii. 8-Zone Residential Hydronic (Engie/SDU):

In pull request, update needed for zone identifiers, no progress (Javier).

iv. Multi-Zone Office Hybrid Simple (KU Leuven/Engie):

Managed to compile the model with JModelica, simulation with JModelica does not run (action Iago/Javier). Have a look at the solvers used. Javier will point to the branch with the latter developments in the associated issue, so other can have a look into that.

Further progress has been made in other models too!

v. BESTEST Hydronic with modulating HP (KU Leuven):

bulk is ready, just a few comments that still need to be figured out, can be merged on the short term (Javier/Filip).

vi. Single-Zone Commercial Building (SDU/LBNL):

Tao is processing the feedback from Dave. Pull request will be sent to Dave in short term.

vii. Multi-Zone Air-Based Prototype Complex (PNNL/KU Leuven):

Sen sent a new version to Iago. Iago starts a detailed review.

viii. Flexible Research Platform - 10 sones office with VAV system (ORNL/SDU):

Yeonjin finalized preliminary building models and starts now comparison with measured data to calibrate some parameters. Ready for review by end of 2020 (Konstantin).

Emulators need to be put on BOPTEST repository in the directory testcases.

Emulators overview:

b. KPIs - scenarios

Choice between TMYx (1946-2018) and TMY3 (1973-2005) weather files. We have a data reader for TMY3 in IBPSA. Both will be compared in the representative days/weeks selection (action PNNL - Yan). No update.

c. BOPTEST:

i. Upgrade to Python 3:

Dave is working on a new version in Docker container that uses Python 3 and PyFMI (which can be installed easily), no JModelica-compiled image needed anymore:

1. Unit tests are passing for BOPTEST Docker container utilizing Python 3.
2. What is more difficult is the update of the parser that creates a new FMU (uses JModelica, which uses Python 2). Dave is working on this, the process works with Linux, not yet with Windows.

ii. Generation of new weather files based on historical data:

to be used for grey and black box model identification (training data that differ from test data). No further progress (Dave). To be discussed in breakout session.

iii. Working group on weather prediction uncertainties (Aritz):

Autoregressive model based on measurements and Q-Q mapping have been tried. Behaviour of the standard deviation has been studied. The error is too large for extreme temperatures. To be continued (Jesus/Aritz).

iv. BS2021 Conference:

Too early for contest now, targeted for 2022.

d. MPC Library (IbpsaMpc):

i. Unit testing framework (Dave):

No update.

e. Issues when using BOPTEST:

Harald brought up issue 213: Method of handling event updates by introducing a sampler with modified sample period. Solution: event update function is available in pyfmi or compile the model as co-simulation FMU. Dave suggests to do the latter, see conclusion in issue 213. Filip suggests that we model as much as possible in Modelica and look for the best solution for each case. Specifically for the hysteresis block we could reformulate the model block. The easiest way to fix is adding a filter on the input.

f. Dissemination:

Broaden the user space (e.g. to Machine Learning people). How to reach these people? Jan: make a first suggestion. Cfr OpenAIGym, to be presented. This will be included in the program of plenary session of the next IBPSA Project 1 (on-line) meeting in October 2020.

3.3. Initiatives joint papers

a. Javier & Lieve submitted an abstract for BS2021: Reinforced Model Predictive Control for Building Energy Management.

b. Filip, Damien & Lieve submitted an abstract for BS2021: Automated workflows for optimal design and control of buildings using Modelica.

c. Konstantin, Toa & Krzysztof submitted an abstract for BS2021 with focus on parallel computation.

d. Dave thinks about a journal paper on BOPTEST.

3.4. Miscellaneous

a. Break-out sessions definition for next IBPSA Project 1 meeting (Lieve)

i. Spreadsheet for internal reporting of BOPTEST test cases

ii. Public reporting (limited information, end-user agreement)

iii. MPC competition: how to set up?

iv. How to deal with uncertainties: forecast of weather, loads, occupants?

v. How to generate weather data for training grey/black box models?

vi. Workshop introducing BOP-TEST at BS2021 (September 2021), as a follow-up of paper at BS2019 in Rome, incl. inspiration towards use cases such as: training building operators, assist teaching ...

vii. How can data be used in MPC?

b. Montreal meeting postponed to October.

c. September 21: Planning Breakout sessions for next IBPSA Project 1 on-line meeting for WP Leaders (Lieve).

d. Next meeting: September 9 at 5 PM (Brussels time), Lieve sends invitation.