WP1.2 Coordination meeting March 23, 2018

Meeting Report WP1.2 ‘Modelica library for MPC’

1 MEETING SUBJECT, DATE

Subject: WP1.2

Date: 23-03-2018

Location: Skype for Business

Minutes taken by: Javier Arroyo (KU Leuven)

2 PARTICIPANTS

Company/Organisation Participant

KU Leuven Lieve Helsen (WP Leader)

KU Leuven Iago Cupeiro

KU Leuven Javier Arroyo

KU Leuven Filip Jorissen

SDU Kzrysztof Arendt

LBNL David Blum

ENGIE Lab Valentin Gavan

ENGIE – Axima (now at LBNL) Lisa Rivalin

Pacific Noerthwest National Laboratory Draguna Vrabie

3E Roel De Coninck

3 AGENDA

Short overview of candidate emulator models, see https://docs.google.com/document/d/1uT17uEteoXmaDZVYm0DxGODje4Pxfe7oqKPCF4oIhmM/edit?usp=sharing

a. ENGIE Lab – validated office building model (hydronic system), 4 construction periods

b. ENGIE Lab – more detailed 8-zones residential building model (hydronic system – hybrid hydronic system, 5 variants), 4 construction periods

c. KU Leuven – Infrax GEOTABS office building (hydronic system)

d. KU Leuven – Solarwind hybrid GEOTABS office building (hydronic)

e. LBNL – 5-zones office floor with VAV system, currently with ideal heating and cooling plant (air-based)

f. PNNL – extension of 5-zones office floor (LBNL) including water loops to VAV boxes (air-based)

g. PNNL – 15-zones (3 floors) office building with single-duct AHU and VAV (air-based)

h. SDU – teaching building with offices, cameras for occupancy detection, EnergyPlus performance gap is 40-60%, Modelica model still needs to be developed (4-zones or single zone) (hydronic system + balanced ventilation)

2. List of requirements for emulator models should lead to template (ACTION Working group Toolchain (David, Filip, Krzysztof, Draguna, Michael, Roel, Damien, Javier, Lisa))

3. Selection of emulator models

4. What is still lacking?

5. Agreement on further steps

6. Other issues?

4 DISCUSSIONS

4.1 SHORT OVERVIEW OF CANDIDATE EMULATOR MODELS

4.1.1 Residential buildings

Valentin: we propose the 8 zones residential building of 100 m2 which is very detailed. It uses the Buildings library. It contains information about the envelop, occupancy… only has a heating system (no cooling) in the current version. It uses a controller that drives the boiler at on/off mode but a VRV system can also be included for cooling. The occupancy profiles are conventional profiles coming from the regulation in France. These schedules can be modified. It is possible to change easily the performance of the building envelope, so it is possible to run different tests with different levels of performance. There are also different possibilities for the heating system, e.g. radiators and radiant ceiling. If needed, those variations can be used, but we feel more comfortable using the boiler and radiators since this is more typical for a residential building.

Lieve: we could consider an extension to a hybrid system.

Valentin: the weakness is the calculation time: around 2 hours for an entire year.

Filip: there seems to be potential to speed up, try Filip’s suggestions.

Lieve: this is a very good candidate for the multi-zone residential building. I propose to use one version with only heating and a second version with the VRV added.

Roel: is VRV actually used in residential buildings? I believe they are more used in tertiary buildings.

Valentin: it is an air conditioner (split air-to-air HP).

Lieve: do we need a single zone residential building too?

Valentin: the model can be easily changed to one zone for the building part, but it is not that straightforward for the production side.

Filip: a single-zone residential building could be interesting as a test case for MPC developers, since it’s simple such that the barrier to step in is rather small.

David: an apartment may be a real one-zone case.

Fillip: we could use the BEST-test as building envelope, and add an energy system.

Krzysztof: we have a validated modelica model of a zone in our building with non-stochastic occupancy already.

Roel: I also believe that having a single zone model is relevant for dwellings and testing.

Lieve: Michael also wanted to have a one zone office building, was it for any kind of standard?

David: around 17% of commercial buildings use this single-zone fashion. The standard that you are referring to may be the ASHRAE guideline 36. This could be taken as the single-zone office building, equipped with VAV.

Lieve: OK, we agree. And for the single zone residential building we use the BEST-test building envelope and we (KU Leuven - Filip) can develop the hydronic system while the air system can be added by someone from the US?

David: yes we can do that.

4.1.2 Office buildings

Valentin presents the 3500 m2 8-zones building, equipped with gas-fired micro CHP, gas boiler, radiators and ventilation unit. The CHP produces electrical and thermal power.

Lieve: we would like to limit the scope in this initial phase to thermal aspects.

Lieve: this is a very complex model. Let’s jump to a less complex office model first.

Iago presents the INFRAX building has a hybrid emission system using concrete core activation and AHU+VAV. The production system is a single geothermal HP unit. Cooling is provided by a cooling tower and geothermal passive cooling. The main weakness is that the initialization time is quite long. Occupancy will be included using CO2 sensors.

David presents a 5-zones small office using VAV with idealized production. It has two versions of control: according to ASHRAE and according to guideline 36. The simulation time is around 2 hours. The model uses the typical office building occupancy schedules.

Draguna: the one that I presented in Berlin is a large commercial office, which is a prototype of what is used in the US. It is not exactly an extension of David’s model. The heating and cooling systems are not idealized, but modelled in detail. This emulator would be very useful to test the scalability of MPC.

Lieve: I would put this last one together with the SolarWind emulator from Filip (which is very hybrid in production and emission) as the most complex models of the BOP-test which both allow testing scalability. So, we still need a simple one for the commercial sector.

Krzysztof presents the teaching building for which different models can be developed:

• complex model with around 100 zones, ventilation control using CO2 measurements and hydronic heating. We know exactly how many people are in the building and even specified for some specific zones.
• single zone model. The main weakness is that the model is not developed yet but we are willing to start developing the models in modelica.

Lieve: it would be nice if you could provide a single zone commercial building model. We still need a single zone office building with hydronic system. For the simple multi-zone building we could use both, Valentin’s and Iago’s or limit the scope to thermal systems at this stage.

Filip: perhaps at this point it is better to confine the scope so rather use only INFRAX building for multizone hydronic.

4.2 SELECTION OF EMULATOR MODELS

The following table summarizes the selection of emulator models from the discussion:

4.3 AGREEMENT ON FURTHER STEPS

Lieve: next steps are

• to define the requirements (interface definition and how to define inputs and outputs in the toolchain? Adjustments of models may be needed) and template for the emulator models. David and Filip will gather the people from the toolchain working group to discuss these requirements and template. As soon as a first draft is ready, the next skype meeting will be planned.

• To check whether we find a candidate (from the US) who can develop a multi-zone residential building emulator with all-air system.

4.4 OTHER ISSUES

Filip: how to distribute the code? Will we use IDEAS for the BEST-test? I would propose to use only one library. I will make a hydronic system with IDEAS and that should be easy to adapt for the air system.

David: ok.