Title of Test

KIT.TC1.TS1 FlexOffice Thermal MPC

Author / Organization

Alexander Engelmann / KIT

Reference to Test Case

Test Rationale

In this test aims at evaluating the capabilities of a MPC to optimize the system behaviour with respect to the previously defines KPIs and the OF. Specifically, this is lowering the fluctuation in the energy demand and to keep the thermal comfort in the building within an acceptable range. The simulation results are subject to a given set of inputs/disturbances defined in this test specification.

Specific Test System

The system under test is the thermal domain of the FlexOffice system configuration as shown in Figure E.2. The thermal behaviour is described on a floor level with influences shown in Figure E.3.

Test and Output Parameters

Test Parameters:

As described in deliverable D4.1 (Appendix D, page 77), the thermal building model has the form

where are initial temperatures and the states

with a temperature vector for each floor with floors (second floor, first floor and ground floor), whereis the temperature of the concrete-core,  is the zone air temperature,is the wall temperature and is the floor temperature of the ?-th floor. The basement temperature vector is defined as

Furthermore, the inputs are

where are the inputs for each floor with concrete-core influxes and radiator influxes

The disturbances are

where is the outdoor temperature, is the ground temperature both in Kelvin, and  are the solar irradiation acting on the roof and the walls respectively, are internal gains modelling heat influxes from occupation and electrical devices.
The parameters (building dimensions, material parameters etc.) for deriving the matrices ?,? and ?, the machine-readable names and the units for inputs and states are provided as attachment to deliverable D4.1 (data file KIT_Thermal-building_data.zip).

Methods for data synthesis are described in D3.4.

Outputs / Measured Parameters:

The outputs of the models are the simulated wall, indoor and concrete core temperatures, i.e. the states ?(?).

Test Design

The disturbance ?(?) for testing represent a typical week of operation such that the test outcome reflects indoor and wall temperatures of a typical winter week. They are shown in Figure E.4.
The inputs ?(t) are kept constant here and can be found in D4.1.

The test is executed as follows: First one has to set up a model representing the relation between room and wall temperatures. Then, a simulation is started using the given disturbances and fixed inputs described above with different or no controllers.
In a second step, the test criterion from the test case are evaluated with the resulting closed-loop trajectories. A model refinement step might be needed in order to match the required accuracy.

Component Models

Initial System State

All temperatures are initialized with 21°C.

Temporal Resolution

Evolution of System State and Test Signals

Figure E.5 shows typical open-loop trajectories for the given
disturbances and fixed input from D4.1.

Figure E.6 shows typical closed-loop trajectories where a tracking MPC scheme is used to compute optimal inputs. One can see that the KPI of maximizing the comfort (i.e., keeping the indoor Temperatures at 21°C) is satisfied to a very high degree. Small peaks deviating from 21°C come from the absence of cooling capabilities in winter time. A variety of MPC formulations, e.g., minimizing fluctuation, are possible. The  predefined KPIs and the objective function values for the different controller  formulations are compared in a second step.

Source of Uncertainty

Stopping Criteria

There are uncertainties in the weather forecast and also in the model parameters (material data).

-

Storage of Data