Civil Engineering Reference
In-Depth Information
used to calibrate the value of the unknown parameters, whereas similar tests are
used to validate the chosen parameter values. Secondly, a range around the estimated
value is chosen (approximately a 25% of the estimated value) and then, the model
is calibrated through the use of genetic algorithms as shown in Sect.
4.3.2
.Inthe
following sections more information about the proposed methodology is provided.
4.3.1 Unknown Parameters Initial Estimation
To obtain an estimation of the values associated with the unknown parameters
,
it is necessary to perform different experimental tests for each involved process.
However, in some cases, these experiments are not easy to perform, since some of
them can have a high economic cost or present a long duration. In addition, as the
meteorological sensors and different actuators state data are available, it would be
feasible to use these data to calibrate the room climate model, without losing the
physical meaning of the involved processes. In order to simplify this problem, some
assumptions have been taken into account:
Ψ
•
Data of the climate variables to model, the disturbances and the actuators state
are measured and stored in a database, hence the main parameters estimation and
calibration problem can be divided into three submodels calibration processes:
(i) indoor air temperature, (ii) indoor humidity ratio and (iii) indoor CO
2
concen-
tration.
•
Some of the processes involved in each of the balance equations do not have any
influence on determined periods of a day or they are not coupled (e.g. the presence
of people into the room or the state of the HVAC system). Thus, all the parameters
of a single submodel do not have to be estimated simultaneously.
•
Some processes have to be modelled in different forms based on determined situ-
ations, such as the convection process between the walls and the indoor air tem-
perature since it depends of laminar or turbulent air flows.
•
Some controlled experimental tests, such as step and impulse response ones, can
be performed in the room to estimate the parameters associated with actuators.
In the case of the model developed in this chapter, to obtain a preliminary approxi-
mation of the unknown parameters values
, a group of experimental tests have been
performed all around the year. Each of these tests is composed by several experiments
in which each contains several days of a certain season of the year. Furthermore, to
select a set of appropriate experimental tests the room has been divided into elemen-
tal systems which can be insulated from the others. More specifically, in a typical
office room the main elements could be the surroundings surfaces (walls, ceiling and
floor), the HVAC system, the windows with their shading devices and the influence of
people, lighting and electrical appliances. After that, and taking as reference values
the guidelines provided by ASHRAE (
2009
), the initial values of the unknown para-
meters have been estimated using a brute force sequential search algorithm. Hence,
Ψ
