Civil Engineering Reference
In-Depth Information
In accordance to the split-range functions defined by Figs.
5.23
and
5.24
, at each
sample time, the split-range control provides a value, for the fan velocity,
V
Fan
,
comprised inside the range [0-100%]. Nevertheless, as pointed out previously, in
general, industrial fancoils are implemented through discrete on/off actions. Hence,
it is necessary to transform the computed fan velocity in such a way that the control
action can be provided through a discrete actuator. The usual way of doing this is
using a PWM signal (Salsbury
2002
). This PWM signal is characterised by having
a cycle period,
C
, of 8 s, which is equal to the sample time of the PI controller,
an internal sample time of a second, a maximum time for which the control signal
is equal to 100%,
T
on
=
8
s
, and a minimum time for which the control signal is
equal to 0%,
T
off
=
0 s. Besides that, in order to avoid stress in the fancoil unit, the
fancoil is not connected unless the fan velocity control signal is greater than 35%
which is equivalent to a global control signal
u
of about 15%, see Figs.
5.23
and
5.24
.
Moreover, this constraint has been included in the anti-windup scheme.
5.4.3 Results
This section presents the real results obtained from the proposed hierarchical control
strategy when it has been tested in a real room of the CDdI-CIESOL-ARFRISOL
building during February, 2013. Therefore, it can be considered that this section
expands the work presented in Sect.
5.3
. However, there exist some important
improvements that justify the development of this work. More specifically, in the
strategy presented in Sect.
5.3
, outside disturbances associated to the outdoor cli-
mate, such as direct irradiance,
I
dr
, and outside air temperature,
T
a
out
, and some
uncontrollable indoor variables, like the number of occupants inside the room,
N
p
,
were not considered by the proposed optimisation algorithms since LTI models were
used. Moreover, in this work a complete inside building climate model based on
variables, outside and indoor disturbances, is used. In addition, from the energy
efficiency point of view, the principal difference between them is that, in the first
approximation, Sect.
5.3
, the water flow valve was always set to 100%, while in the
controller presented in this approach, the use of water flow is minimised as much as
possible. Hence, the main control objective was to find out if the developed strategy
was able both to maintain thermal comfort taking into account different disturbances
and, to reduce the energy consumption derived from it.
Therefore, several tests have been carried out during different working days inside
the selected room. More specifically, each test has a total duration of approximately
five hours, from 9 a.m to 14 p.m., that is, the typical morning schedule of an
office. The tests were performed with a sample time of 240 s in the upper layer,
that is, the PNMPC optimiser. This sample time value was chosen based on the
desired closed loop time constant defined by the PI parameters,
ʸ
cl
=
80 s, that is,
240 s
×
ʸ
cl
which is 95% of the new steady-state value that the PI must reach.
As mentioned previously, the sample time of the lower layer, the fancoil MISO
=
3
