Civil Engineering Reference
In-Depth Information
y
1
y
2
y
3
y
4
setpoint
22
20
18
10
20
30
40
50
60
70
80
90
time [min]
3
u
1
u
2
u
3
u
4
2
1
0
10
20
30
40
50
60
70
80
90
time [min]
10
solar plant temperature output
8
6
10
20
30
40
50
60
70
80
90
time [min]
solar plant flow output
8
6
4
10
20
30
40
50
60
70
80
90
time [min]
Fig. 6.2
Subsystem outputs and controls, solar plant temperature and water flow with independent
MPC controllers.
Source:
As a courtesy of the authors (Scherer et al.
2014
)
the data exchange between agents. In this case, during the periods of time where the
temperature of the input flow of chilled water was low enough to supply the demand
of energy of all the subsystems, all controllers provided optimal performances. This
can be observed in Fig.
6.2
in the intervals 5-15min and 50-68 min approximately.
Notice that the available energy is in this case proportional to the fluid temperature
since the input flow is almost constant during the whole experiment. On the other
hand, when the water temperature increases, for example between 20 and 50min,
temperature of subsystem 4 cannot be maintained in the setpoint. Because of the
spatial distribution of the subsystems, when the energy is not enough for all, the
last agent is the first to be harmed. A given subsystem will be harmed only when
there is no flow passing to the next subsystem. Clearly, a control strategy based on
independent controllers is not suitable for this type of network. Fancoil systems can
be operated with independent controllers only when the chilled water produced by
the solar plant is oversized.
In the second test, similar operation conditions as in the previous case were con-
sidered (using the same output reference for all the subsystems) but with a network
of distributed agents enabled. When the DMPC strategy was active, i.e. agents could
communicate, the controlled system presented the results shown in Fig.
6.3
. As can
be seen, the same behaviour as the one obtained with independent controllers was
observed when the allowable energy was enough for all the agents, which means
that all the subsystems can follow their references. The characteristic of dependence
between subsystems appears only in the saturated case, for high values of the input
flow temperature, for example, between 40 and 80min. When this happens, the
advantages of the distributed strategy appears.
