You are on page 1of 17

PROJECT SYD118S 2024: Multi-agent model of

an industrial power consumer

1.8k 15

MODEL DESCRIPTION

Industrial enterprises, as a rule, are large energy consumers and the calculation of
electricity costs differs somewhat from household ones. The fee is charged for
electricity every hour at different prices. In addition, it is also necessary to pay for
power during peak hours in the region. Thus, the cost of electric power is directly
dependent on the power consumption distribution during week. To minimize the
electricity costs, it is necessary to optimize the consumption of electric power from
the external grid.

1
Possible solution is to add a solar and wind generation utilities to the power feed
from external grid. Energy generated by wind turbine and solar panel are first stored
in the battery and from battery energy comes to the manufacturing plant.
However, limited quantities of energy can be accumulated in batteries.
Therefore, a balance should be maintained between the produced and consumed
electricity. During periods when battery storage is exhausted, electric power is fed
from an external power grid, otherwise the electricity is accumulated in the battery,
and then used when required.

Internal power generation block

Wind and solar generators are the agents of the generation block. The generated
power is determined by climatic conditions: wind speed and direction determine the
amount of power generated by wind turbine, cloudiness and day length are
controlling factors for the solar panels.

Figure 1 Illustrative Example for wind turbine and solar panel statecharts.

2
Figure 2 Another illustrative example how wind turbine model can look like. Here
wind turbine is modelled by a statechart and action chart.

Wind speed and corresponding generated power are given by table function
Wind_Table.
Illumination of PV panel at different time of the day is given by table function
Illumination_PV.

With insufficient wind speed or low light, the wind turbine or solar panel are idle or
generate a small amount of electric power for production needs. In this case energy
supply from external grid is required. Like any other equipment, power generating
installations (turbine and solar panel) need maintenance and repairs. These
activities are carried out on a specific schedule, which is also implemented in the

3
model. The service schedules are given by table functions service_Turbine and
service_PV_panel correspondently. The technical maintenance (service) is carried
out once in 2 weeks’ time. The service period for wind turbine is 5 hours; for solar
panel is 2 hours.

Figure 3 Illustrative Example of battery charging/discharging model

Thus, the behaviour of generating agents in the model is described by agent-based


models (statecharts), and system dynamics model. The parameters of turbine and
PV panel, as well as battery are included in the model. The weather conditions (wind
speed, sunlight illumination) are also given as table functions.

4
Wind speed and turbine power
Wind velocity is based on experimental data

Figure 4 Wind velocity - experimental measurements

The variation of wind velocity can be approximated by normal distribution


where mean value mu = 9 m/s, variation sigma = 3.8 m/s

Wind velocity
Cumulative normal distribution

1.2

0.8
Axis Title

0.6

0.4

0.2

0
0 5 10 15 20
Velocity, m/s

ANYLOGIC has a variety of probability distribution functions.


Normal distribution formula: normal(double sigma, double mean).
In our case normal(3.8,9)
5
Turbine power is non-linear function of wind velocity, but can be approximated as
given in the Figure.

Figure 5 Turbine power curve


Where Pr = maximum turbine power, P is current turbine power.

YOU CAN CHOOSE YOUR OWN MAXIMUM TURBINE POWER Pr!

6
Battery charging table function
The realistic model of battery charging should include dependence of charging rate
on charging current, environment temperature, and battery electrochemical
processes, linking current, voltage, and amount of stored energy. Since only an
approximate model is used, we neglect the peculiarities of the accumulation of
electricity, believing that both a complete battery discharge and an unlimited
accumulation of energy are possible. The simplified model for battery charging is
used in this project. This model links Battery charging rate to the instant battery
charge. This dependence is expressed as a table.
Battery charging rate (SOC%/hour) as function of instant battery charge (SOC%) is
given as Table function. The state of charge (SOC) is a measurement of the
amount of energy available in a battery at a specific point in time expressed as a
percentage. When battery is fully charged SOC=100%. Charging rate depends on
amount of stored energy. The amount of energy in the battery is calculated as
SOC*Pmax, where Pmax has units [Watt*hour].

YOU CAN CHOOSE YOUR OWN VALUE OF Pmax for your battery!

7
Figure 6 Battery charging rate as function of instant battery charge

The Table data are given in Appendix and in attached EXCEL file.

8
Photovoltaic panels
There are several solar panels installed in your plant.
The total amount of energy generated by all panels, is given in the Table as
example.
Example: Illumination during the 24hours period and generated power, Watts
Table 2 Photovoltaic panel data (to be used in ANYLOGIC as Function Table)

YOU CAN CONSTRUCT AND USE YOUR OWN TABLE for the distribution of
illumination during the day and generated solar power.
9
Energy Consumption block (Manufacturing plant)

The production shop is a consumer of electric power; therefore, it is the main agent
of the consumption block. In a real factory, such shops may be a few dozen.
The production starts by receiving the order – how many details are required to be
manufactured.
The agent Order defines the number of details to be manufactured, and determines
the sequence of operations on a particular machine, as well as the duration of these
operations.

It is allowed that some number of machines (agents) of 3 types is installed in the


shop. Each type of machine is designed to perform only one operation. The
operation has certain the duration and the power consumed by the machine during
machining.

The parameters of the agents are given in Table.


Table 1 Parameters of machines (agents)
# Agent Power, Watt Operation
duration, hours
1 Machine of Type 1 5000 4
2 Machine of Type 2 8000 1.5
3 Machine of Type 3 9000 3

The sequence of operations is set at the input to the process by the agent Order,
arriving at the shop floor every 6 hours.

10
Simulation of the workshop activity, and, consequently, the electric power
consumption is set using the Process Modeling Library.

Figure 7 Illustrative example: workshop model

Agent Order - is an object of technological process. It defines number of details


coming to the Source block. And it defines which Machine Type (1, 2 or 3) must be
used to manufacture the ordered detail.
There are three kinds of operations, one for each type of machine. Production
machines are the main consumers of electricity in the shop, as well as the resource
of the technological process. It is assumed that three different types of machines are
installed in the shop, two of each type - a total of 6 units. Each type of machine is
represented as a population of agents (each block on the discrete-event model is a

11
specialized agent). Machine behaviour is described by interactions between
discrete-element blocks (queue, delay, service, resources…) and some parameters.
The machine, when turned on, can be “seized” for the operation or run idle. It is
important to note that when modelling the work of the machines in the workshop, the
enterprise’s work schedule has to be taken into account. The working day starts
from 8:00 and ends at 17:00, and there is a lunch break from 12:00 to 13:00. The
electricity consumption in the workshop must be recorded and displayed on the
graphs in the model.

Model building and simulation – WHAT DO YOU NEED TO DO

Develop a model which simulates generation and distribution of electric power in a


manufacturing plant. Simulate and observe the plant operation. Make conclusion
about energy savings by using internal energy generation (solar panel and wind
turbine).
All panels of the simulation model must have:
text fields showing energy generated by solar panel, by wind turbine, energy to
charge battery, energy supplied from battery to the workshop, energy supplied from
external grid, graphs and sliders for controlling system parameters (if any).
Please see the possible model interface
https://cloud.anylogic.com/model/64c8ca21-2bc9-49bf-9b8c-
27d0ea7c0e71?mode=SETTINGS

CONCLUSION

Multi-agent method is quite effective for the modeling of the power


generation/consumption task. With the help of the developed model, it should be
possible to analyze the consumption or generation of electricity playing the “what if
...?” scenario, and identifying the power consumption peaks.

12
APPENDIX A
Battery charging rate (SOC %/hour) vs battery instant charge (SOC %)
y=d(SOC %)/d(Time,
x = SOC % hours)
0 5
0.01 5
0.1 5
0.260416667 5
0.520833333 5
1.041666667 5
1.5625 5
2.083333333 5
2.34375 5
2.864583333 5
3.125 5
3.645833333 5
4.166666667 5
4.6875 5
4.947916667 5
5.208333333 5
5.729166667 5
6.25 5
6.510416667 5
7.03125 5
7.552083333 5
8.072916667 5
8.333333333 5
8.854166667 5
9.114583333 5
9.635416667 5
9.895833333 5
10.41666667 5

13
10.67708333 5
11.19791667 5
11.71875 5
11.97916667 5
12.5 5
13.02083333 5
13.28125 5
13.80208333 5
14.0625 5
14.32291667 5
14.58333333 5
15.10416667 5
15.625 5
15.88541667 5
16.40625 5
16.92708333 5
17.44791667 5
17.70833333 5
17.96875 5
18.48958333 5
19.01041667 5
19.53125 5
20.05208333 5
20.3125 5
20.57291667 5
21.09375 5
21.35416667 5
21.875 5
22.13541667 5
22.39583333 5
22.65625 5
22.91666667 5
23.17708333 5
23.2 5
23.69791667 5

14
24.21875 5
24.73958333 5
25.26041667 5
25.52083333 5
25.78125 5
25.8 5
26.30208333 5
26.82291667 5
27.08333333 5
27.34375 5
27.5 5
27.60416667 5
27.86458333 5
28.125 5
28.38541667 5
28.90625 5
29.16666667 5
29.6875 5
29.8 5.000805412
29.94791667 5.022867117
30.46875 5.088916454
30.5 5.195220318
30.72916667 5.373557485
30.98958333 5.690431053
31.25 6.150097387
31.51041667 6.719081672
31.77083333 7.392233128
31.8 8.193314123
32.29166667 9.142645275
32.55208333 10.20166542
33.33333333 11.32616247
33.59375 12.45272564
34.11458333 13.56363579
34.63541667 14.71805517
35.15625 15.88783776

15
35.67708333 16.99857795
36.19791667 18.00624862
36.71875 18.97516395
37.23958333 19.89636793
38.28125 20.65117191
38.80208333 21.19850555
39.84375 21.61556899
41.14583333 22.11455672
41.66666667 22.80312003
42.70833333 23.68439161
43.22916667 24.716107
44.01041667 25.88289877
45.05208333 27.24599555
45.57291667 28.75183311
46.35416667 30.22042857
46.875 31.44912268
48.17708333 32.36335634
49.21875 33.02531315
50.26041667 33.51833662
51.82291667 33.81082595
52.86458333 33.87244845
54.42708333 33.86726774
55.46875 34.0068048
56.51041667 34.33816204
57.03125 34.71941763
58.07291667 35.05267057
59.11458333 35.35973433
60.67708333 35.70292885
61.97916667 36.08663846
63.28125 36.33416526
64.32291667 36.31459804
65.625 36.11097954
66.66666667 35.81100948
67.96875 35.40571145
69.27083333 34.86006755

16
70.83333333 34.18898853
72.39583333 33.4588957
73.17708333 32.75005208
74.47916667 32.07141029
76.04166667 31.31310157
77.08333333 30.48766746
77.86458333 29.69213969
79.16666667 28.87142903
79.6875 27.95562899
80.46875 27.00249987
81.77083333 26.1358597
82.29166667 25.27543834
83.07291667 24.27307002
84.375 23.15062241
85.15625 21.86142711
86.71875 20.37523186
88.02083333 18.7184817
89.32291667 16.89543363
89.58333333 15.00421345
90.36458333 13.18107434
91.40625 11.50096716
92.70833333 9.931804152
93.48958333 8.449549281
94.79166667 7.110880133
95.3125 5.937581412
95.83333333 4.878955348
96.61458333 3.854565513
97.13541667 2.862735382
97.39583333 1.979359572
98.17708333 1.253438898
98.69791667 0.701182388
99.21875 0.301359368
100 0.1
110 0.1

17

You might also like