Modelling and Analyzing a Renewable Energy System using Simulation Approach

Introduction

From reliability perspective, solar systems are generally robust against failures. Yet from an operation perspective, the system is not stable as energy source depends on weather conditions.

This presentation describes a simulation approach to quantify how the variations of 1. input energy, 2. load demand, and 3. storage usage, affect the availability of output energy.

The Scenario

A solar energy system consists of battery storage, backed up by traditional grid power intended to supply a load continuously. Figure 1 depicts the digital twin representation of such a system. In this article, the energy unit is normalized to some arbitrary unit for simplicity.

Analyzing Renewable Energy System through its digital twin

In this scenario, the digital twin contains the following information:

  1. A schematic diagram defining the flow of energy.
  2. The energy profile of the source: Source (Solar).
  3. The energy consumption profile for the load, Load.
  4. The logic to turn-on backup energy source Grid (Meter), when Storage (Battery) becomes empty.

  5. Resources_Renewable01.png
    Figure 1. The digital twin of the solar system created using AeROS software

    The solar panel can only produce energy between 8:00 to 17:00 (9 hours) daily. The power is averaged over 9 hours.


    Resources_Renewable02.png
    Figure 2. Profile describing the ability of the solar panel to generate power

    The Daily Average Power (units/hour), from 8:00 to 17:00 hours, produced by the solar panel were collected over last 30 days.

    Resources_Renewable_Tbl1
    Table 1. Dataset of Daily Average Power from 8:00 to 17:00 hours express in unit/hour

    The power consumed by the load is 60 units/hour from 8:00 to 17:00 hours, and 30 units/hours for the other time of the day.

    Resources_Renewable03
    Figure 3. Power consumption profile of the load

    Information Sought

    How much power do I have to rely on from the grid supply?

    What is the usage profile of the battery?

    The Approach

    We start with creating an input energy profile for the Source (Solar) node as mentioned in Figure 2.

    The Daily Average Power dataset is fitted with Weibull distribution.

    Resources_Renewable04.png
    Figure 4. The Daily Average Power follows Weibull distribution with beta=7, and eta=100 units/hour

    Resources_Renewable05.png
    Figure 5. The Probability-Weibull and PDF of the Daily Average Power distribution

    Note that, from this distribution, the mean Daily Average Power is 93.3 units/hours.

    Now we are able to complete the power generation profile of the Source (Solar) node, by providing the Daily Average Power distribution.

    Resources_Renewable06.png
    Figure 6. The Source (Solar) node's power generation profile with varying amplitudes following Weibull distribution with beta=7, eta=100 units/hr.

    The mean Daily Average Power generated by solar panel is 93.3 units/hours. The Storage (Battery) node is designed to store this amount of energy over 9 hours, which is 840 units (of energy).

    Finally, the Grid (Meter) is set to standby, and will be activated when the battery is depleted, and reset back to standby when the battery is being charged above, say, 2%.

    Now, the digital twin has been setup with all the necessary information for simulation.

    Resources_Renewable07.png
    Figure 7. Run a 720 hours (1 month) simulation with 1000 executions

    The Results

    For this setting, the simulation generates and stores performance metrics for each execution. The simulation results are the collection of these metrics averaged over 1000 times.

    AeROS software provides an interesting visualization of operating profiles of individual node over the simulated period, as shown here.

    Resources_Renewable08.png
    Figure 8. The operating profile of the last execution of the simulation

    The nodes Load and Grid (Meter) are called Regular Node in AeROS. It can have 3 states: Operating, Standby, and Out-of-Service. Since in this scenario, nodes’ failure behaviours are not defined, they will not fail. The blue area under these nodes represents Flowrate (Power) being “processed” (consumed).

    Storage Node cannot fail by design. The plot for Storage (Battery) shows its charging and discharging profile.

    Profile Node cannot fail by design. Source (Solar) and Load Profile are profile nodes. Each profile node contains an operating profile defined by user. The blue areas under these nodes represents Flowrate (Power) being “processed” (consumed).

    Resources_Renewable09.png
    Figure 9. Regular Node tab of Simulation results dialog. It displays the statistics associated to Regular Nodes

    How much power do I have to rely on the grid supply?

    From the above figure, Grid (Meter) node supplies 4,416 units of energy over a month.

    Resources_Renewable_img1.png

    The Grid (Meter) node is turned-on when the battery is depleted. Although it can supply 60 units/hour, only 30 units/hour are consumed. This is because the battery depletion occurs after daylight, and the Load consumption is reduced to 30 units/hours.

    What is the usage of the battery storage?

    Resources_Renewable10.png
    Figure 10. Simulation results for the Storage (Battery) node

    Over a month of operation, the number of days that the storage is full, is virtually zero.

    The cumulative number of days that the storage is depleted is 5.7 days.

    The battery average storage level is about 18%.

    The total amount of energy consumed by the load is 29,700 units (Figure 9) over a month. The amounts supplied by power grid and solar power are 4,416 units and 25,284 (energy consumed by the Load: 29,700 – energy supplied by Grid (Meter): 4,416) units respectively.

    Supposing 1 unit of energy costs $1, so the monthly power bill would be $4,416.

    If you are planning to double the capacity of the solar panel, how much more can you save from the $4,416? How long does it take to recover your investment? These answers can be obtained by modifying the digital twin, and running a new simulation.

    Conclusion

    This case study demonstrates a statistical approach for quantifying the operational impact of a solar power system based on supply, usage demand variations, and buffering storage design. Cost-benefit analysis, what-if analysis, and design optimizations, can also be performed using this modelling and analysis tool.

    -End-