4a. Tutorial - Optimal Unit Commitment

Introduction

In this tutorial, you will learn how to use backbone for optimizing the unit commitment in an energy system. Information on the theoretical background of optimization and unit commitment you can find in the videos LINK.

Backbone is an open source energy system optimization model, developed by the VTT in Finnland (see https://gitlab.vtt.fi/backbone/backbone and [1]). Backbone is programmed in GAMS. You can build a model of your energy system and then use backbone to find the cost optimal solution of your system.

Note: The case study works best on Windows machines. However, it also can be completed on Mac or Linux machines. The only drawback here is, that the automated export of results to Excel does not work. To manually export data to Excel, you must use the export button in GAMS studio (more on that later on).

[1] Helistö, N.; Kiviluoma, J.; Ikäheimo, J.; Rasku, T.; Rinne, E.; O’Dwyer, C.; Li, R.; Flynn, D. Backbone—An Adaptable Energy Systems Modelling Framework. Energies 2019, 12, 3388. https://doi.org/10.3390/en12173388

System Overview

In this tutorial, we will continue to use the energy system of Tudo and Rubbo. Unfortunately, the OCGT plant is broken down and cannot be used anymore. To run an energy system optimization with backbone, we have to perform the following three steps:

1. Fill in the backbone input file in excel.
2. Run backbone.
3. Display and interpret results.

Fill the backbone input excel

The following tasks aim to help you with filling in this input data file. First, you should download the template file "Input data Tutorial unit commitment" from moodle and open it in Excel or LibreOffice. In this part of the tutorial, we will only fill in the blue sheets. Be careful to use the exact names for girds, nodes and unit as they are defined in the yellow sheets.

p_gn

First, we will take a look at the table p_gn. Here, nodes are connected to the grids they belong to. Write all nodes in the column “node” and write the corresponding grids in the column “grid”. The parameter “nodeBalance” determines, whether input and output must always be equal for the node. For all nodes, where energy balance is required, write 1 in the column “nodeBalance”, for all other nodes, write 0. Hint: electric nodes always require nodeBalance.

p_gnn

In the sheet p_gnn connection between nodes are defined. Fill in all connection between nodes with a transfer capacity of 400 MW and a transfer loss of 0%. Lines must be made available by adding a 1 in the column “availability”.

p_gnu_io

p_gnu_io gives information about units in connection with their nodes and grid.

1. Fill in all units with the nodes and grids they are connected to. If a unit is connected to multiple nodes, it must also have multiple entries in p_gnu_io.
2. If a unit gives energy to the corresponding grid, write “output” in the column “input_output”. If it takes energy from this grid, write “input”.
3. Write 1 in the column “conversionCoeff” for every unit.[1]
   The following information shall only be added for output units to avoid double entries:
4. unitSize equals the size of one unit. As all our plants only consist of one unit, unitSize equals capacity.
5. Fill in capacities and variable costs (vomCosts). Use the following table for the values:
   

   	Lignite powerplant	Hard coal powerplant	CCGT powerplant
   Capacity (MW)	240	240	180
   Variable costs (€/MWh)	30	40	60

   
   

p_unit

p_unit gives unit specific information which do not depend on the corresponding node. Fill in all units and make them available. In following table, you find information on the effectivity and minimum load parameter.

In this case, we will assume the same efficiency for the complete operating range. Fill in all parameters with the information from the following table:

ts_influx

In ts_influx, influxes for nodes can be added. In our energy model, we have consumers at Tudo and Rubbo. In this assignment we will only consider consumers in Rubbo. Each timestep equals one hour. Load must have a negative sign, as this is energy, that drains from the node.[1] Use the following load curve:

Run Backbone

Open GAMS studio. Open the file “backbone.gms”, which is included in the backbone folder. In GAMS studio, you have a command line (marked with the big red arrow in the following picture) and a green start button (also marked in the following picture.

In the command line, you must type:
--input_file_excel="path/to/your_excel_file.xlsx" --tutorial=1

path/to/your_excel_file.xlsx must be replaced with the path, that leads to the excel input file you filled in the previous tasks. Make sure, the path is correct and leads to an existing backbone excel file, otherwise backbone will not work. It should also not contain spaces or special characters like ö/ü/ß etc.

For all tutorial runs, you have to use the flag --tutorial=1. For the case study runs, you must not use that flag.

Common errors

If backbone does not run sucessfully, there can be multiple reasons. On this page, we show you the most common ones. To see error messages, you have to look into the process log of GAMS Studio (on the right). Always start with solving the first error, that occured, because later errors might just be resulting from this first error.

Display and analyize results

After sucessfully running backbone, you will find a file named "results.gdx" in the folder backbone/output. This is the main output file of backbone. It contains all results from the optimization. You can open it with GAMS studio to look into the results. On the left, you will see a table containing all result tables. Next to the tables, there is a description of the results table (maybe you have to enlarge this part of the window to see this information). If you click on a result table, it will be displayed on the right. There are a lot of very small numbers in the results. This is due to the solving algorithm. You can ignore every value, that is smaller than 1.

The gdx file provides a good overview over all results, however, it is not optimal to analysis results. Therefore, the most important tables are also saved as excel files. This tables are:
r_gen_gnuft - energy generation from a unit (MW)
r_invest_unitCount_u - number of invested units
r_state_gnft - node state at timestep t (MWh)
r_qgen_gnft - dummy energy generation/consumption at timestep t (MW)
r_cost_realizedCost - total realized system cost over the simulation (MEUR)

For the analysis of the first tutorial, we will only need r_gen_gnuft. The meaning of the other tables will be explained in further tutorials.

Note: On Mac and Linux machines, no Excel files are exported automatically. To manually export the sheets you need, use the Export button in GAMS studio.

Task: Plot your results

Solution: Plot your results

• Graphs should always contain axes labels and units
• Stacked area graphs are more insightful, if the units are sorted in a senseful way (e.g. base load plants at the bottom, peak power plants on top)

Task: How much lignite is used?

Task: Which case did you model?