Welcome to the showcase!
Step forward: SPACE | Step backward: SHIFT+SPACE
Chapters: ← / → | Subslides: ↑ / ↓
Overview: Esc
About Draf¶
What is Draf?¶
Draf stands for "Demand Response Analysis Framework" and is an easy-to-use open source decision support tool that enables the economic and ecological evaluation of industrial demand response potential. It is devoloped in Python and uses the power of Gurobi to build and solve multi-objective optimization models of industrial energy systems and production processes to quantify the cost and green-house-gas saving potentials of price based demand response. Target groups are researchers, engineers, and decision-makers of companies in industrial or commercial sectors with some basic knowledge of the Python programming language.
Draf is developed by Markus Fleschutz in a cooperative PhD between Cork Institute of Technology and the Karlsruhe University of Applied Sciences and under the supervision of Dr. Michael D. Murphy and Dr.-Ing Marco Braun. More information can be found on the website.
Why Draf?¶
- The core-idea of demand response is to shift electric demand to hours with cheap electricity prices e.g. when shares of renewable energy sources are high.
- For industrial companies, price-based demand response is an essential and promising approach to reduce both electricity costs and – with increasing relevance – carbon emissions.
- The problem with the quantification of demand response potential is that it requires the modeling of the inventives (here: day ahead market prices and carbon dynamic emission factors) and the flexible entities (batteries, heatpumps etc.) with its most relevant constraints which can be very complex and time consuming.
- Draf solves this problem by making the quantification process of demand response potentials automated and easy-to-use. It provides market data, predefined models, modeling assistance, and visualization tools and helps the user to focus on the important aspects.
Core areas¶
Draf combines several task areas that are important for the quantification of demand response potentials. The most important ones are:
- Easy access to electricity market data.
E.g. if a case study is configured with a specific year and country, Draf downloads the according day-ahead market prices and historic national electricity generation data, preprocesses it, and prepares the real-time prices and carbon emission factors that are needed for the analysis. - Several predefined models.
... of typical industrial energy systems that can easily be parameterized. - Fully automated scenario generation.
In a scenario, only a few parameters are changed. - Several built-in analysis and visualisation tools.
A quick-start showcase¶
We start with a straightforward model. Let's say we have a 100 kWpeak Photovoltaic System (PV), a 1 MWhel Battery Energy System (BES), and given electricity demand and want to know the cost and carbon reduction potential for price based demand response considering a 50 €/kW peak-load price and a day-ahead market price. We can use the "PV_BES"-model of draf:
Schematic depiction of model "PV_BES" with its parameters and variables. Case study setup & scenario generation¶
These few lines of codes import the model PV_BES, initializes a case study, adds a reference scenario, sets the default parameters from the PV_BES-model, changes the user-defined parameters P_PV_CAPx_(existing 'PV power) and c_GRID_peak_(electricity peak price), builds 8 scenarios (2 pricing tariffs with 4 pareto points each), improves the pareto weighting factors so that the pareto points are more equally distributed, sets the model, optimizes it and saves the whole case study.
import draf
from models import PV_BES as mod
cs = draf.CaseStudy(name="ShowCase", year=2017, freq="60min", country="DE")
cs.add_REF_scen(doc="no BES").set_params(mod.params_func).update_params(
P_PV_CAPx_=100, c_GRID_peak_=50)
cs.add_scens([("c_GRID_T", "t", ["c_GRID_RTP_T", "c_GRID_TOU_T"]),
("E_BES_CAPx_", "b", [1000])], nParetoPoints=4)
cs.improve_pareto_and_set_model(mod.model_func).optimize(mod.postprocess_func).save();
Results¶
The pareto curve is the main resulting diagram of the draf-analysis. It plots the total costs versus the total carbon emissions for all scenarios. Scenarios can be grouped according to key-words in order to help identify individual pareto fronts. Here, the pareto-weighting factors are denoted with α.
cs.plot.pareto_curves(groups=["REF", "TOU", "RTP"], label_verbosity=4)
More plotting¶
Parameter (input)¶
Parameter scalars for all scenarios¶
This table shows all parameters of all scenarios with its values, unit and description.
cs.plot.table("p")
For all non-scalars entities, the describe function gives an overview of one ore many scenarios:
cs.plot
sc = cs.scens.sc1
sc.plot.describe(include_vars=False)
Heatmaps¶
Heatmaps are the perfect plot type to see seasonal and daily patterns of a whole year's timeseries.
sc.plot.heatmap(ent_name="E_dem_T")
sc.plot.heatmap(ent_name="E_PV_T")
sc.plot.heatmap(ent_name="c_GRID_RTP_T")
sc.plot.heatmap(ent_name="ce_GRID_T")
sc.plot.heatmap(ent_name="E_BES_in_T")
sc.plot.heatmap(ent_name="E_BES_out_T")
Variables / Results (output)¶
Scalars for all scenarios¶
This code produces a table of all scalar variables of the model with its resulting values, units, and descriptions.
cs.plot.table("v")
Variables Summary for scenario "sc1"¶
cs.scens.sc1.plot.describe(include_pars=False)
Pareto front¶
cs.plot.pareto_curves(groups=["REF", "TOU", "RTP"])
Sankey¶
The sankey plot shows the yearly energy sums of key energy flows from left (source) to right (cradle). You can hover over to see the numbers:
cs.scens.sc1.plot.sankey(mod.sankey_func)
Features / modules¶
Saving & Loading of case studies¶
At any time, you can save the case study to hard disk like so:
cs.save("test");
... and open it later again:
import draf
cs = draf.CaseStudy(fp="D:/mf/draf/results/ShowCase/2019-12-10_143657_test.p")
Data export¶
Draf is compatible with the powerful xarray-module. Here, parameters and results are comined into one big dataset:
cs.scens.REF.get_xarray_dataset()
Spot market prices¶
Draf provides day ahead spot market prices for several European countries and years.
sc = cs.scens.REF
sc.plot.fig_size = (12,1)
sc.prep.add_c_GRID_RTP_T()
sc.plot.heatmap(ent_name="c_GRID_RTP_T")
Carbon emission factors¶
Draf provides dynamic carbon emission factors for several European countries and years.
sc.prep.add_ce_GRID_T(name="ce_GRID_avg_T", method="average")
sc.plot.heatmap(ent_name="ce_GRID_avg_T")
sc.prep.add_ce_GRID_T(name="ce_GRID_mar_T", method="marginal")
sc.plot.heatmap(ent_name="ce_GRID_mar_T")
Timeseries Preparation¶
Draf provides some useful tools for demand estimation. These are especially useful when only aggregated values are availible.
First, we prepare an electricity demand from a industrial standard load profile "G1" which stands for Business on weekdays 08:00 - 18:00 and an given annual energy demand of 5 GWh/a.
sc = draf.CaseStudy(year=2017).add_REF_scen()
sc.prep.add_E_dem_T(profile="G1", annual_energy=5e6, holidays_province="BY")
sc.plot.fig_size = (12,1)
sc.plot.heatmap(ent_name="E_dem_T")
sc.params.E_dem_T.sum()
Now let's create an heat demand timeseries from the annual energy, the target_temperatur, the heating threshold and the location where the ambient temperature is taken.
sc.prep.add_H_dem_T(annual_energy=5e6, target_temp=20.0, threshold_temp=10.0, location='Rheinstetten')
sc.plot.heatmap(ent_name="H_dem_T")
This is the ambient temperature that was used as basis:
sc.plot.heatmap(timeseries=draf.prep.io.get_amb_temp(year=cs.year, freq=cs.freq, location="Rheinstetten"))
Model templates¶
Draf comes with some out-of-the-box models for different relevant aggregates in the industrial environment that can easily be parameterized. One of the models is "DER_HUT" - Distribted Energy Ressources(DER) and Heat upgrading Technologies (HUT). Here is a schematic:
By the way:¶
is a submodel of the DER_HUT:
Access to Draf-objects¶
In the following, access and preview of draf-objects (CaseStudy, Scenario, etc.) is demonstrated.
Case Study¶
cs
Scenarios¶
cs.scens
Scenario¶
cs.scens.REF
Params¶
cs.scens.REF.params
choose one of them, e.g. P_PV_CAPx_:
cs.scens.REF.params.P_PV_CAPx_
Results¶
cs.scens.REF.res
choose one of them, e.g. C_:
cs.scens.REF.res.C_