This module provides both data and a method to represent key integration challenges of wind and solar power in largescale energyeconomy models.
Cite: Ueckerdt, Pietzcker et al. (n.d.) when using any RLDC data or the described method to implement wind or solar variability.
The main idea of the module is that the variable renewable energies wind and solar are challenging to integrate in power systems because of their variability and the necessity to match load and generation at all times. VRE influence key power system parameters such as peak residual demand, fullload hours of dispatchable power plants, storage requirements and curtailment. Residual load duration curves (RLDCs) contain the information how these key system parameters change depending on the share of wind and solar in electricity generation (Ueckerdt et al., 2015). Using the RLDCs, we describe below how to represent this information in an IAM (Integrated Assessment Model), using a polynomial formulation suited for a nonlinear model (Ueckerdt, Pietzcker et al., n.d.). The results of different implementations of the RLDC data in several IAMs as well as other approaches to represent variability are evaluated in (Pietzcker et al., n.d.).
Files required to use the module
All files required to use the module are available here.
Residual load duration curves
Data
This module contains the RLDCs for the eight world regions SubSaharan Africa, China, Europe, India, Japan, Latin America, Middle East and North Africa, and the USA.
ADVANCE_VREintegration_RLDCdata.xlsx
ADVANCE_VREintegration_monthlyRLDCdata.xlsx
Data Description:
 The full description of the RLDC data and its calculation can be found in (Ueckerdt, Pietzcker et al., in this issue), a preprint of which can be accessed here [please insert link to preprint on the ADVANCE page. The pdf to upload on the ADVANCE page can be downloaded at https://www.pikpotsdam.de/members/robertp/ueckerdtetaldecarbonizingglobalpowersupply.pdf ]
 Each row contains the information for one assumed gross share of wind and PV, both of which are varied in steps of 10%points. “Gross” in “gross VRE/wind/PV share” (columns CE) means that curtailment and storage losses have not been deducted. “Net” in “net VRE share” (column H) means that curtailment and storage losses were deducted
 The RLDCs were calculated by taking the load time series, subtracting the rescaled wind and PV time series from it, and reordering from highest to lowest. The values are rescaled so that 100% is the peak load in a system without wind and PV
 Column F shows if in this scenario the DIMES model was allowed to invest into shortterm storage to reduce total system costs, or not.
Overview of key system parameters for each RLDC:
ADVANCE_VREintegration_RLDCdata.xlsx → worksheet: “overview_results”
 “Total Storage Investment costs in a system with 100kW peak load”: as DIMES sees increasing shortterm storage costs with increasing storage deployment, we report also the total investment cost of the installed storage capacity and reservoir.
 “Residual peak demand after VRE+ storage”: even at 0% wind and solar contribution (row 6), it would be costoptimal for the model to invest in shortterm storage to reduce peak demand slightly”
Residual load duration curves:
ADVANCE_VREintegration_RLDCdata.xlsx → worksheet: “detailed RLDCs”
 Columns L to AHC give the residual load in % of peak demand in 10hbins to reduce data size
 In the “monthly” file, the reordering from highest to lowest residual demand was done for each month, not for the full year.
How to use the module
Data can be used to calibrate integration challenges for wind and solar, including
 peak residual demand after wind and solar feedin (w/ and w/o shortterm storage)
 curtailment (w/ and w/o shortterm storage),
 shortterm storage requirements,
 longterm storage,
 and fullload hour reduction of dispatchable power plants
An example for how this information can be implemented in an IAM is described below. Other examples such as the representation of RLDCs in a linear IAM (Johnson et al, n.d.) are compared in (Pietzcker et al., n.d.).
Residual load duration curves – implementation example in a nonlinear model (REMIND)
Data
The module contains:
 The thirdorder polynomial fit of the values of the following key system parameters for eight world regions SubSaharan Africa, China, Europe, India, Japan, Latin America, Middle East and North Africa, and USA:
 peak residual demand after wind and solar feedin
 curtailment
 shortterm storage requirements
 and fullload hour reduction of dispatchable power plants
 Figures of the abovementioned polynomial fits compared to the raw data from the RLDCs:
ADVANCE_VREintegration_RLDCpolynomial_fit_figures.zip
Description:
 The full description of the nonlinear implementation of RLDCs via load boxes and storage and curtailment equations can be found in (Ueckerdt, Pietzcker et al., n.d.), a preprint of which can be accessed here [please insert link to preprint on the ADVANCE page. The pdf to upload on the ADVANCE page can be downloaded at https://www.pikpotsdam.de/members/robertp/ueckerdtetaldecarbonizingglobalpowersupply.pdf ]
 The dependence of key electricity system parameters on the gross share of wind and solar in electricity generation can be represented by thirdorder polynomials
 The parameters are
 Curtailment (“curt” = curtailment as share of demand; “curtShVRE” = curtailment as share of gross VRE production)
 shortterm storage (“shtStor” = storage capacity, “STSRes2Cap” = storage reservoir per capacity in hours, “STScost” = total storage investment costs in $/W peak demand of the electricity system)
 5 load band heights (peak residual demand plus 4 load boxes of 700, 2200, 4400 and 8760 FLh). For implementation reasons, the load band height parameters in the Excel file are normalized such that the sum of the four RLDC boxes is the share of total load that needs to be covered by nonVRE generation (i.e. 1 for a system with no VRE). Accordingly, the average capacity factor of load is 1/residual peak load at 0% VRE.
 The plots visualize the polynomial fits, with coloured surfaces representing the polynomial fits, compared to the raw data from the RLDCs which is displayed with blue +

The fits were calculated with higher weights on low VRE shares, and lower weights on high VRE shares. This was done to a) balance the higher influence of high VRE shares that arises from combinatorial reasons, and b) represent the larger uncertainty about system configurations at high VRE shares.
How to use the module
Any or all of the polynomial equations can be implemented in an energyeconomymodel to improve the dynamic representation of the effects of increasing wind and solar share on
 peak residual demand after wind and solar feedin
 curtailment
 shortterm storage requirements,
 and fullload hour reduction of dispatchable power plants
References
 Johnson, N., Strubegger, M., McPherson, M., Parkinson, S.C., Krey, V., Sullivan, P., n.d. A reducedform approach for representing the impacts of wind and solar PV deployment on the structure and operation of the electricity system. Energy Economics. doi:10.1016/j.eneco.2016.07.010
 Pietzcker, R.C., Ueckerdt, F., Carrara, S., Sytze de Boer, H., Després, J., Fujimori, S., Johnson, N., Kitous, A., Scholz, Y., Sullivan, P., Luderer, G., n.d. System integration of wind and solar power in Integrated Assessment Models: A crossmodel evaluation of new approaches. Energy Economics. doi:10.1016/j.eneco.2016.11.018
 Ueckerdt, F., Brecha, R., Luderer, G., Sullivan, P., Schmid, E., Bauer, N., Böttger, D., Pietzcker, R., 2015. Representing power sector variability and the integration of variable renewables in longterm energyeconomy models using residual load duration curves. Energy 90, Part 2, 1799–1814. doi:10.1016/j.energy.2015.07.006
 Ueckerdt, F., Pietzcker, R., Scholz, Y., Stetter, D., Giannousakis, A., Luderer, G., n.d. Decarbonizing global power supply under regionspecific consideration of challenges and options of integrating variable renewables in the REMIND model. Energy Economics. doi:10.1016/j.eneco.2016.05.012