OPERA: a New High-Resolution Energy System Model for Sector Integration Research

Abstract

This article introduces and describes OPERA, a new technology-rich bottom-up energy system optimization model for the Netherlands. We give a detailed specification of OPERA’s underlying methodology and approach, as well as a description of its multiple applications. The model has been used extensively to formulate strategic policy advice on energy decarbonization and climate change mitigation for the Dutch government, and to perform exploratory studies on the role of specific low-carbon energy technologies in the energy transition of the Netherlands. Based on a reference scenario established through extensive consultation with industry and the public sector, OPERA allows for examining the impact of autonomous technology diffusion and energy efficiency improvement processes, and investigating a broad range of policy interventions that target greenhouse gas emission abatement or air pollution reduction, amongst others. Two characteristics that render OPERA particularly useful are the fact that (1) it covers the entire energy system and all greenhouse gas emissions of the Netherlands, and (2) it possesses a high temporal resolution, including a module for flexibly handling individual time units. The simulation of the complete Dutch energy system and the ability to represent energy supply and demand on an hourly basis allow for making balanced judgments on how to best accommodate large amounts of variable renewable energy production. OPERA is therefore particularly suited for analyzing subjects in the field of system integration, which makes it an ideal tool for assessing the implementation of the energy transition and the establishment of a low-carbon economy. We outline several near-term developments and opportunities for future improvements and refinements of OPERA. One of OPERA’s attractive features is that its structure can readily be applied to other countries, notably in Europe, and could thus relatively easily be extended to cover the entire European Union.

Appendix

The allocation of hours into time-slices relies on the idea that hours with a similar character can be grouped together, without significantly affecting the level of insight in the final results. In this context, we use the word “character” to designate all features of a certain hour in terms of energy supply and demand. Table A1 lists the features explicitly modeled in OPERA to define the character of a certain hour, along with a short description and an explanation for their relevance.

Table 5 Features used to characterize time-slices

For features 1 through 3, the user is free to choose what subdivision to apply. For example, it is possible to group spring and autumn into one single intermediate season (feature 1), or to divide the 24 h of a day into 12 day-hours and 12 night-hours without further subdivision (feature 3). Features 4 through 8 can be specified using offset parameters. For example, with reference to feature 4, the user can provide a threshold above which the hours should be characterized as peaks.

Within this framework, the user is able to choose the desired number of time-slices and devise a suitable mix of features to characterize them. For example, it is possible to only choose features 1 and 3 from Table A1 and divide the hours statically into six time-slices:

• a. Intermediate day

• b. Intermediate night

• c. Summer day

• d. Summer night

• e. Winter day

• f. Winter night

Alternatively, the user can devise a more sophisticated algorithm to define the character of each time-slice by taking into account all features at the same time. An example of such an algorithm, also based on 6 time-slices, might be:

• Allocate all extreme hours (feature 8) to time-slice 1;

• Single out all peaks (feature 4) and valleys (feature 5), and allocate the peaks that are close to a future valley (feature 6) into time-slice 2 and all other peaks to time-slice 3; Further allocate all the valleys that are close to a peak in the past (feature 6) to time-slice 4, and all other valleys to time-slice 5;

• Put all the remaining hours into time-slice 6 (feature 7).

Note that in this case, the time-slice allocation is dynamic, in the sense that it will change if the user provides a different set of temporal profiles for energy supply and demand in the model input. It is possible to further split the hours introducing other dimensions, such as any of features 1 through 3, and by defining different threshold levels for the other features. For instance, with reference to feature 4, one can split the likelihood of excess supply into three tiers, high, medium, and low.

The possibility to use demand shift has been formulated separately for different energy service demands. Demand shift that is expressed in terms of final energy is represented differently from demand shift that is expressed in another unit (see Table 2). For energy service demands that are expressed in terms of final energy demand, the constraints used are:

$$ {D}_{r, fo,e, ts}={\mathbf{FD}}_{r, fo,e, ts}+{D}_{r, fo,e, ts}^{+}-{D}_{r, fo,e, ts}^{-},{\forall}_r,{\forall}_{fo},{\forall}_e,{\forall}_{ts}, $$

(12)

$$ {D}_{r, fo,e, ts}^{+}\le \mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{+},{\forall}_r,{\forall}_{fo},{\forall}_e,{\forall}_{ts}, $$

(13)

$$ {D}_{r, fo,e, ts}^{-}\le \mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{-},{\forall}_r,{\forall}_{fo},{\forall}_e,{\forall}_{ts}, $$

(14)

$$ {\sum}_{ts}\left({D}_{r, fo,e, ts}^{+}-{D}_{r, fo,e, ts}^{-}\right)=0,{\forall}_r,{\forall}_{fo},{\forall}_e, $$

(15)

in which:

fo:

is the final energy demand option, fo ∈ o,

D_r,fo,e,ts:

is the variable representing demand for energy carrier e,

FD_r,fo,e,ts:

is the parameter representing final demand for energy carrier e, [0,∞],

${D}_{r, fo,e, ts}^{+}$:

is the auxiliary variable representing an increase in demand for energy carrier e,

${D}_{r, fo,e, ts}^{-}$:

is the auxiliary variable representing a decrease in demand for energy carrier e,

$\mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{+}$:

is the parameter representing an increase in demand for energy carrier e, [0,∞],

$\mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{-}$:

is the parameter representing a decrease in demand for energy carrier e, [0,∞].

These variables and parameters apply to all regions r, time-slices ts, and in principle to all energy carriers e, but final energy demand only applies to electricity and heat. Via the parameters $\mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{+}$ and $\mathbf{\Delta }{\mathbf{FD}}_{r, fo,e, ts}^{-}$, the user can decide to what extent the final demand per time-slice can deviate from the final demand given by parameter FD_r,fo,e,ts. The demand variable D_r,fo,e,ts is simply related to the activity variable given in Eq. 2 by:

$$ {D}_{r, fo,e, ts}={A}_{r, fo, ts}{\mathbf{EA}}_e,{\forall}_r,{\forall}_{fo},{\forall}_e,{\forall}_{ts}, $$

(16)

in which EA_e is the energy allocation for energy carrier e, which is simply 1 for both electricity and heat.

In case the energy service demand is expressed via another unit, like the amount (in Mt) of steel that needs to be produced, the following constraint applies:

$$ {\sum}_{ts}{A}_{r,o, ts}{\mathbf{COS}}_{o,s}={\sum}_{ts}{\mathbf{SD}}_{r,s, ts},{\forall}_r,{\forall}_s $$

(17)

in which:

s:

is the service demand represented by a physical unit (like Mt),

COS_o,s:

is the binary parameter indicating if option o can deliver service demand s,

SD_r,s,ts:

Parameter representing the service demand s in time-slice ts,

and A_r,o,ts is the activity variable given in Eq. 2. If the service demand s corresponds to steel, the options o that can deliver steel have a value 1 for parameter COS_o,s.

In the case of, for example, steel production, by default a flat activity profile is assumed. Via the following constraints, one can deviate from the default profile:

$$ {A}_{r,o, ts}{\mathbf{COS}}_{o,s}\le {\boldsymbol{SD}}_{r,s, ts}^{\mathrm{max}},{\forall}_r,{\forall}_s,{\forall}_{ts} $$

(18)

$$ {A}_{r,o, ts}{\mathbf{COS}}_{o,s}\ge {\boldsymbol{SD}}_{r,s, ts}^{\mathrm{min}},{\forall}_r,{\forall}_s,{\forall}_{ts} $$

(19)

in which:

${\boldsymbol{SD}}_{r,s, ts}^{\mathrm{max}}$:

is the parameter representing the maximum service demand s in time-slice ts,

${\boldsymbol{SD}}_{r,s, ts}^{\mathrm{min}}$:

is the parameter representing the minimum service demand s in time-slice ts.

The user has the possibility to indicate to what extent the maximum and minimum service demand can deviate from the default service demand SD_r,s,ts by choosing values for respectively ${\boldsymbol{SD}}_{r,s, ts}^{max}$ and ${\boldsymbol{SD}}_{r,s, ts}^{min}$.

The prices used for primary energy carriers as used in section 4.3 are given in Table A2.

Table 6 Prices for primary energy carriers in 2030 and 2050 [€/GJ]

Values for energy service demand as used in section 4.3 can be found in Table A3.

Table 7 Energy service demand in 2030 and 2050