The dynamic scheduling process is modelled looking ahead over a 36 hour period at the demand profile to be met and associated reserve requirements. The model then schedules generation, storage and demand response for each 24 hour time horizon to meet these requirements. The actual day-ahead is varied by the stochastic modelling of the energy output from the renewable generation sources. The stochastic framework allows a number of renewable output realizations to be evaluated for each hour looking forward 36 hours. The generation and responsive demand resources in each region are simultaneously scheduled in order to consider multiple renewable generation output conditions for a prescribed set of network constraints. The model takes account of losses and costs incurred through the use of demand response and storage resources. The system operation model for scheduling generation and operating reserves in each region exploits the diversity of demand and renewable outputs across Europe to minimize operating costs while significantly enhancing the ability of the system to
accommodate the output of variable renewable generation sources.
The stochastic modelling of intermittent renewable generation results in an optimal allocation of long- term operating reserve between standing reserve and synchronised spinning reserve plant to maintain supply/demand balance. Any inadequacy in terms of the ability of the system to meet the demand given the need for reserve is managed by appropriate augmentation of generation capacity. The scheduling of reserves imposes further constraints on system operation for the following reasons. Reserve scheduling causes generation output deviations from the optimal generation schedule in order to provide sufficient flexibility for generation output to either be increased or decreased in response to variations in demand and/or supply. The operating characteristics of reserve generation introduce further constraints
including reducing the generation capacity available to supply demand and imposing limits on the lowest output to be delivered from flexible generation. The first effect can lead to requirements for greater generation capacity within the system either within each region or via interconnecting transmission. The second effect can lead to increased curtailment of variable renewable generation as the system must maintain adequate reserves, which will require flexible plant to be readily dispatchable. Where reserve generation is constrained by minimum stable operating limits, this can displace renewable generation unless sufficient transmission capacity is available to facilitate exports outside the node or sufficient storage is available within the node.
The key outputs of the stochastic scheduling and reserve model include hourly dispatch of each generation technology; hourly utilization of storage and demand response in each region; hourly allocation of operating reserves, renewable curtailment assessment and associated costs; transmission flows and congestions (flow duration curves); disaggregated total system operational costs per year including; start-up, no-load, fuel, losses and cost of renewable energy curtailment.
In order to deal with the uncertainties associated with conventional generation availability, demand fluctuations and variability of output of (variable) renewable generation three types of operating reserve are modelled:
Automatically activated frequency containment and frequency restoration reserves18 (“response”) that can be activated in a timeframe from several seconds to a few minutes;
Manually activated frequency restoration reserves (“Fast reserves”) with an activation time of less than 30 minutes; and
Replacement reserves with an activation time of several hours ("Back up") that are used to mitigate unforeseen imbalances between demand and supply over longer time horizons.
The initial reserve requirements have been determined based on existing operating practices, i.e. the volumes of different ancillary services that are currently procured and held by TSOs in each country. For future years, the basic reserve requirements are endogenously adjusted on an hourly basis, based on the simulated output variable RES-E in each hour. For each reserve type, the total volume is increased, if necessary, such as to cover the largest possible variation over the corresponding time horizon, using the persistence of short-term fluctuations.
On the supply side, the availability of different types of ancillary services varies by technology and depends on the underlying assumptions on ramp rates, start-up times and the ability for proiding frequency response and regulation. Moreover, a conservative approach has been followed by not including the contribution of any frequency sensitive loads towards frequency regulation (for example smart refrigerators). Moreover, the model allows sharing the different types of ancillary reserves between different regions across interconnectors. This is based on the overall co-optimisation of energy and reserves for each hour and ensures that a certain share of interconnector capacity may have to be reserved for reliability purposes.
At present, European TSOs generally procure ancillary services on a national basis. But there are also several examples of regional integration, for instance in the Nordic region and Central Europe. Moreover, the regional exchange of reserves and balancing services represents a core objective of the European Framework Guidelines for Electricity Balancing, which furthermore stipulate a set of different timelines for the different processes. In line with these developments and requirements, the simulations in this study are based on the assumption that reserves are regionally shared between different countries and regions. More specifically, back up capacity can be freel shared between all countries, subject to the availability of sufficient interconnector capacity. In addition, we consider the following regions, within which response and fast reserves are shared on a regional basis:
Nordic countries (NO, SE, FI, DK-East); Baltic (EE, LV, LT);
Central Western Europe (DK-West, DE, NL, BE, LUX, FR, CH, AT); Central Eastern Europe (PL, SK, CZ, HU);
GB + IE
Iberia (ES + PT); Italy; and
2.2 Transmission Modelling
For the modelling of the transmission grids, we have developed a simplified grid model that is based on a zonal transport model with a total of 74 individual nodes and approx. 165 existing and potential (inter-) connections between these nodes. As illustrated by Figure 22 this grid model covers all Member States that are physically part of the interconnected electricity market within the EU19. By also including Norway, Switzerland, Albania and the remaining members of ENTSO-E in South-Eastern Europe, the model
provides for a comprehensive coverage of the continental European grid.
Figure 22 Topology of Regional Transmission Model
In many cases, the networks of individual Member States are represented by two or more network zones, in order to better reflect internal congestion and the potential need for transmission expansion especially within larger countries20.
Similar to the approach chosen by ENTSO-E for the 10-Year Network Development Plan, the capacity of each existing (inter-) connection is determined by the Grid Transfer Capability (GTC). The GTC specifies the ability of the grid to transport electricity across a given boundary, i. e. from one area (a country or an area within a country) to another. The use of GTCs effectively corresponds to the notion of transfer capacities, which are commonly used for determining the transmission capacity available for cross-
19 Including Croatia but excluding Cyprus and Malta 20
In contrast, we have partially aggregated the countries in the Southern part of former Yugoslavia, due to the small size of the corresponding power systems and a limited expected impact on the development of the transmission grids at the European level.
border trading in the European electricity markets. The corresponding values have been derived from the NTC21 values published by the European TSOs, the incremental GTC's specified in the latest version of the 10-Year Network Development Plan (from July 2012) and from our own analysis of the existing and planned transmission infrastructure.
Besides the GTC of current and planned infrastructure, each (inter-) connection is characterized by the specific costs of capacity expansion, in terms of €/MW/km. The corresponding values are based on typical values, which have also been used by various other studies and which reflect the fundamental difference between AC and DC lines as well as topography; see Table 12.
Table 12 Basic Assumptions for Cost of Transmission Expansion
Technology Unit Costs
Overhead line (AC), normal conditions(a) M€/MW/km 0.500 Overhead line (DC), normal conditions(a) M€/MW/km 0.150
Submarine cable (DC) M€/MW/km 1.5
Additional costs for rough terrain 35%
Additional costs for extreme conditions 75%
Discount for use of guyed towers -35%
Converter station (AC/DC) M€/MW 0.075
(a)– Including cost of switchgear
Source: DNV GL assumptions
2.3 Distribution Modelling
2.3.1 Distribution Models used for the Analysis
The distribution analysis serves to understand the impact of distributed generation and load growth (e.g. electrification of heat and transport sectors) on future distribution network operation and investment, to quantify the benefits of alternative distribution network control approaches, including active network management, demand response and application of smart grid network technologies, and to assess the cost and performance characteristics of different distribution network design strategies, including optimisation of the number of voltage levels, equipment design approaches etc.
Apart from the analysis of technical and operational measures, the primary objective of the distribution analysis is to estimate the need and cost for distribution expansion in different scenarios. In addition, this analysis will also reveal whether and to which extent the use of distributed generation and their treatment in the market can help to avoid costs linked to additional grid build out at the distribution level, whilst the corresponding effects on the transmission level will be captured by the generation and
transmission model discussed in Sections 2.1 and 2.2 above. In both cases, our analysis will not address the principal presence of such effects but also help to quantify their impact.
As illustrated by Figure 23 the overall approach taken can be summarised as follows:
In a first step, we have collected information on typical network design policies and standards in different Member States and carried out a statistical analysis of population density (as a proxy for load density) in each country (see Section 2.3.2 below for further details);
This information is then used to create a set of typical networks that can be expected to be representative of the real situation in each of the Member States; and
These representative networks then provide the basis for the detailed distribution analysis and the determination of network reinforcement costs.
Figure 23 General Approach for Distribution Analysis
Source: Imperial College
Figure 24 below shows another view of our approach: for a given scenario, load characteristics of four network user categories, associated with individual local authority areas are specified for each year across the period: (i) domestic, commercial and industrial consumers, (ii) electric vehicles, (iii) heat pumps and (iv) various types of distributed generation. All users are allocated to distribution sites in relation to the scenario considered. In addition, the overall approach supports to different operation paradigms (e.g. business as usual or 'smart'). The chosen paradigm as well as the level of
responsiveness of demand assumed will then drive peak network demand and the corresponding levels of network reinforcement.
Figure 24 Modelling approach for distribution modelling
Source: Imperial College
The Dynamic Distribution Investment Model (DDIM) tests whether thermal, voltage and/or fault level constraints are violated and proposes appropriate upgrades of assets based on a defined reinforcement strategy. Finally, the model produces reports on network upgrades identified, an associated schedule, together with equipment utilisation profiles. This also includes modelling of alternative network
reinforcement and design strategies, quantifying the potential benefits of alternative mitigation measures such as demand response and other active network management techniques.
The developed modelling approach includes three distribution network models: Low Voltage (LV) network model;
Medium Voltage (MV); and High Voltage (HV) networks.
The LV network model is based on representative fractal networks with the parameters that represent the key characteristics of typical LV networks (0.4 kV) supplied from individual distribution transformers. The MV network model contains feeders with typical voltages of approx. 6 – 36 kV starting from
secondary busbars in the HV/MV substations and finishing with distribution substations. The HV network