Conclusions Related to Methodologies

In this research work various approaches related to methodological improvement for long- term energy system planning are outlined focusing on large-scale RE integration issues.

Methodological improvements are undertaken in terms of a) adopting finer temporal as well as satial resolution in the long-term planning model b) quantifying intra-regional RE potential using GIS tools and incorporating them in planning model in required temporal and spatial resolution c) develop and utilize a separate operational model to analyze the operational impact of system portfolio calculated by the planning model d) outlining a bi-directional linking method to incorporate the results calculated by the operational model into the planning model to recalculate new capacity values. Following three subsections presents overall summary and conclusion related to each method.

7.1.1 Endogenous Improvement of Energy System Planning Model

Compared to usual practice of long-term energy system modeling and planning studies, present exercise considers substantially higher number of annual time slices and model regions for a regional-scale planning study. In India, for the first time such high-resolution model settings are adopted in long-term energy system planning focusing on power sector.

There are several benefits of the endogenous modeling improvements. Higher number of annual time slices helps to capture the seasonality of demand and RE generation, and provide robust output compared to models with limited time slices. Intra-day time slices defined at hourly level helps to capture realistic variations of demand as well as RE generation at intra-day level. Regional definitions considered at state level helps to track inter-regional energy flow, regional annual generation mix, and technology activity profiles. These higher modeling settings help the planning model to ensure demand and supply balance for each region for every time slice, which improves calculation of technology capacity.

The finer modeling settings also help to perform other methodological improvements, such as incorporation of intra-regional RE variability and linking with an operational model.

Higher temporal and spatial definitions help to incorporate intra-regional RE generation as well as capacity potentials calculated at grid-cell level. It also helps to track activity profiles (i.e. dispatch) of technologies at higher temporal and spatial detail and compare them with operational model results. Following subsections details these aspects.

7.1.2 Linking Energy System Planning Model with GIS Based Tools

Consideration of intra-regional RE variability further enhances planning model’s capability to quantify RE penetration level and overall system portfolio. GIS methodologies adopted for this purpose do not precisely mimic the assumptions in official estimate calculations due to the difference in granularity and unavailability of data; but it outlines an effective way to incorporate intra-regional class wise RE potential limit in a planning model. The methodology can be scaled up or down according to requirement and data availability. Here, the intra-regional potential of RE classes is quantified at 1-degree by 1-degree geographical grid cells. Terrain suitability and exclusion criteria are also considered according to the available data resolution. It is possible to adopt a much higher resolution and employ several other exclusion/ suitability criteria to make the calculation process robust.

The planning model employs a substantially higher number of annual time slices com- pared to what is usually adopted in a large-scale, long-term planning study. To make use of this high temporal definition, time slice specific capacity factors are developed for each geographical grid-cells. The calculation utilizes sufficiently large quantum of historical

7.1 Conclusions Related to Methodologies 129 hourly time series data of wind to develop the corresponding CF values. The data utilized in this case is openly available and satellite-derived via remote sensing methods; hence resolution and reliability are lower than ground measured data. Though National Institute of Wind Energy (NIWE) provides historical ground measured data for some selected Indian locations; it is not useful in the present study as its spatial resolution is not uniform. In case of PV, ready-made long-term historical hourly time series (either satellite-derived or ground measured) was unavailable in open domain. Hence a widely used tooli.e., PVWatts has been utilized. This uses a representative years’ data derived from historical data sets to calculate annual generation. The accuracy of solar and wind CF calculation can be improved further if long-term historical ground measured data of solar radiation and wind speed is utilized.

These detailed RE related information has been incorporated into the planning model by creating various technologies to represent different RE classes. Additional user constraints are used to define region wise capacity potential. Time slice specific capacity factors are also provided per region per RE class.To the author’s knowledge, it is the first attempt to capture intra-regional spatial and temporal variability using GIS approach for a large-scale, long-term planning model for India. It outlines that open domain spatial data can be utilized by GIS methods to develop RE related information, which is not generally available at desired spatial or temporal resolution. Future studies in this regard should target to develop better assumptions, consider additional GIS data layers, and ground measured historical RE generations to develop more realistic RE resource potentials. These issues are highlighted in Section 7.3.

7.1.3 Linking between Planning and Operational Model

Linking of the system and operational models involve rigorous data preparation process, other than development of the operational model itself. Compared to existing studies in this regard, the operational model has several intra-regional nodes and it operates at hourly resolution. As the operational model operates with higher spatial and temporal resolutions compared to the planning model, availability of data at desired spatial and temporal level, development of robust assumptions and rules for data preparationetc., are crucial.

In the model linking process, technological capacity related data for the operational model are extracted from the planning model results. This information along with additional data, assumptions and rules, is used to generate intra-regional nodes, their location, and their spread. Rules are applied to convert aggregated capacity values into realistic unit sizes and allocate them at suitable nodes. In operational model also, grid-cell specific RE generation potential is considered to capture realistic RE variability.

The operational model portrays system operational insights at higher spatial and temporal resolution. Due to applied constraints, it ensures that the dispatch profiles of the generators are within their technical limits and more realistic power flow between the nodes. This leads to calculation of different activity profiles of the technologies compared to the planning model. Due to a considerable number of nodes, generators, and transmission lines, volume of the operational model is large. It is computationally challenging to solve the model for multiple milestone years of the planning model and perform multiple scenario analysis within reasonable time frame. Reformulation of model and refinement of model solving methods needs to be targeted. Incorporation of specific operational constraints in the planning model itself can lead may lead to more accurate dispatch decisions of generators, regional power exchange and help streamlining data exchange between the two types of models. There is a need to identify specific constraints to incorporate in the planning model based on trade-off between additional computational complexity and result accuracy. These are further elaborated in Section 7.3.

Bi-directional method illustrates how output of an operational model working at finer spatial and temporal level can be utilized in a planning model with coarse modeling definitions.

It also outlines how additional user constraints can be constructed from operational model results to update technology wise, time slice wise and region wise capacity factor in the planning model for each model iteration for attempting new solution.