diff --git a/doc/about.rst b/doc/about.rst index f6f5c538..57add017 100644 --- a/doc/about.rst +++ b/doc/about.rst @@ -1,229 +1,274 @@ .. _theoretical_background: -###################### -Theoretical background -###################### +############################ +Background and Methodology +############################ +Ding0 is a tool designed for the generation of synthetic distribution +networks, serving as a valuable resource for simulating and analyzing electrical +distribution systems. This section provides an in-depth understanding of the +theoretical foundations and core processes that underlie the functionality of ding0. +The fundamental data basis is described in [Huelk2017]_ and its extension is +detailed by [Amme2017]_. Further refinements are contributed by [Dubielzig]_ and [John]_. +:ref:`Ding0 flowchart` indicates the methodology of ding0 which is described more in detail within this section. -Data basis -========== + .. _Ding0 flowchart: + .. figure:: images/flowchart_mv_grids.png -The fundamental data basis is decribed in [Huelk2017]_ and its extension is -detailed by [Amme2017]_. Further extensions and additional details are provided -in the sections below. + Ding0 flowchart -:ref:`definition-of-terms` introduces terms we stick to in the following text. +Throughout the tool, a nesting has been implemented that is shown in +:ref:`Ding0 Structure`. -.. _definition-of-terms: -.. figure:: images/mvgd_la_lvgd-mit_ortsnamen.png + .. _Ding0 Structure: + .. figure:: images/ding0_basic_structure.png - Definition of names + Ding0 Structure -.. _lv-trafo-and-grid-district: +1. Data Import and Integration +=============================== -MV/LV Substations and LV grid district --------------------------------------- +The first step in the ding0 workflow involves the import of essential input data from two primary sources: -Medium-voltage/low-voltage (MV/LV) substations are located on a equidistant -grid of points with an interval of 180m within the load areas. -Cable length in low-voltage (LV) grids ranges from 100-1.500m (see [Kerber]_, -[Scheffler]_, [Mohrmann]_). -According to [Scheffler]_, a cable length of 200 m to 300 m is most typical. -Furthermore, we foud a difference between the cable length and the line over -ground is 72% (1.39 Umwegfaktor), see master thesis Jonas Gütter. -This seems plausible compared to the value for the MV grid of 77% (1.3). -The chosen value concludes in cable lengths of 250m at the shortest distance and 283m at the longest distance between the middle point of the square and its outer line. +1.1 **Open Energy Platform (OEP)** + OEP serves as a comprehensive data repository, supplying critical grid-related + information. For each distribution network, the following data types are imported: -* Finding LV-Grid districts (LV-GD): - We define Voronoi polygons within the load areas based on a grid of points with an interval of 180m. -* Assign consumption to the LV-GD: - This works analogously to the methods for the MV-GD, as described in "Allocation of annual electricity consumption and power generation capacities across multi voltage levels in a high spatial resolution" (Huelk) -* Assign peak load + ======================== ===== + Data Type Definition + ======================== ===== + Load Areas Geographical clusters designed for electrical loads + Building Loads Electrical loads assigned to specific buildings + MVGD Polygon defining the geographical area of the Medium Voltage Grid District (MVGD) + HV/MV station Location of the grid connection point between high -and medium voltage level + Generators Renewable and conventional power generation sources located in the regarded MVGD + ======================== ===== +1.2 **Open Street Maps [OSM]_ ** + OSM provides the street network data, forming the geographical basis of the distribution network. -Medium-voltage grids -==================== +2. Data Processing +=================== +2.1 **Allocation of Loads and Generators to belonging voltage level** -Methodological details and exemplary results are presented in [Amme2017]_. + **Loads**. High-resolution load data, which are allocated to specific buildings within the network area is categorized. + This categorization involves associating them with a specific voltage level, thereby determining the type of + grid connection they require. All loads **>1MVA** are considered for the MV ring topology (if connectivity is high). + Others are connected via stub connection to the assigned grid level. -Low-voltage grids -================= + ================== ================== ========================= + Voltage level Nominal capacity Allocation target + ================== ================== ========================= + HV-MV (20kV) 5.5 - 20 MW HV/MV substation + HV-MV (10kV) 3 - 11 MW HV/MV substation + MV (20kV) 0.2 - 5.5 MW MV grid + MV (10kV) 0.2 - 3 MW MV grid + MV-LV (0.4kV) 0.1 - 0.2 MW MV/LV substation + LV (0.4kV) ≤ 0.1 LV grid + ================== ================== ========================= -The topology of low-voltage grids is determined on the basis of typified grid -models that are vastly available for the residential sector and partially -available for other sector retail, industrial and agricultural. -The mentioned sectors are modeled differently: the grid topology of residential -sector loads founds on typified grid models from [Kerber]_. Retail and -industrial sector are treated as a single sector and use same methodology to -determine grid topology as applied for the agricultural sector. -Loads of each sector are located in separate branches - one for each sector. -In the following its creation is described in detail. + **Generators**. Depending on their nominal capacity, power generation sources are tagged with different allocation targets. + Ding0 does not contemplate generators when building the initial grid topology. They are connected through stub + connections to the grid topology afterwards as they play an important role for power flow analysis. + + ================== ================== ========================= + Voltage level Nominal capacity Allocation target + ================== ================== ========================= + level 4 (HV-MV) 4.5 - 17.5 MW HV/MV substation + level 5 (MV) 0.3 - 4.5 MW MV grid + level 6 (MV-LV) 0.1 - 0.3 MW MV/LV substation + level 7 (LV) ≤ 0.1 LV grid + ================== ================== ========================= + +2.2 **Clustering, Partitioning and Positioning** + + **Clustering and Partitioning.** The clustering process divides the road graph into subnetworks while considering + the inherent structure of the graph during cluster analysis. These resulting clusters, when + combined with the associated buildings, collectively constitute the Low Voltage Grid Districts (LVGDs) + within a Load Area (LA). + + **Positioning.** In each LVGD, the street point located at the load center of the LV grid is calculated, + considering only LV loads. This load center is utilized to determine the positioning of the MV/LV substation. + + .. figure:: images/clustering_positioning_partitioning_.png + + Clustering, Partitioning and Positioning of substations + + Details are presented in [Dubielzig]_. + +2.3 **Parametrization and Validation** + + **Parametrization**. In the parameterization phase, technical specifications for the MVGD are determined, including voltage levels + based on load density and distance between LA centers and the HV/MV substations. Load density exceeding 1 MVA/km² or distances + below 15 km results in an operating voltage of Vₙ = 10 kV; otherwise, Vₙ = 20 kV. This defines the use of underground cables (10 kV) + or overhead lines (20 kV) in the network. **Aggregated LAs** are identified when + + .. math:: + PLA \geq \frac{Imax_{th}}{3\sqrt{V_n}} + + + classifying them as urban regions. PLA refers to the peak load of the regarded LA. Transformer types are chosen based on peak loads, + with HS/MS transformers operating at up to 60% load and MS/NS transformers at up to 100% load, ensuring redundancy and (n-1) security + for the substation. +3. Low Voltage (LV) Grid Construction +======================================= + .. figure:: images/LVGD.png -.. Kerber describes 8 rural and 3 village and 8 suburban LV Grids; each with several branch lines. The exemplary grids are based on 132 real MV/LV Substations data in south Germany. -.. Scheffler gives statistical data about technical parameters of LV grids divided on 8 types of settlement areas. -.. Mohrmann discribes statistical data about technical parameters of LV grids based on 2700 LV-Grids. -.. Demirel -.. VNS + Example of LVGD + +3.1 **Clustering and partitioning** + + - The LV grid generation process begins with the clustering of LV Load Areas (LAs) based on the capacity of LV loads within each LA. + - Clustering results in the formation of clusters, and a clustered graph is positioned according to street topology. + - These clusters serve as the foundation for creating LV Grid Districts (LVGDs). + +3.2 **MV/LV substation Placement** + + - LVGDs are defined, partitioning LAs into LVGDs, each of which is associated with a load center. + - MV/LV stations are strategically positioned at the load centers of LVGDs. + +3.3 **Building the LV Grid** + + - The LV grid is then constructed within each LVGD, utilizing the OSM network. + - Loads with a demand of less than 100 kW are directly connected to the street graph. + - Loads with a demand between 100 kW and 200 kW are linked directly to the MV/LV station. + - Branching occurs from the MV/LV station based on capacity constraints, finalizing the LV grid topology. + - LV generators are integrated into the LV grid topology, with two possible connection levels: + - Level 6: Connects generators to LV stations. + - Level 7: Connects generators to the closest LV-grid node. + +4. Medium Voltage (MV) Grid Construction +========================================= + + .. figure:: images/MVGD.png + + Example of MVGD + +**4.1 MV Grid in rural areas (regular & satellite LAs)** + + **Assumptions** are established that underlie the entire MV grid generation process + + ======================================== ================================================================================================== + Assumption Value + ======================================== ================================================================================================== + Type of topology Open ring topology + Voltage Level (MV) 20 kV (if load density > 1 MVA/km² or distance between LA centers < 15 km), 10 kV (otherwise) + Preferred Cable Type (10 kV) Earth Cables + Preferred Cable Type (20 kV) Overhead Lines + Maximum Line Loading normal Up to 60% + Maximum Line Loading failure Up to 100% + Maximum Voltage Drop normal 5 % + Maximum Voltage Drop failure 10 % + Detour Factor 1.3 + Reactive Power requirements loads cos(φ)=0.9 + Reactive Power requirements generators cos(φ)=1 + ======================================== ================================================================================================== -However, a method to generate a representative variation of LV-grids, that can be assigned to the modeled LV/MV substations cannot be found. -Given data on MV/LV substations: + **Rounting.** The routing for the initial grid topology in the context of the ding0 tool is + based on the Capacitated Vehicle Routing Problem (CVRP) formulation which is solved by a two-stage + metaheuristic approach. The classic CVRP optimization problem is adapted for designing the MV grid, + with the objective of determining the most efficient routes for supplying electricity to different + MV gird connection points (MV/LV substations and MV loads). + First, the initial routes are constructed using a parallel savings heuristic of + Clarke and Wright. This heuristic identifies potential savings by combining routes and iteratively + improves the solution. Only centers of LAs are regarded as potential points for the algorithm. + Second, local search heuristics are used to refine the routes further. -* land use data divided in industry, commercial, agriculture and residential -* population -* peak load -* Define transformer + ======================================== =============================== ==================================================================== + CVRP notation Appliance to grid planning Explanation + ======================================== =============================== ==================================================================== + customers Centers of LAs Location that needs to be visited + depot HV/MV substation Central point from which the MV grid routes originate and return + ======================================== =============================== ==================================================================== -Branches of sector residential ------------------------------- + Throughout the routing process, various technical constraints are considered, including current carrying capacity, + voltage stability, load factor, line loading, and operational modes (normal and faulty). These constraints ensure that the designed grid + remains technically feasible and reliable. -.. TODO: Editha + **Grid extension.** The initial MV grid topology is extended by those MV grid connection points that are adversed above + due to proximity reasons or technical constraints. Three sequential steps are executed to connect satellite LAs, MV/LV substations, + and generation units to the existing grid are proceeded by applying the order of connection respectively. -#. LV-Branches + **(1)** Nodes within a proximity of ≤ 100 meters to an existing grid route are integrated into the grid by adjusting the route's path. - We are using the LV-Branches of Kerber from the grids. They should be assigned to the most plausible types of settlement areas. + **(2)** Geographic Information System (GIS) methods are used to find and connect remaining nodes, starting with a search radius of 2000 meters. + The radius expands incrementally if no suitable points are found. -#. Define the type of settlement area - - To decide if a LV-grid district is most likely a rural, village or suburban settlement area we are using the population value combined with statistical data. Statisticly, there are 2.3 persons per appartment and 1.5 appartments per house. [see BBR Tabelle B12 http://www.ggr-planung.de/fileadmin/pdf-projekte/SiedEntw_und_InfrastrFolgekosten_Teil_2.pdf] [DEMIREL page 37-41, average has been coosen]. (This is not valid for urban areas.) With this we estimate the amount aus house connections (HC). + **(3)** If the above connection options are infeasible due to technical constraints, nodes are directly connected to the main route using + separate branch lines and the standard line type. - This value can also be found at the explenation of the database of the "Kerber"-grids and is assinged to the type of settlement area: + Details on routing principles for MV grid topology are presented in [Amme2017]_. Be aware that there have been major changes in + methodology since the publishing of this paper: - * Rural: 622 HC at 43 MV/LV substations results in an average amount of 14.5 HC/substation - * Village: 2807 HC at 51 MV/LV substations results in an average amount of 55 HC/substation - * Suburban: 4856 HC at 38 MV/LV substations results in an average amount of 128 HC/substationTher + ================================================== ====================================================================================================== + Initial methodology Update + ================================================== ====================================================================================================== + Sector-specific electricity demand High resolution load data for each building + Equidistant grid of points for MV/LV substations Location of MV/LV substations are based on load center of each LVGD + Voronoi partition for LVGD definition Definition of LVGDs as a result of clustering by loads + No MV loads Heavy load electricity consumers are classified as MV loads and connected within that voltage level + ================================================== ====================================================================================================== - With the resulting trendline of this three point, [the Polynomial degree 2 [ 16.127*(x^2)-7.847*x+6.1848 ] whereas x is the type of of settlement area], we difine the border values for the typ of settlement area at: +**4.2 MV Grid in urban areas (aggregated LAs)** + +Aggregated Local Areas (LAs) are characterized by a high cumulative power demand, classifying them as urban regions. +In these LAs, each is connected to the HV/MV substation through at least one direct connection, which depends on the cumulative load. +To design the MV grid within aggregated LAs, actual road network distances (OSM) are considered. The graph representing this network +is preprocessed and divided into two components: G_core, which contains potential MV-ring topology customers, and G_stub, which +includes the remaining customers for MV grid connection via stubs. + +Initially, all well-connected customers (those with at least two neighbors) are primarily assigned to G_core. +Subsequently, the stub connection criteria are evaluated, considering a load threshold of <1 MVA. Customers exceeding +this threshold are also moved from G_stub to G_core to ensure grid stability. The final G_core graph encompasses all customers considered +for the initial MV ring topology. The routing procedure aligns with that used in rural areas, with the additional constraint of +network distances being considered instead of air distances. The Dijkstra Algorithm plays a crucial role in determining the precise +road geometries of ring circuits and their lengths. + + ======================================== =============================== ==================================================================== + Criterion Rural area Urban area + ======================================== =============================== ==================================================================== + Distance calculation Street network Air distances * detour factor + Customers for initial ring topology LA centers Final G_core components (MV/LV stations + MV loads) + Customers for stub connection missing LA centers, Final G_stub components (MV/LV stations + MV loads) + generators, MV/LV stations + Voltage level 10kV and 20kV mostly 10kV + Connection type Cables and overhead lines mostly cables (underground) + ======================================== =============================== ==================================================================== + +Details on MV grid design in urban areas are presented in [Dubielzig]_. + +5. Grid Analysis and Reinforcement +=================================== + + +**5.1 Relocating Switch Disconnectors and Power Flow Analysis** + +Switch disconnectors are used to realize redundancy within the grid in case of an outage of a component. +MV-rings can be operated as isolated half-rings.The grid thus does +meet minimum technical requirements. A ring’s switch disconnector was initially located on the line where power +flow is minimal in closed condition. Due to the extension of the grid, additional loads are +connected to the rings. Therefore, the switch disconnectors are subsequently relocated to fulfill +this requirement. For further power flow analysis, results are exported to a PyPSA format. + +**5.2 Grid Reinforcement** + +Grid reinforcement measures are employed to enhance the MV grid robustness (LV grid reinforcement and transformer reinforcement is not implemented yet). +The main purpose is to address overloading issues within the grid and ensure stable operation. +The following steps are applied: + + **(1)** Identify critical branches and stations with overloading issues. + + **(2)** Reinforce critical branches by selecting appropriate cable types. + If no suitable cable type is found for a branch, the branch's original type is retained. + + **(3)** Run a power flow analysis to check for voltage issues. + + **(4)** Select larger cable types for the critical branches to address voltage issues. + For each node where voltage exceeds 3 % of nominal voltage in feed-in case or 5 % + of nominal voltage in load case, branch segments connecting the node with the substation + are reinforce until no further issues remain. - * Rural <31 HC/substation - * Village <87 HC/substation - * Suburban >=87 HC/substation - -#. Assinging grid branches to the Substations - - within the "Kerber"-model-grids several grid branches are found. - - * Rural: 5 branches (with l>=78m & l<=676m) - * Village: 7 branches (with l>=102m & l<=588m) - * Suburban: 15 branches (with l>=85 & l<=610m) - - -Strangzuweisung -Zu jeder ONS werden in Abhängigkeit von Netztyp und HA, NS-Stränge zugewiesen -Eine Verteilung des Aufkommens der Stränge anhand von der Gesamtstranglänge geschieht mit Hilfe der Scheffler Angaben (Abbildung Länge der Netzstrahlen für ausgewählte Siedlungstypen [44]) - - #. Categorising grid branches form "Kerber" model grids - -Hinzu kommen auf Basis von kerber interpolierte stränge um Lücken in der Vollständigkeit zu schließen - -Branches of sector retail/industrial and agricultural ------------------------------------------------------ - -Creating individual LV grid branches for the sectors retail/industrial and -agricultural applies the same methodology. -The topology of these grid branches determines by the sectoral peak load that -is available at -high spatial resolution (see [Huelk2017]_). Furthermore the number of land-use -areas (taken from [OSM]_) of each of the sectors determines the number -individual loads connected to one or more of these sectoral branches. - -The topology of each sectoral branch is affected largely by assumptions on -parameters that are provided in the table below. - -.. _assumptions: - -========================================================= ===== -Parameter Value -========================================================= ===== -Max. load in each branch 290 kVA -Max. branch length retail/industrial :math:`L_{R/I,max}` 400 m -Max. branch length agricultural :math:`L_{A,max}` 800 m -Length of branch stub 30 m -PV peak power <= 30 kW residential -PV peak power > 30 kW <= 100 kW retail/industrial or agricultural -PV peak power > 100 kW MV/LV station bus bar -========================================================= ===== - -In each LV grid district (LVGD) (see :ref:`lv-trafo-and-grid-district`) sectoral -peak load of sectors retail+industrial and agricultural are analyzed. The -number loads of each sectors determines by dividing sectoral peak load by -number of land-use area found in this grid district. - -.. math:: - N_{loads} = P_{sector} \cdot N_{land-use} - -In the next step individual loads are allocated to branches considering the -limit of max. 290 kVA peak load connected to a single branch. If a single load -exceeds the limit of 290 kVA, it is halved until it is smaller than or equal -to 290 kVA. -Loads are distributed equidistant on the branches while the branch does not -necessarily take the maximum length defined in the table above. -The distance defines as - -.. math:: - d_{sector} = \frac{L_{sector,max}}{N_{loads} + 1} - -Single loads are connected to the branch line by stubs of a length of 30 m. - -Photovoltaic (PV) power plants are allocated to different sectoral LV grid branches -depending on the nominal power. The allocation by the nominal power is provided -in the above table. It follows a simple assumption: smaller PV power plants are -allocated to LV grid branches of sector residential, larger power plants -are allocated to branches of the other sector, and really large ones are -directly connected to the bus bar of the MV-LV substation. - -Grid stability and equipment -~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - -During build of LV grid topology equipment is chosen with respect to max. -occurring load and generation according to current grid codes (see [VDEAR]_). -Nevertheless, some overloading issues may remain. In addition, voltage issues -may arise that can't be considered during grid topology creation. Therefore, we -adhere to the regulatory framework of [DINEN50160]_ which is simplified by -[VDEAR]_. -According to [DINEN50160]_ voltage deviation is limited to +/-10 % of nominal -that is for practical use divided into voltage drop/increase for each voltage -level and the associated transformers. The allowed voltage increase in the LV -grid level is limited to 3 % of nominal voltage. The allowed voltage drop is -limited to 5 % as detailed in [Zdrallek]_. - -.. as detail by :numref:`Image of Sphinx (Fig. %s) `. - - - -.. .. _voltage-deviation: -.. .. figure:: images/DIN_EN_50160_voltage_drop_scheme.png -.. -.. Allowed voltage bandwidth in distrivution grids detailed for each grid -.. level. Source [Zdrallek]_ - - - -Following steps do apply during reinforcement of Ding0 LV grids - -#. Checks for **overloading** issues at branches and MV-LV transformers first -#. Critical branches (those with line overloading) are extended to appropriate - size of cable to transport connected load and generation. Note, if connected load or - generation capacity is still exceeding capacity of largest cable type. We keep - largest available cable type and the issue most probably will remain -#. Stations are tested for overloading issues for generation and load case as - well. If nominal apparent power of transformers of a substation is not - sufficient a two-step procedure is applied - - #. Existing transformers are extended (replaced) to comply with load and - generation conencted to subsequent grid. - #. If Step 1 does not resolve all issues additional transformers are build - in the substation -#. Subsequently **over-voltage issues** are analyzed for all grid nodes -#. For each node where voltage exceeds 3 % of nominal voltage in feed-in case or - 5 % of nominal voltage in load case, branch segments - connecting the node with the substation are reinforce until no further issues - remain. If a over-voltage issue cannot be solved by installing largest - availabe cable (NAYY 4x1x300) this type of cable still remains as well as - the overvoltage issue -#. Substations are checked for over-voltage issues at the bus bar individually. - Identified issues are resolved by extending nominal apparent power of - existing transformer. A ultimately build up to two new transformers in the - substation. +Once grid reinforcement is complete, switch disconnectors are closed, finalizing the network configuration. References ---------- @@ -237,21 +282,9 @@ References generation capacities across multi voltage levels in a high spatial resolution*. International Journal of Sustainable Energy Planning and Management Vol. 13 2017 79–92, `doi:10.5278/ijsepm.2017.13.6 `_ -.. [Kerber] G. Kerber: Aufnahmefähigkeit von Niederspannungsverteilnetzen für - die Einspeisung aus Photovoltaikkleinanlagen, Dissertation, TU München, - 2011 -.. [Scheffler] J. Scheffler: Bestimmung der maximal zulässigen - Netzanschlussleistung photovoltaischer Energiewandlungsanlagen in - Wohnsiedlungsgebieten, Dissertation, TU Chemnitz, 2002 -.. [Mohrmann] M. Mohrmann, C. Reese, L. Hofmann, J. Schmiesing: Untersuchung - von Niederspannungsverteilnetzen anhand synthetische Netzstrukturen. In: - Proceedings of VDE ETG Kongress, 2013 +.. [Dubielzig] P. Dubielzig: Modellierung synthetischer Verteilnetztopologien + in urbanen Gebieten, Dissertation, TU Berlin, 2022 +.. [John] R. John: Planning of Synthetic Low Voltage Networks + with Geographical Constraints, Dissertation, Offenburg University, 2021 .. [OSM] OpenStreetMap contributors: `Open street map `_, 2017 -.. [VDEAR] VDE Anwenderrichtlinie: Erzeugungsanlagen am Niederspannungsnetz – - Technische Mindestanforderungen für Anschluss und Parallelbetrieb von - Erzeugungsanlagen am Niederspannungsnetz, 2011 -.. [DINEN50160] DIN EN 50160 Merkmale der Spannung in öffentlichen - Elektrizitätsversorgungsnetzen, 2011 -.. [Zdrallek] Planungs und Betriebsgrundsätze für ländliche Verteilungsnetze -- - Leitfaden zur Ausrichtung der Netze an ihren zukünftigen Anforderungen, 2016 diff --git a/doc/images/LVGD.png b/doc/images/LVGD.png new file mode 100644 index 00000000..485c4c67 Binary files /dev/null and b/doc/images/LVGD.png differ diff --git a/doc/images/MVGD.png b/doc/images/MVGD.png new file mode 100644 index 00000000..9dcb79e9 Binary files /dev/null and b/doc/images/MVGD.png differ diff --git a/doc/images/clustering_positioning_partitioning_.png b/doc/images/clustering_positioning_partitioning_.png new file mode 100644 index 00000000..aa64a47e Binary files /dev/null and b/doc/images/clustering_positioning_partitioning_.png differ diff --git a/doc/images/ding0_basic_structure.png b/doc/images/ding0_basic_structure.png new file mode 100644 index 00000000..2493f308 Binary files /dev/null and b/doc/images/ding0_basic_structure.png differ diff --git a/doc/images/flowchart_mv_grids.png b/doc/images/flowchart_mv_grids.png new file mode 100644 index 00000000..d8ce1406 Binary files /dev/null and b/doc/images/flowchart_mv_grids.png differ