Double-layer Planning of Distribution Network considering Flexible Supply and Demand

To improve the flexibility of the system, flexible resources are added to the planning of renewable energy systems for collaborative optimization. Firstly, three quantitative indicators of flexible resources are introduced to build a distributed double-layer distribution network planning model that considers flexible resources and a more detailed cost model is considered based on existing research. Secondly, the scheduling and optimization of controllable loads are coordinated to achieve a fully integrated configuration of the distribution network. Then, the APSO algorithm is used to solve the problem. Finally, test results verify that the proposed planning approach is effective in improving system flexibility, economy, and reliability.


Introduction
With growing load demand and increasingly diversified energy use worldwide, the integration of distributed power sources, demand response, and energy storage into the planning and dispatch of distribution networks can effectively reduce risks and improve the reliability of distribution network operations [1].Due to the volatility, intermittency, and uncertainty of distributed power supply output, and its large-scale access to the distribution network, the node voltage climbs and the network loss also increases.We promote the grid integration and consumption of distributed power supply, coordinating the source network, load, and storage resources and improving the system flexibility and regulation capacity as well as the stable and economic operation of the distribution network [2], which becomes the focus of this paper.
There is a lot of research on distribution network planning issues involving distributed renewable energy.In [3], distributed renewable energy access and dispatch of interruptible loads were considered to solve the optimal installation location and capacity of distributed power supply.In [4], a variety of flexibility resources were integrated to reduce network losses and wind and light abandonment rates, and the role of micro gas turbines in system security was analyzed.In [5], a multi-objective optimal configuration model that considers economy and reliability was summarized to mitigate the effects of voltage variations.In [6], a double-layer optimization model was proposed to reduce annual comprehensive costs, but the problem of wind and light abandonment caused by distributed renewable energy integration was not considered.To summarize, the above studies had not fully exploited the potential of various types of resources of the system to regulate the power grid, which motivates the study of this paper.
Based on the above research, a double-layer planning model for distributed power supplies is proposed in this paper, which considers the uncertainty of distributed renewable energy output from the perspective of economic and flexible supply and demand balance.

Distribution network operation flexibility indicator
To ensure the flexible supply and demand balance of the distribution network system and avoid the occurrence of cutting machine load and abandonment of wind and light, it is necessary to ensure that the adjustment capacity and flexible supply capacity of flexible resources in the distribution network are greater than the flexibility demand of net load fluctuation [7].The fluctuation degree of the net load curve is less than the maximum climbing rate allowed by the system, which requires adding appropriate flexibility constraints to the distribution network optimization planning model to meet the flexibility demand of net load fluctuation.

Netload time domain volatility
Netload time domain volatility refers to the rate of change of net load in the distribution network in a unit of time, reflecting its degree of fluctuation per unit of time.
where FRB t F is the net load time domain volatility; B t P is the net load of the system at the current moment.

Netload time domain volatility limits
Within the range of wind power photovoltaic output fluctuations, the climbing rate of the net load curve of the distribution network cannot exceed the maximum climbing rate of the adjustable power of the system.The net load time domain fluctuation limit reflects the flexible adjustment ability of the distribution network system to withstand uncertain fluctuations during operation, that is, the climbing ability, as shown in Equation (2).P is the allowable climbing rate of the distribution network system.

Lack of flexibility
In the process of system operation, there will be an imbalance between the supply and demand of flexibility.When the supply of flexibility is greater than the demand, it means that the system is flexible enough, and the failure indicates insufficient system flexibility.To visually represent the supply and demand of system flexibility, the flexibility absence rate is introduced as an evaluation index.The combination of upward and downward adjustment of the flexibility loss rate constitutes the system flexibility loss rate, namely:

Distribution network operation flexibility indicator
To reduce the impact of the uncertainty fluctuation of renewable energy output on the system, a doublelayer optimization planning model is constructed [8].

Upper planning model
Coordinating and optimizing various controllable resources in the dispatch distribution network can improve the system economy, and the minimum comprehensive investment cost of the distribution network can be taken as the objective function: (5)

Flexibility resource call cost
where ep ( ) where U m and D m are the penalties for insufficient flexibility in upward and downward adjustments.
In addition, other costing equations refer to [9].

Lower planning model
The lower model considers suppressing the net load time domain fluctuation, sets the optimization target object to the net load time domain volatility with the minimum, and comprehensively considers the safety operation constraint and the equipment operation constraint.Other constraints are shown in [10].

The solution to the model
In the double-layer planning process, the planning level considers the comprehensive input cost of the distribution network as well as the installation location and capacity of the DRE.The operational level integrates the net load time-domain volatility and flexibility deficiency indicators for collaborative regulation and optimization by adaptive particle swarm algorithms to obtain optimal results.For convenience, the block diagram of the proposed planning algorithm is presented in Figure 1.

Basic parameter settings
The model used in this paper was validated by simulation using the standard IEEE 33 node arithmetic and examined scenery data from four scenarios, as shown in Figure .2.The nodes to be selected for installation are [5,13,15,30] and [3,12,18,26,29], with a maximum installed capacity of 5 kW of distributed renewables.The investment cost of WTG is RMB 4, 550/kW and the maintenance cost is RMB 0.032/KWH, and the investment cost of PVG is RMB 4, 100.The investment cost for the WTG is RMB 4, 550/kW and the maintenance cost is RMB 0.035/KWH.
The wind and PV output models can be found in [11].The energy storage system ESS is the original equipment of the system, the installation location is 17 nodes, the rated capacity, and the upper capacity limit is 1 MW/1.5 MWH, and the discharge efficiency is 0.9.The upper and lower limits of SOC are [0.1, 0.9], and the interruptible load installation nodes are 10, 13, 26, and 31, with a maximum curtailment rate of 50%.

Simulation analysis
The adaptive particle swarm algorithm and the CPLEX solver are used to solve the two-layer planning model.The number of particles in the particle swarm algorithm is set to 50 and the maximum number of iterations is 10.The iterative solution of the two-layer model is shown in Figure 2.

Figure 2. Iterative solution diagram
To facilitate comparative analysis, the following 2 scenarios are considered being planned in this paper: Scenario 1 is the planning model proposed in this paper; Scenario 2 is the planning model with the minimization of network loss costs as the lower-level objective function.I, the overall DG access capacity of Option 1 is greater than that of Option 2. Due to the consideration of the net load fluctuation rate, the system flexibility is enhanced and the system's ability to accept unstable power sources such as scenery can be improved.Due to the existence of the penetration rate constraint, the PVG access capacity of Option 1 is smaller than that of Option 2, but the PVG output is relatively more regular and less burdensome to the system operation.While for the fluctuating WTG, Option 1, which has a stronger acceptance capacity, can access more WTG, while Option 2 accesses less.As can be seen from Figure 3, the overall net load time volatility of Option 1 with flexibility is less than that of Option 2, especially during the mid-morning and mid-day peak periods around 8 am and 12 pm.The net load time volatility can be well mitigated, reducing the pressure on system operation and promoting the consumption of new energy sources such as scenery.The graphs of energy storage charging and discharging and SOC variation for Scenario 1 and interruptible load dispatching results are represented.As can be seen from Figure 4, energy storage started discharging at the two load peaks at 11:00 and 20:00 to smooth out peak-valley differences and scenery uncertainty.SOC reached a maximum at the peak load moment and decreased thereafter.As can be seen in Figure 5, the increasing demand for interruptible load calls at peak load moments promotes the role of interruptible load to alleviate supply and demand problems at peak moments.In Figure 6, we can see that Scenario 1, which considered flexibility, can reduce the risk of voltage overrun, and the voltage at each node in Scenario 1 was more stable at all times.As can be seen from Figure 7, while for Scenario 2 without considering flexibility, more obvious fluctuations can be seen near node 12 and node 13 with DG access, which fully demonstrates that the model was insufficient to maintain stable operation of the system under today's planning scenarios, and proved the usefulness of this paper's model.

Conclusion
In this paper, for the distribution network flexibility planning, a double-layer planning model taking into account the flexibility supply and demand was proposed, where net load time domain fluctuation rate, netload time domain fluctuation limit, and flexibility deficiency were introduced.The model proposed in this paper integrated multiple flexibility resources to achieve a rational use of resources, which could effectively improve the economy of system planning and the flexibility and stability of operation.The effectiveness of the proposed model considering economy, safety, flexibility, and reliability was verified through simulation experiments, which improved the computational efficiency and obtained more reliable simulation results.

FF
are the upward and downward loss rates of the distribution network system, respectively; are the probability of the scene occurring, respectively; S,U all F and S,D all F are the sum of the flexibility requirements in the upper and lower directions in all scenarios, respectively.

Figure 3 .
Figure 3.Comparison of net load volatility for different scenarios

Figure 4 .
Figure 4. Graph of energy storage charging and discharging and SOC variation Figure 5. Diagram of interruptible load dispatch

Figure 6 . 1 Figure 7 .
Figure 6.Node voltages during the running time of Option 1 Figure 7. Node voltages during the running time of Option 2 t is the time-of-use electricity price at t time; ESS C is the energy storage call cost per unit of electricity; ESS M is the number of energy storage charges and discharge cycle lifetimes; sp C is the discharge compensation price; IL ( ) C t is the price of interruptible load call at a timet .

Table 1 .
Data sheet on model scenarios