1.1. Literature Review and Contributions

TEP is an optimization problem that finds the optimal solutions to problems such as when, where, and how many circuits are needed to build new transmission lines based on different objective functions and constraints [4, 5]. The TEP problem is multiobjective, binary, nonlinear, nonconvex, constrained, and complex by its nature [6]. In traditional transmission planning studies, the main component of the objective function is to minimize the total cost of new transmission lines. However, in recent years, new objective functions and constraints have been included in the TEP with changing grid and electricity market conditions [7]. In [8], the objective function includes loss of load cost (LOLC) as a reliability index in addition to the line cost. In [9], the operation cost due to line congestion is merged into the objective function. In [10, 11], another critical issue, the N-1 security constraint in the system security criterion, is included in the TEP.

TEP studies are conducted for different time horizons. Medium-term TEP is conducted less than 5-year periods, while long-term TEP is implemented for periods of more than 5 and 10 years [12]. Regarding the time slots selected in planning horizons, TEP is traditionally classified as static (single-stage) and dynamic (multistage). If the planner is looking for optimal solutions for a single time slot in the planning horizon, it is defined as static TEP. Here, the planner is not interested in when new transmission lines will be installed since the final solution will be obtained for a single static future state. Dynamic TEP differs from static TEP in which the planning horizon is divided into many time horizons, and optimal solutions are sought for each time horizon. In this case, the planner has to deal with the location of the new transmission line, the number of circuits, and its timing [13].

For the transmission grid, the DC model based on DC load flow equations and the AC model based on AC load flow equations are used in the TEP studies [14]. The AC model allows planners to analyze parameters such as grid loss, reactive power flow, and voltage stability. The DC model is a linearization and simplification of the AC model with some assumptions [15]. The AC model accurately represents the grid, but the DC model is widely used due to the complexity of the TEP problem and the large computational burden [16].

In solving the TEP problem, efficiency is crucial, along with obtaining accurate results. The optimization algorithms to solve the problem should be efficient enough [17]. Although various algorithms are used in modeling and solving the TEP problem, these algorithms are generally categorized as mathematical and metaheuristic. Mathematical algorithms use mathematical techniques to solve the TEP problem and are called conventional methods. Among the mathematical algorithms, there are linear programming (LP) [18], nonlinear programming (NLP) [19], mixed integer linear programming (MILP) [20], mixed integer nonlinear programming (MINLP) [21], hierarchical techniques (HT) [22], benders decomposition (BD) [23], and branch bound (BB) [24]. Mathematical algorithms give good results for small and less complex systems. Considering the complexity of the TEP problem and the size of the search space, the solutions take a long time, which reduces the solution’s efficiency, and the divergence problem might become a current issue. In recent years, metaheuristic algorithms have been gaining popularity as an alternative to overcome the drawbacks of mathematical algorithms in the solution of TEP [20]. The development of metaheuristics generally begins with inspiration from nature, offering better performance and higher quality solutions in a shorter time. In the solution of TEP, novel metaheuristic algorithms such as constructive metaheuristic algorithm (CMA) [25], tree searching heuristic algorithm (TSHA) [26], orthogonal crossover-based differential evolution (OXDE) [27], teaching learning-based optimization (TLBO) [28], shuffled frog leap algorithm (SFLA) [29], and salp swarm algorithm (SSA) [30] have been widely used in recent years. In addition to novel approaches, relatively old metaheuristic algorithms such as particle swarm optimization (PSO) [31], nondominated sorting genetic algorithm-2 (NSGA-2) [32], ant colony optimization (ACO) [33], artificial bee colony (ABC) [34], and grey wolf optimization (GWO) [35] continue to be used in TEP studies.

In the literature, TEP studies are generally classified according to algorithms, objective functions and constraints, network models, planning horizons, and test systems. Table 1 summarizes the classification based on the characteristics of algorithms used in recent years in TEP studies.

The forensic-based investigation (FBI) algorithm is a newly developed efficient optimization algorithm inspired by the suspect investigation, location, and pursuit process in criminal cases, which is being applied to solve continuous optimization problems [38]. The FBI algorithm has been used effectively in solving optimization problems in various fields, such as the optimal design of frequency-constrained dome structures, parameter extractions of solar cell modules, selection of structural design parameters, classification of patched and unpatched potholes, and identification of the ideal parameters for a fractional-order PID-based maximum power point tracking system for proton exchange membrane (PEM) fuel cells [39–43]. The FBI algorithm is effective because it needs a few control parameters to achieve the optimal solution for the problems. Moreover, it is quite capable and accurate in solving large-scale and high-dimensional problems [44].

In this study, a new binary version of the FBI algorithm called the BFBI algorithm is proposed to solve the TEP problem. The original FBI algorithm is not able to solve binary optimization problems. To overcome this drawback and use it in solving the TEP problem, we have developed the BFBI algorithm, which is proposed as a new binary variant of the original FBI algorithm. In addition, GA and PSO algorithms are used as benchmarks to evaluate the performance of the BFBI algorithm in solving the TEP problem. GA and PSO algorithms are relatively old metaheuristic optimization algorithms that perform well and are widely used to solve continuous problems. Many variants of the traditional GA and PSO algorithms have been developed and used to solve binary problems such as TEP [45, 46]. This study aims to achieve robust and efficient transmission expansion for future planning horizons by considering reliability and security criteria in transmission grids.

The rest of the paper is organized as follows: the TEP problem formulation is presented in Section 2. The BFBI algorithm is described and the general strategy for implementing the algorithms is given in Section 3. The obtained results are discussed in Section 4. Finally, conclusions are given in Section 5.

2.1. Multiobjective Function

The security approach in transmission grids considers the continuation of the operation without any outage in the grid in case one or more elements are out-of-service in a grid with N elements. N-1, N-2, and N-1-1 are system security criteria frequently used in transmission grid studies. In TEP studies, the N-1 security criterion is commonly used, which makes the TEP problem more complex and the solution to the problem more difficult. In the TEP study for the Garver 6-bus test system, the out-of-service of each line in the grid is included in the TEP problem as a security criterion, while in the TEP study for the Turkish grid, the N-1 contingency line list is created according to the reference limits determined depending on the loading of the lines in the grid in the N case and included in the TEP problem.

Lack of capacity of transmission lines in the transmission grid can cause line congestions. When line congestions occur in N or N-1 cases, generation and/or load must be rescheduled to ensure system reliability. Essentially, redispatching causes additional operational costs for TSOs. In addition, the inadequacy of the grid, especially in N-1 cases, may lead to the formation of isolated regions in the grid that are regionally disconnected from the main grid. This is an undesirable operational condition in terms of the system operation and system reliability. Thus, eliminating isolated regions in the grid also incurs new additional operational costs for TSOs.

Two different objective functions are used in this study: the LOL cost and investment cost are considered in equation (7) and the operational cost and investment cost are considered in equation (8). The transmission system planner creates an investment plan according to the result obtained by minimizing the total costs in the objective function in equations (7) and (8).

2.2. Constraints

Equations (11) and (12) denote the power flow constraints between bus i and bus j in the N and N-1 cases, respectively. Maximum limits of the power flow between bus i and bus j are calculated by using thermal capacity of the lines. Equation (13) shows the maximum and minimum generation limits at each bus. Generation limits are calculated by summing the generators connected to each bus. Equation (14) shows the loss of load limits, and equation (15) shows the new line limits that can be added for each right-of-way (corridor).

3.1. Original FBI Algorithm

In recent years, numerous metaheuristic optimization algorithms inspired by natural and human-based processes have been developed. The FBI algorithm was recently proposed by Chou and Nguyen as a human-based metaheuristic optimization algorithm in 2020 [38]. The algorithm is based on the process by which police investigate, pursue, and locate a suspect in criminal cases [42]. FBI is powerful, efficient, stable, and user-friendly than other algorithms to solve continuous nonlinear problems [43].

The FBI process can mainly be divided into six steps: opening a case, interpretation of findings, inquiry direction, actions, extension of actions, and prosecution. The process starts when a criminal event occurs and is reported and ends after the case is solved and the truth comes out. The FBI process has two main phases: the investigation phase and the pursuit phase. In the investigation phase performed by the investigation team, interpretation of findings and direction of investigation are carried out, whereas in the pursuit phase performed by the pursuit team, action-oriented steps are taken, starting with the most promising search direction first and collecting the data about the case. At the end of each action, the investigation team interprets new findings and information gathered, and new investigation directions are determined. The investigation phase corresponds to the exploration phase, and the pursuit phase corresponds to the exploitation phase in the metaheuristic optimization literature.

Figure 1 shows the steps of the FBI process. The middle four steps of the process are seen as a cycle that continues until the truth is revealed. The investigation team performs Step 2 and Step 3, whereas the pursuit team performs Step 4 and Step 5. Step 1 is the occurrence of the incident, the filing of a complaint and the opening of a case file, while Step 6 is the final step, which is the uncovering of the truth and the prosecution of the suspect. Those steps are briefly explained as follows [39, 40, 47]:

3.2. Binary Transformation of the Original FBI

Optimization problems are divided into two main categories regarding search space: binary and continuous problems. In binary problems, search agents take only 0 or 1 in the search space, while in continuous problems, agents might take fractional values in the search space. The FBI algorithm was initially developed to solve continuous problems. However, numerous binary optimization problems exist, such as TEP, in which decision values take only binary values (1 or 0) since the problem is adding or not adding a new transmission line. Therefore, a transformation is required to move particles to binary space. We have introduced a new binary variant of the FBI algorithm by integrating a transfer function into the original FBI algorithm to solve the binary TEP problem.

Transfer functions are among the most widely used techniques because of their simplicity as an operator in the binary transformation [48]. Various transfer methods have been introduced from continuous search space to binary space, such as v-shaped [49], s-shaped [50], o-shaped [51], z-shaped [52], x-shaped [53], cosine transfer function [54], and hyperbolic tangent transfer function [55]. In this study, the hyperbolic tangent transfer function is used to convert the continuous search space to a binary one since it has some advantages, such as a few control parameters, an effective convergence rate, and the simplicity of implementation.

In the TEP process, the binary value of an individual, which is obtained at the end of the transformation process, contains information about whether a candidate line will be added to the grid and how many lines will be added to the grid in the planning horizons. A representation of the triple binary structure of the individuals can be seen in Figure 2.

In the TEP study for the Turkish grid, as the new line limit that can be added for each corridor is decided as one, each triple binary value of an individual contains just information about which candidate line in the candidate set will be added to the grid. In the TEP study for the Garver 6-bus test system, as the new line limit for each corridor is decided as seven, each triple binary value of an individual contains both information of which the candidate line in the candidate set will be added into the grid and the number of the related candidate line to be added with their decimal conversion.

3.3. TEP Problem Solving Flowchart

The general flow diagram of the TEP implementation model is shown in Figure 3, which allows the BFBI algorithm to be applied along with other algorithms to solve the TEP problem. The model is initialized by defining the data about the grid (line, bus, load, generation, and so on) and the optimization algorithm parameters to be used (number of iterations, number of populations, upper and lower bounds, and so on) as input to the model. The initial population of the algorithm is generated in a randomized fashion to obtain an initial solution. Then, the iterative iteration loop is started. In the loop, lines from the candidate line set are added to the model one by one and the model is run according to the objective function to be used in the model. If the objective function includes only the investment cost of the new lines to be added to the grid, only investment costs are calculated in the model, and the total cost is taken. If the objective function of the model includes operational costs, loss of load costs, and N-1 in addition to investment costs, the total cost is obtained by calculating the costs separately for these objectives. This iterative process continues until the number of iterations determined at the beginning of the model is completed and the solution set is obtained. After the stopping condition required for the completion of the model is met, the optimal result is found from the solution set. The model is completed by static verification of the optimal result based on DC power flow in the PSS/E power system analysis program.

4.1. Application to the Garver 6-Bus Test System

The standard Garver 6-bus test system is used as one of the test systems in implementing the BFBI algorithm. The system, whose base topology is shown in Figure 4(a), consists of five buses and six lines connecting; however, the sixth bus is initially considered disconnected. In the initial configuration, the system’s total installed generation capacity and load are 270 MW and 190 MW, respectively. In the planning horizon, it is planned that the system load will be increased by four times and reach 760 MW, and the total installed capacity will reach 1110 MW by connecting the power plant at the sixth bus to the system. Furthermore, fifteen candidate corridors are identified where a maximum of seven circuits of the line can be installed in each corridor to solve the TEP problem. The dataset and detailed system description can be found in [56]. In the second stage, the results obtained with the BFBI algorithm are compared with results obtained using conventional and metaheuristic algorithms such as MILP, MINLP, BD, tabu search (TS), modified gases Brownian motion optimization (MGBMO), artificial neural network (ANN), linear success history-based differential evolution (LSHADE) in the literature. The statistical results are obtained and analyzed to demonstrate the effectiveness of the BFBI algorithm.

4.1.2. Performance Evaluations of Algorithms

Performance indicators including convergence, population diversity, exploration and exploitation are used to evaluate the performance of the BFBI algorithm. In order to understand the trade-off between diversification and intensification, population diversity, exploration, and exploitation behaviors have been examined. In addition, the convergence curve has been analyzed to show how the solution converges to the optimal location. Obtained figures related to these factors are analyzed and interpreted based on the approach offered in [61]. The defined values for the parameters of the BFBI algorithm are given in Table 3.

Table 3. The BFBI algorithm parameters.

Parameter	Value
Population of individuals	50
Number of iterations	500
Number of runs	30

Figure 5 gives a quantitative comparison of algorithm performances by considering the N-1 contingency. The convergence plot of the algorithms for the best run among all runs is presented in Figure 5(a). GA is the fastest algorithm, reaching the optimum solution in the 22nd iteration, while the BFBI algorithm reaches the optimum solution in the 28th iteration. The PSO algorithm is the worst, reaching the optimum solution with the 386th iteration. In Figure 5(b), the population diversity measurements can be seen for the best run among all runs. The diversity approaches zero in GA by finding the most promising search directions. However, for the BFBI algorithm and PSO algorithm, it does not decrease to zero even after all iterations. Furthermore, decreasing the diversity before converging to the optimum solution in the problem-solving process is undesirable. Although the optimum solution is found in the 28th iteration, the BFBI algorithm manages to keep the population diversity high. Figure 5(c) depicts the exploration and exploitation capabilities of the algorithms for the best run among all runs. The BFBI algorithm performs quite a trade-off balance between exploration and exploitation. Figure 5(d) shows the best fitness values obtained from each run.

Average computational times required for solving the TEP problem by BFBI, GA, and PSO algorithms for N-1 contingency are presented in Table 2. It can be observed that the BFBI algorithm needs more computational time than GA and PSO algorithms. However, in the real case, since it takes many years (5, 10 years) for a transmission line to be planned, invested, and installed, the calculation times given in Table 2 for BFBI are acceptable for transmission system planners.

Based on the analyzed results, due to its powerful abilities, the BFBI algorithm can be accepted as an accurate, convergent, balanced, and effective metaheuristic optimization algorithm for solving the TEP problem.

4.2. Application to the Modified 400 kV Turkish Grid

This section includes evaluations of the adequacy of existing transmission lines and the optimal determination of new transmission lines needed to maintain the system balance for a realistic bulk grid in planning horizons. The BFBI, GA, and PSO algorithms are tested on the modified 400 kV Turkish grid for seasonal scenarios. Each season is built based on the actual generation and consumption values representing the entire grid to find optimal investment solutions in the 5- and 10-year planning periods. In the TEP problem, to converge this study to real planning conditions, the N-1 security criterion is taken into consideration. Furthermore, in addition to the investment cost, additional operational costs incurred in redispatching to ensure operational security are included in the objective function. Hence, the objective function includes penalties for both N and N-1 contingency cases to alleviate line congestions and prevent the formation of isolated parts in the grid. Overall results are assessed by comparison.

4.2.1. System Description

The Turkish transmission grid comprises approximately 75,000 km of transmission lines and 1500 substations at 400 kV and 154 kV. Since 2010, the Turkish grid has been synchronously connected to the CESA (The Continental Europe Synchronous Area) through 400 kV interconnection lines with Bulgaria and Greece. In addition to these connections, there are also interconnections with other neighboring countries at 400 kV and 154 kV. By the end of 2023, Türkiye’s total consumption reached 322 TWh, and the total installed capacity reached 107 GW [62]. Figure 6 shows distribution of total installed capacity and generation by resources in Türkiye by the end of 2023.

TEIAS is the transmission system operator in Türkiye and is responsible for secure and reliable transmission grid operation while maintaining its stability. One of the tasks included in this responsibility is to carry out long-term transmission system planning studies to ensure adequate transmission capacity for the increasing demand and generation of electricity in the future. Long-term transmission system planning studies are carried out within the scope of the Turkish grid code, which includes transmission grid design principles.

In this study, 400 kV line planning is considered as the main backbone of the Turkish transmission grid. Since the realization of 400 kV transmission line investments takes 5 and longer periods, 5- and 10-year planning periods are determined. In this context, the main objective of this study is to determine the optimal line investments in the 400 kV grid for 5- and 10-year planning periods.

As a first step, the reference grid used as a test system in this study has been created doing some modifications on the 400 kV Turkish grid. These modifications include merging some generation and load buses (substations) to form aggregated buses, and moving generation and load values at the 154 kV and 36 kV grids to the 400 kV grid using the bottom-to-up method. The reference grid topology consists of 187 buses and 319 lines. The reference grid’s total load and installed capacity are 53.2 GW (peak value of the year) and 100.6 GW, respectively. At the end of the 10-year planning period, 23 new generations and load buses will be added to the reference grid in the light of load and generation data for the current and planning periods. The number of candidate lines is 283, including 187 lines selected from existing ones and 96 new corridors. Weather conditions such as strong winds, heavy snowfalls, and ice loads can cause physical damage to transmission lines. Damage and destruction of transmission lines can directly affect system security. In order to increase system security, it is not preferred to install double-circuit transmission lines on a single pole and to install more than one transmission line from the same corridor. Thus, due to system security concerns, it is assumed that a single circuit line can be installed in each right-of-way. Figure 7 shows the corresponding single-line diagram of the reference grid.

Since the capacities (thermal capacity) of the lines in the grid vary seasonally, the values in the Turkish grid code are used in this study and the thermal loading thresholds for each line type are given in Table 4. For planning studies, it is assumed that the new line type to be added to the grid will be 3 bundle pheasant (3P), and its cost will be 1 M$/km.

Table 4. Thermal loading thresholds for 400 kV lines according to their type.

Line type	Thermal loading thresholds of the line (A)
	Spring	Summer	Fall	Winter
2R (rail)	1436	1200	1436	1962
2C (cardinal)	1450	1210	1450	1962
3C (cardinal)	2179	1830	2179	2987
3P (pheasant)	1634	2199	1634	3579
Underground cable (2000 mm2)	1500	1500	1500	1500
Submarine cable (1600 mm2)	1500	1500	1500	1500

4.2.2. Seasonal-Based Scenarios

To maintain the secure and stable operation of the grid in future planning horizons, TEP studies should be conducted by creating scenario sets that include generation and load conditions for different operating points in the grid [63]. In this context, eight scenarios have been created seasonally based on generation and load expectations for 5- and 10-year planning horizons. For each scenario, generation and load values are determined based on statistical data to reflect the seasonal characteristics. Thus, the optimum new line requirements are found by stressing the grid for different generation and load conditions. A comprehensive approach has been considered in defining and implementing the scenarios.

(1) Spring Scenario. In this scenario, while the minimum load (lowest condition) occurs in the spring months, hydro-based power plants, located in the north and east of the country, operate intensively. The effects of increasing the energy generation of hydroelectric power plants in the lowest load conditions on the grid are analyzed with this scenario.

(2) Summer Scenario. Since the maximum load (peak condition) in the Turkish grid occurs in the summer months, the summer scenario is created to determine the need for new lines by maximizing the interregional lines in the load flows from the regions with high generation to the regions with increased load. The summer scenario can be considered as the worst-case scenario.

(3) Fall Scenario. In the fall months, the load is higher than that in the spring, hydro generation is lower, and wind and solar generation is higher than that in the spring. The fall scenario determines the line requirements between regions when wind and solar generation and load are at the year’s average.

(4) Winter Scenario. In the winter months, the load is less than that in the summer months. Fossil-fired generation plants operate intensively to meet demands, thanks to the lack of hydro and other renewable generations. This scenario has led to the need for new lines between regions with fossil-based generation and regions with high load.

Possible differences are identified creating the scenarios. In this context, with these scenarios, it is aimed to cover the areas for all possible generation and load conditions of the grid. Table 5 summarizes the impact factors that cause stress to the grid in each scenario.

Table 5. Scenario-building synthesis.

	Spring scenario	Summer scenario	Fall scenario	Winter scenario
Load	−−	+++	+	++
Fossil fired generation	−−	++	+	+++
Hydro generation	+++	+	+	−−
Wind generation	+	+++	+++	++
Solar generation	−−	+++	+++	−

• + stands for moderate; ++ stands for high; +++ stands for highest; − stands for low; −− stands for lowest.

In seasonal-based scenarios, it is essential to determine generation and consumption profiles in which generation and consumption balance is achieved. For each scenario, the load value and the generation dispatch to meet this load value are obtained by considering the impact factors given in Table 5.

Figure 8 shows the total load values used for each scenario and it is assumed that the peak load will occur in the summer scenario and the minimum load in the spring scenario.

Figure 9 shows total installed capacity development by resources used for generation dispatch in the 5- and 10-year planning periods. At the end of the 10-year planning period, the installed capacity of the grid is assumed to be 141.7 GW. Compared to the reference grid, the installed capacity of fossil fuels (natural gas and coal excluded nuclear) is assumed not to increase, whereas RES (wind and solar) capacity is assumed to reach 48.3 GW. Considering the increase in RES capacity at the end of planning periods and RES generation contribution in each scenario, it is aimed to evaluate the impact of RES on TEP studies.

4.2.3. Optimization Results

The TEP problem is solved using the BFBI, GA, and PSO algorithms for the case where N-1 contingency is considered. Table 6 presents results, including optimal investment cost, operational cost, and total expansion line length to be added to the modified 400 kV Turkish grid. The results in Table 6 indicate that in some scenarios, the BFBI algorithm suggests cheaper solutions than the GA and PSO algorithms; however, in some scenarios, GA algorithm suggests cheaper solutions. The PSO algorithm suggests the most expensive solutions for all scenarios. Furthermore, for some scenarios, GA and PSO algorithms propose solutions with additional operational costs to achieve a less investment cost. This results in a less investment cost and shorter transmission lines, while the additional operational cost increases the total cost. The risk to system security is significantly increased when lines are overloaded and congested, which is an undesirable operational state for TSOs. In addition, forming isolated parts separated from the main grid is also undesirable for the system operation. These two crucial operational states cause additional operational costs incurred by TSOs in redispatch to ensure operational security, eliminate the line congestions, and prevent the formation of isolated parts in the grid. As an important output, it can be deduced that the results obtained with the BFBI and GA algorithms converge to the real states in real and large grids, while the PSO algorithm does not.

Table 6. Optimal expansion results.

Scenario	Planning period	Investment cost (M$)			Operational cost (M$)			Total expansion (km)
		BFBI	GA	PSO	BFBI	GA	PSO	BFBI	GA	PSO
Spring	5 year	2777.17	2727	10170.93	0	0	0	2777.17	2727	10170.93
	10 year	4440.71	4648.46	10652.32	0	0	0	4440.71	4648.46	10652.32
Summer	5 year	2646.45	2573	4205.8	0	1010.36	2020.72	2646.45	2573	4205.8
	10 year	5112.64	4205.8	12693.27	0	0	0	5112.64	4205.8	12693.27
Fall	5 year	1909.54	2069.25	2008.45	0	0	2000	1909.54	2069.25	2008.45
	10 year	3752.65	3787.5	9606.06	0	0	0	3752.65	3787.5	9606.06
Winter	5 year	1844.8	1843.9	3446.21	0	0	2000	1844.8	1843.9	3446.21
	10 year	4386.52	4098.66	11640.05	0	0	0	4386.52	4098.66	11640.05

Among the global optimal expansion solutions considering N-1 security for the Turkish case in the 10-year planning period, the fall scenario solved with the BFBI algorithm has the minimum total cost with 3752.65 M$; in contrast to this, the summer scenario solved with the BFBI algorithm has the maximum total cost with 5112.64 M$.

The solutions relevant to different seasons might include the same expansion investments. Therefore, it is necessary to list the same transmission lines available in different scenarios. Then, only different investments obtained for particular scenarios should be added to the base list. In this way, the total investment cost covering all possible seasons can be determined. Hence, at the end of the 5- and 10-year planning periods, to ensure the reliable and secure operation of the system for all scenarios, the total cost and length of new transmission lines required are 3132.31 km-M$ and 6436.9 km-M$, respectively. These results are offered by the BFBI algorithm as the best expansion plans for the modified Turkish grid. Figure 10 depicts the new grid with the optimum expansion plan for 5- and 10-year planning periods.

This study is significant in the sense of revealing the necessity of determining the optimum expansion plans for grid investments by considering various generation and consumption scenarios depending on the changes in electricity generation and consumption in line with national energy policies. In this context, it is evaluated that the results obtained for the Turkish grid can meet the adequate capacity requirements of the grid by covering necessary grid investments to meet the expected changes in generation and consumption for future planning periods.