Increased penetration of renewable resources and new loads have increased the uncertainty levels ... more Increased penetration of renewable resources and new loads have increased the uncertainty levels in low voltage distribution systems (LVDS). This requires considering LVDS planning as a stochastic problem. Low voltage photovoltaics (PV) hosting capacity (HC) calculation is such a planning problem. Traditionally, this is solved by using the iterative Monte Carlo method, which requires solving the power flow equations thousands of times. This paper proposes a chance-constrained optimization-based hosting capacity calculation technique, which eliminates the necessity of repetitive solving of power flow equations. General polynomial chaos expansion is used to translate the input uncertainties defined by probability density function to the hosting capacity of the network. Chance constraints are applied for the nodal voltages and thermal overloading. A case study for an actual LV feeder shows that the computational time of using the Monte Carlo based method is reduced from days to the or...
For market-based procurement of low voltage (LV) flexibility, DSOs identify the amount of flexibi... more For market-based procurement of low voltage (LV) flexibility, DSOs identify the amount of flexibility needed for resolving probable distribution network (DN) voltage and thermal congestion. A framework is required to avoid over or under procurement of flexibility in the presence of uncertainty. To this end, we propose a scenario-based robust chance-constrained (CC) day-ahead flexibility needs assessment (FNA) framework. The CC level is analogous to the risk DSO is willing to take in flexibility planning. Multi-period optimal power flow is performed to calculate the amount of flexibility needed to avoid network issues. Flexibility is defined in terms of nodal power ramp-up and ramp-down and cumulative energy needs over a full day for each node. Future uncertainties are considered as multiple scenarios generated using multivariate Gaussian distribution and Cholesky decomposition. These scenarios are utilized to solve the flexibility needs assessment optimal power flow (FNA-OPF) problem. Zonal clustering of an LV feeder is performed using electrical distance as a measure and spatial partitioning. The FNA tool calculates ramp-up and ramp-down flexibility's power and energy requirements. Energy and power needs are often valued differently in many energy markets. We identify the marginal value of flexibility associated with energy and power needs separately. From numerical results for an LV feeder, it is observed that zonal flexibility needs assessment is more immune to uncertainty than nodal flexibility needs, making it more useful for DSOs to evaluate day-ahead flexibility procurement. We also propose a Pareto optimal mechanism for selecting CC level to reduce flexibility needs while reducing DN congestion.
This paper introduces a decoupled method of calculating the PV hosting capacity of low voltage di... more This paper introduces a decoupled method of calculating the PV hosting capacity of low voltage distribution system (LVDS) feeders for new solar photovoltaics (PV) installations. The hosting capacity calculation of LVDS is a multidimensional stochastic problem. A general polynomial chaos-based probabilistic power flow is used to solve this problem, as it allows for fast computation times without any compromise in accuracy. Two types of uncertainties exist in the hosting capacity calculation problem: planning level uncertainties such as size, location, type, and number of PV installations and operational uncertainties such as consumer load and PV generation. These two types of uncertainties are usually sampled together in probabilistic hosting capacity approaches. In this paper, a decoupled approach is presented where the impact of planning level scenarios on the probability of violation of operational limits is studied. The highest total PV among the planning scenarios inside the probabilistic bound of the operational limit is termed as the hosting capacity of the feeder. The results show that hosting capacity depends upon the planning uncertainties and operational variables, and decoupling them gives a more concise picture of the impact of different uncertainties.
Proceedings of the Thirteenth ACM International Conference on Future Energy Systems
In low-voltage distribution networks, the integration of novel energy technologies can be acceler... more In low-voltage distribution networks, the integration of novel energy technologies can be accelerated through advanced optimizationbased analytics such as network state estimation and networkconstrained dispatch engines for distributed energy resources. The scalability of distribution network optimization models is challenging due to phase unbalance and neutral voltage rise effects necessitating the use of 4 times as many voltage variables per bus than in transmission systems. This paper proposes a novel technique to limit this to a factor 3, exploiting common physical features of lowvoltage networks specifically, where neutral grounding is sparse, as it is in many parts of the world. We validate the proposed approach in OpenDSS, by translating a number of published test cases to the reduced form, and observe that the proposed "phase-to-neutral" transformation is highly accurate for the common single-grounded low-voltage network configuration, and provides a high-quality approximation for other configurations. We finally provide numerical results for unbalanced power flow optimization problems using PowerModelsDistribution.jl, to illustrate the computational speed benefits of a factor of about 1.42. CCS CONCEPTS • Theory of computation → Continuous optimization; Network flows; • Software and its engineering → Domain specific languages; • Computing methodologies → Model verification and validation; • Applied computing → Multi-criterion optimization and decision-making.
Voltage unbalance in distribution networks (DN) is expected to grow with increasing penetration o... more Voltage unbalance in distribution networks (DN) is expected to grow with increasing penetration of single-phase distributed generation and single-phase loads such as electric vehicle chargers. Unbalance mitigation will be a significant concern as voltage unbalance leads to increased losses, reduced motor and inverter efficiency, and becomes a limiting factor for DN operation. The true definition of the unbalance metric needs phasor measurements of network voltage and current. However, such phasor measurements are generally not available in real life and as such approximate definitions are widely used due to their simplicity. This work aims to compare the true voltage unbalance definition and approximate unbalance metrics derived from phase voltage magnitude, as phase voltage magnitudes are commonly measured by digital metering infrastructure. For the comparison, multi-period power flow simulations are performed for 161 Spanish distribution feeders with R/X ratios varying from 2.87 to 14.68. We observe that phase magnitude-based unbalance metrics reasonably approximate the true unbalance for higher R/X ratios with a varying load power factor in a DN. Furthermore, the approximate unbalance metrics slightly improve for a low DN power factor due to the increase in DN unbalance. Therefore, the phase magnitude-based unbalance metric can be utilized for approximating DN unbalance.
2019 International Conference on Smart Energy Systems and Technologies (SEST), 2019
This paper discusses the methods of modelling a European style TT grounded low voltage (LV) distr... more This paper discusses the methods of modelling a European style TT grounded low voltage (LV) distribution system without increasing computational complexity and with increased accuracy. The conventional Kron's reduction neglects the effect of the neutral current return path through the neutral conductor. While a four-wire model gives accurate power flow results with the possibility to monitor the neutral voltage, it also increases the computational burden. In this paper, a novel reduction method is proposed for European style TT grounded low voltage distribution systems to an equivalent three-wire system without neglecting the impact of the neutral voltage.
Recent evolutions in low voltage distribution system (LVDS), e.g., distributed generation and ele... more Recent evolutions in low voltage distribution system (LVDS), e.g., distributed generation and electric vehicles, have introduced a higher level of uncertainty. To determine the probability of violating grid constraints, e.g., undervoltage, such system must be assessed using a probabilistic power flow, which considers these uncertainties. Several approaches exist, including simulation-based and analytical methods. A well-known example of the simulation-based methods is the crude Monte Carlo (MC) approach which is very common in scientific computation due to its simplicity. Recently, analytical methods such as the general polynomial chaos (gPC) approach have gained increasing interest. This paper illustrates the effectiveness of the gPC approach compared to the MC method in determining the uncertainty of certain grid measures. Both methods are compared with respect to computational time and accuracy using a small test case with stochastic input which coheres to a univariate continuous...
Increased penetration of renewable resources and new loads have increased the uncertainty levels ... more Increased penetration of renewable resources and new loads have increased the uncertainty levels in low voltage distribution systems (LVDS). This requires considering LVDS planning as a stochastic problem. Low voltage photovoltaics (PV) hosting capacity (HC) calculation is such a planning problem. Traditionally, this is solved by using the iterative Monte Carlo method, which requires solving the power flow equations thousands of times. This paper proposes a chance-constrained optimization-based hosting capacity calculation technique, which eliminates the necessity of repetitive solving of power flow equations. General polynomial chaos expansion is used to translate the input uncertainties defined by probability density function to the hosting capacity of the network. Chance constraints are applied for the nodal voltages and thermal overloading. A case study for an actual LV feeder shows that the computational time of using the Monte Carlo based method is reduced from days to the or...
For market-based procurement of low voltage (LV) flexibility, DSOs identify the amount of flexibi... more For market-based procurement of low voltage (LV) flexibility, DSOs identify the amount of flexibility needed for resolving probable distribution network (DN) voltage and thermal congestion. A framework is required to avoid over or under procurement of flexibility in the presence of uncertainty. To this end, we propose a scenario-based robust chance-constrained (CC) day-ahead flexibility needs assessment (FNA) framework. The CC level is analogous to the risk DSO is willing to take in flexibility planning. Multi-period optimal power flow is performed to calculate the amount of flexibility needed to avoid network issues. Flexibility is defined in terms of nodal power ramp-up and ramp-down and cumulative energy needs over a full day for each node. Future uncertainties are considered as multiple scenarios generated using multivariate Gaussian distribution and Cholesky decomposition. These scenarios are utilized to solve the flexibility needs assessment optimal power flow (FNA-OPF) problem. Zonal clustering of an LV feeder is performed using electrical distance as a measure and spatial partitioning. The FNA tool calculates ramp-up and ramp-down flexibility's power and energy requirements. Energy and power needs are often valued differently in many energy markets. We identify the marginal value of flexibility associated with energy and power needs separately. From numerical results for an LV feeder, it is observed that zonal flexibility needs assessment is more immune to uncertainty than nodal flexibility needs, making it more useful for DSOs to evaluate day-ahead flexibility procurement. We also propose a Pareto optimal mechanism for selecting CC level to reduce flexibility needs while reducing DN congestion.
This paper introduces a decoupled method of calculating the PV hosting capacity of low voltage di... more This paper introduces a decoupled method of calculating the PV hosting capacity of low voltage distribution system (LVDS) feeders for new solar photovoltaics (PV) installations. The hosting capacity calculation of LVDS is a multidimensional stochastic problem. A general polynomial chaos-based probabilistic power flow is used to solve this problem, as it allows for fast computation times without any compromise in accuracy. Two types of uncertainties exist in the hosting capacity calculation problem: planning level uncertainties such as size, location, type, and number of PV installations and operational uncertainties such as consumer load and PV generation. These two types of uncertainties are usually sampled together in probabilistic hosting capacity approaches. In this paper, a decoupled approach is presented where the impact of planning level scenarios on the probability of violation of operational limits is studied. The highest total PV among the planning scenarios inside the probabilistic bound of the operational limit is termed as the hosting capacity of the feeder. The results show that hosting capacity depends upon the planning uncertainties and operational variables, and decoupling them gives a more concise picture of the impact of different uncertainties.
Proceedings of the Thirteenth ACM International Conference on Future Energy Systems
In low-voltage distribution networks, the integration of novel energy technologies can be acceler... more In low-voltage distribution networks, the integration of novel energy technologies can be accelerated through advanced optimizationbased analytics such as network state estimation and networkconstrained dispatch engines for distributed energy resources. The scalability of distribution network optimization models is challenging due to phase unbalance and neutral voltage rise effects necessitating the use of 4 times as many voltage variables per bus than in transmission systems. This paper proposes a novel technique to limit this to a factor 3, exploiting common physical features of lowvoltage networks specifically, where neutral grounding is sparse, as it is in many parts of the world. We validate the proposed approach in OpenDSS, by translating a number of published test cases to the reduced form, and observe that the proposed "phase-to-neutral" transformation is highly accurate for the common single-grounded low-voltage network configuration, and provides a high-quality approximation for other configurations. We finally provide numerical results for unbalanced power flow optimization problems using PowerModelsDistribution.jl, to illustrate the computational speed benefits of a factor of about 1.42. CCS CONCEPTS • Theory of computation → Continuous optimization; Network flows; • Software and its engineering → Domain specific languages; • Computing methodologies → Model verification and validation; • Applied computing → Multi-criterion optimization and decision-making.
Voltage unbalance in distribution networks (DN) is expected to grow with increasing penetration o... more Voltage unbalance in distribution networks (DN) is expected to grow with increasing penetration of single-phase distributed generation and single-phase loads such as electric vehicle chargers. Unbalance mitigation will be a significant concern as voltage unbalance leads to increased losses, reduced motor and inverter efficiency, and becomes a limiting factor for DN operation. The true definition of the unbalance metric needs phasor measurements of network voltage and current. However, such phasor measurements are generally not available in real life and as such approximate definitions are widely used due to their simplicity. This work aims to compare the true voltage unbalance definition and approximate unbalance metrics derived from phase voltage magnitude, as phase voltage magnitudes are commonly measured by digital metering infrastructure. For the comparison, multi-period power flow simulations are performed for 161 Spanish distribution feeders with R/X ratios varying from 2.87 to 14.68. We observe that phase magnitude-based unbalance metrics reasonably approximate the true unbalance for higher R/X ratios with a varying load power factor in a DN. Furthermore, the approximate unbalance metrics slightly improve for a low DN power factor due to the increase in DN unbalance. Therefore, the phase magnitude-based unbalance metric can be utilized for approximating DN unbalance.
2019 International Conference on Smart Energy Systems and Technologies (SEST), 2019
This paper discusses the methods of modelling a European style TT grounded low voltage (LV) distr... more This paper discusses the methods of modelling a European style TT grounded low voltage (LV) distribution system without increasing computational complexity and with increased accuracy. The conventional Kron's reduction neglects the effect of the neutral current return path through the neutral conductor. While a four-wire model gives accurate power flow results with the possibility to monitor the neutral voltage, it also increases the computational burden. In this paper, a novel reduction method is proposed for European style TT grounded low voltage distribution systems to an equivalent three-wire system without neglecting the impact of the neutral voltage.
Recent evolutions in low voltage distribution system (LVDS), e.g., distributed generation and ele... more Recent evolutions in low voltage distribution system (LVDS), e.g., distributed generation and electric vehicles, have introduced a higher level of uncertainty. To determine the probability of violating grid constraints, e.g., undervoltage, such system must be assessed using a probabilistic power flow, which considers these uncertainties. Several approaches exist, including simulation-based and analytical methods. A well-known example of the simulation-based methods is the crude Monte Carlo (MC) approach which is very common in scientific computation due to its simplicity. Recently, analytical methods such as the general polynomial chaos (gPC) approach have gained increasing interest. This paper illustrates the effectiveness of the gPC approach compared to the MC method in determining the uncertainty of certain grid measures. Both methods are compared with respect to computational time and accuracy using a small test case with stochastic input which coheres to a univariate continuous...
Uploads
Papers by Arpan Koirala