By Vince Martinelli
Many distribution utilities are faced with the challenge of integrating increasing amounts of distributed energy resources (DER). These DERs, especially smaller-scale (e.g. <10 kW) solar photovoltaics (PV) connected to the low voltage (LV) secondary network, however, can place stress on both medium voltage (MV) primary distribution feeders and the LV secondary circuits to which they are directly connected. The primary MV feeder side also faces challenges including increased voltage rise and/or fluctuation; increased reverse real power flow (excessive loading), greater use of existing mechanical equipment such as voltage regulators, capacitor banks and load tap changers; the need to replace, reprogram and/or desensitize protection equipment (e.g. because of changes in fault currents); and degradation of power factor (PF) and power quality.
|Figure 1. Double-chain secondary network|
While there have been many studies focusing on these MV challenges, they have generally ignored the local impacts of small-scale residential PV. Hosting capacity also critically depends on the secondary circuit construction and the constraints it places in terms of thermal limits and local voltage rise seen at the actual customer point of common coupling (PCC). Dynamically neutralizing this variable voltage rise can eliminate one of these constraints, significantly increasing overall hosting capacity.
Reducing Voltage Rise Enhances Residential Photovoltaic Hosting Capability
Unlike commercial and utility-scale PV, which often undergo rigorous connection screens, residential PV screens are typically based on aggregate feeder constraints, such as whether the addition of a particular residential PV installation will violate the constraint of PV generation exceeding 20 percent of the feeder's minimum daytime loading. While these aggregate screens are useful in minimizing feeder-wide risks, they rarely consider local voltage issues, such as whether a newly added PV asset (along with all other PV) attached to a secondary line will cause any PCC voltage violation because of local voltage rise across secondary wiring and distribution transformers.
This screen is not often performed for two reasons. First, it is rare that the construction of the secondary circuit is explicitly known and captured in utility circuit model databases. Second, the voltage at the residential PCC depends on the maximum feeder voltage at that specific distribution transformer, which is often difficult to discern from MV feeder models unless solid GIS and phase models are maintained and loading conditions are well known. With low residential PV penetration and in relatively short secondary circuits, secondary voltage rise is rarely a concern and can safely be ignored. For secondary circuits with high PV penetration and/or for longer secondary circuits, however, excessive voltage rise may occur. If this is not addressed, it will lead to high voltage violations and potential customer complaints if smart inverters or other equipment trip off or become damaged because of high voltage.
Defining Low Voltage Secondary Hosting Capacity
To better understand secondary hosting capacity, it's helpful to look at an example. Figure 1 shows a secondary network split into two runs, each with four homes (nodes).
For purposes of this study, the hosting capacity of this secondary network is defined as the number of nodes (out of a maximum of eight) that can each host 8 kW of PV. There are two constraints on the hosting capacity:
- Voltage constraint: The voltage at any PCC must not exceed 1.05 p.u. (5 percent above the nominal service voltage).
- Capacity constraint: The reverse power flow through the transformer must not exceed 125 percent of the transformer's kVA rating. Likewise, the current through each conductor must not exceed 125 percent of its ampacity rating. (The 125 percent factor has been chosen to reflect the widespread practice of allowing slight overloading at peak times).
These two constraints share the same worst-case scenario: that of peak reverse power flow, i.e., maximum PV output (8 kW per node with PV) and light load (defined as 0.5 kW per node in this example). With respect to the capacity constraint, the transformers and conductors in each example have been sized to allow all eight nodes to each host 8 kW PV at unity PF.
With respect to the voltage constraint, the PCC voltage is the sum (in p.u.) of the MV voltage and the secondary voltage rise caused by reverse power flow through the conductors and transformer. While the secondary voltage rise can be calculated from the example, the MV voltage is an independent variable, determined by primary-feeder-wide events and largely unaffected by what happens in the secondary network. The hosting capacity, therefore, is best described as a function of the MV voltage delivered to this secondary network.
Figure 2 shows the hosting capacity of this split secondary network. The "base" case represents the network as presented in Figure 1. As the bars for the base case show, hosting capacity rapidly decreases from eight nodes (at MV = 1 p.u.) to zero nodes (at MV = 1.04 p.u. or above) because of the voltage constraint. In other words, if this network were situated at a point in the primary feeder with 1 p.u. MV voltage, its hosting capacity is already at maximum (all eight nodes can host PV) without any further enhancements. If this network were situated at a point in the primary feeder where MV voltage is expected (or modeled) to be high during times of feeder-wide maximum PV and light load, however, then its hosting capacity will be severely limited, and various mitigation strategies must be considered.
The other bars in Figure 2 represent several sample mitigation techniques:
- XF 100--upgrading the transformer from 50 kVA to 100 kVA;
- cond 4/0--upgrading the main secondary conductor from 1/0 to 4/0;
- XF + cond--upgrading both the transformer and the main secondary conductor as in the previous techniques;
- PV PF 0.9 (or 0.8)--programming each PV smart inverter to operate at fixed PF of 0.9 or 0.8, i.e., sourcing real power but consuming reactive power (in an attempt to bring down voltages); and
- IPR--deploying Gridco Systems' In-line Power Regulators (IPRs).
Each mitigation technique improves hosting capacity, but their effectiveness varies widely, with the transformer upgrade being least effective and the IPR being most effective in this example. Note that in the cases of PV PF 0.9 or 0.8, the additional reactive power increases current at the (50 kVA) transformer, limiting the hosting capacity (to seven or six nodes, respectively) because of the high apparent power capacity of the transformer at low MV voltages.
|Figure 2. Hosting capacity of the double-chain network|
As shown in Figure 2, the IPR achieves maximum hosting capacity (i.e., all eight nodes can host PV), regardless of MV voltage. This is because the IPR acts as a fast LV voltage regulator, providing sub-cycle and continuous series voltage regulation across a wide +/- 10 percent range. In other words, as long as the MV voltage is 0.90-1.10 p.u., the IPR (installed at the LV output of the transformer) can keep its output (load-side) LV voltage at 1 p.u. When added to the worst-case voltage rise (because of secondary conductors only, since the IPR sits behind the transformer), this regulated voltage maintains all PCC voltages below 1.05 p.u. In essence, the IPR can decouple the MV voltage from the LV voltage, entirely removing the voltage constraint.
In contrast, other mitigation techniques typically lower the PCC voltage by some 1 percent to 2 percent. As a result, they can only increase PV hosting capacity when the MV voltage is not too high. In the split secondary example, these other techniques are marginally useful at MV = 1.03 p.u. and completely ineffective at MV = 1.04 p.u. or above. Similar results are obtained for other secondary topologies, including a non-split secondary network (i.e., all nodes on one secondary branch) and a star network (i.e., direct buried).
In reality, mitigation options must be selected based on both the MV voltage and the upgrade costs involved. When operating PV at non-unity PF, one should also consider the need to source and transmit additional dynamic VARs--from somewhere else in the primary feeder/transmission network--and the resulting additional resistive loss. In addition, operating the smart inverters at 0.9 or 0.8 PF during a time of peak real power output means the inverter kVA rating must be oversized by 111 percent or 125 percent, respectively, compared to the PV kW rating. Various techniques can also be combined (e.g., transformer upgrade and operating PV smart inverters at PF 0.8), with costs being roughly additive, while the combined effectiveness can be calculated. The IPR is a special case. Because it already eliminates the voltage constraint--by essentially decoupling LV and MV voltages--it doesn't need to be combined with other mitigation techniques except to alleviate capacity constraints.
Other Benefits of the In-line Power Regulators
While this article focuses on the most common barrier to increasing secondary PV hosting, i.e., the PCC voltage constraint during peak reverse power flow, there are other barriers that can be addressed by the IPR. The IPR's sub-cycle voltage regulation, for example, can effectively minimize voltage fluctuations caused by sudden changes in loads and/or solar irradiance. In addition, the IPR has various harmonic cancellation capabilities and can increase PV hosting capacity in secondary networks where excessive harmonic content is the limiting factor.
There are many potential technical barriers to accepting a high penetration of distributed PV sources, and not all of them can be overcome by the IPR. For the most common scenario where excessive voltage rise or fluctuation is the main obstacle to increasing PV penetration, however, the IPR offers a cost-effective solution because of its sub-cycle, dynamic and broad +/-10 percent series voltage regulation capability. In comparison, other approaches such as transformer upgrades, reconductoring, or operating PV smart inverters at 0.9 PF can only achieve some 1 percent to 2 percent voltage improvement, making them useful only when resolving small voltage violations. The IPR effectively decouples the MV voltage from the LV voltage and eliminates any voltage concern caused by high penetration of PV in a secondary network, thereby maximizing hosting capacity.
About the author: Vince Martinelli is the director of Product Management and Solution Engineering for Gridco Systems. He is responsible for the product roadmap and developing services that leverage power systems modeling and data analytics to build compelling business cases for utilities. Before entering the electric power industry, Martinelli led similar practices helping to introduce new technologies and products for successful start-up companies in telecommunications and e-commerce.