Extending Grid Capacity: How Phase Balancing Delays Expensive Transformer Upgrades

Introduction

As the demand for energy continues to rise due to electrification of heating and rise of electric vehicles, as the global economy is moving away from fossil fuels, the electricity grid, especially distribution networks are becoming increasingly complex, dynamic, and burdened. Our low-voltage networks, the backbone of energy distribution, face unprecedented challenges. 

The escalation in demand and renewable energy sources, such as residential solar, not only complicates the grid but also increases the necessity for costly and time-consuming cable and transformer upgrades. However, given the high costs and time-consuming nature of asset upgrades, these should be viewed as a last resort—especially when low-voltage networks are already the most expensive segments to maintain owing to their vast size.

As cases of limited capacity in distribution networks grow, system operators are finding themselves under increasing pressure. Power quality phenomena such as voltage violations (voltage swells and sags) are becoming more and more common. The conventional route—upgrading assets through cable replacement, reconductoring, or new transformer installations—often takes considerable time, largely due to the lengthy permit acquisition process rather than financing. 

To address these delays, operators have been exploring a broad range of non-conventional measures that can extend existing capacity and postpone the need for major asset upgrades by a few years. This additional time typically allows for more strategic planning of eventual upgrades while also ensuring that both human and financial resources are directed toward the most critical needs first.

Phase load balancing is one such non-conventional solution to constricted capacity scenarios. Phase load imbalance is caused by single-phase loads in low-voltage networks. This often results in one phase nearing its thermal or voltage capacity limits, while the remaining phases stay underutilized as depicted in Figure 1. By statically redistributing loads across phases to balance utilization, additional headroom can be released and peak loading reduced, thereby deferring expensive asset reinforcements.

This solution, while not applicable in every scenario, is cost effective and can be implemented quickly. In the following sections we will take a look at how such a process can be implemented - specifically how the constricted capacity scenarios are evaluated, what are the necessary prerequisites for process implementation and how Dewesoft tools can streamline and simplify the process.

Capacity constraints in low-voltage networks

At any point in a low-voltage network, the capacity to consume or supply power is limited. This limitation occurs due to two main factors:

1. First, the thermal (current) limits of network assets can restrict power flow.

2. Second, voltage rise or sag at that point can impose a constraint. The extent of this voltage change depends on the impedance of the power delivery network.

Figure 2 shows a simplified distribution network from primary substation to the end customers. The only points where the voltage profile is discontinuous are at the voltage regulator on the primary substation transformer and at the tap changing transformer at the secondary substation.

In a low-voltage network, the ability to consume or supply power is always limited by certain factors.

1. One key factor is the thermal (current) limits of network components, which can restrict the amount of power flowing through them.

2. Another factor is voltage rise or sag at a specific point in the network. This occurs due to the impedance of the power delivery system, which affects how voltage changes as power flows.

Capacity constraints through asset investment planning

When Distribution System Operators (DSOs) encounter capacity constraints, they typically evaluate these scenarios through the lens of asset investment planning. In cases driven by voltage limitations, the focus often falls on specific low-voltage (LV) feeders; however, when the constraint is current-related, attention shifts to either LV feeders or distribution transformers (DTs). 

This application note concentrates on thermal constraints in DTs. Standard power ratings in distribution networks are typically 35kVA, 50kVA, 100kVA, 160kVA, 250kVA, 400kVA, 630kVA, 1000kVA, 1250kVA and 1600kVA. When a capacity issue arises, the DSO usually considers installing a unit one or two sizes larger than the existing one. 

Another key factor is transformer mounting - units below 400 kVA are typically pole-mounted (Figure 4), whereas those 400 kVA and above are pad-mounted (Figure 5). Upgrading from a pole- to a pad-mounted transformer entails considerably higher costs and can extend the project timeline by two years or more due to the need for additional permits. 

Table 1 summarizes the cost estimates for each potential upgrade, including both equipment and labor.

Evaluation of constrained cases

We analyzed the phase load currents of 1,564 distribution transformers (DTs) at the customers network, representing roughly 60% of all DTs in their network. Our evaluation process identifies any continuous time interval where at least one phase current exceeds 80% of the transformer’s nominal line capacity (calculated as nominal power / 3 / nominal voltage) for a duration of at least one hour. Whenever such a high-load interval is detected, we compute the average phase and neutral currents exclusively from data within that interval.

Among the 1,564 DTs analyzed, 415 met these criteria and were flagged as candidates for load balancing. For each candidate, we then estimated both the utilization and the imbalance levels, along with the expected time horizon for necessary transformer replacement.

We also calculated the Additional Reinforcement Cost (ARC), which reflects the extra capital investment needed to upgrade a DT earlier than in the case the loads were phase balanced. By deferring that upgrade through effective load balancing, a DSO can achieve measurable cost savings, captured in the difference between the “balanced” and “imbalanced” replacement timelines.

Transformer utilization levels in both balanced and imbalanced scenarios can be estimated via the following formula:

A. Transformer Utilization (Balanced)

B. Transformer Utilization (Imalanced)

Here S3ϕ is the three‐phase total power, Sn​ is the neutral‐line power, and C is the transformer’s rated capacity. 

We then calculate the time horizons Tbal and Timb for a DT replacement under each scenario by applying an annual load growth rate g (typically 3%). 

Finally, we determine the Additional Reinforcement Cost (ARC) by taking the difference in net present values (NPVs) between balanced and imbalanced cases considering the discount rate d (taken at 4%). This calculation includes both standard replacement expenses (based on Table 1) and any added cost of transitioning to a pad-mounted unit. Essentially, ARC captures the cost that would have been avoided if no early reinforcement were necessary. 

C. Time Horizon (Balanced)

D. Time Horizon (Imbalanced)

E. Present Value of Reinforcement Cost (Balanced)

F. Present Value of Reinforcement Cost (Imlanced)

Several examples of high-load DTs selected by the algorithm—along with their extended operating horizons and corresponding ARC values—are shown in Figures 5,6,7, and 8. Of the 415 candidate DTs, 377 were found to have a time-horizon difference of more than two years when comparing balanced versus imbalanced scenarios; these 377 were subsequently prioritized for load balancing. Altogether, their combined ARC amounted to roughly €750,000.

A pie chart of total ARC distribution per transformer rating depicted in Figure 9, reveals that a striking 46% of the overall ARC stems from just 24 units rated at 250 kVA. This cost spike occurs primarily because a 250 kVA transformer must be upgraded to a 400 kVA unit, which shifts from pole to pad mounting. This brings added expenses from construction work, permitting, and trenching. In some instances, DSOs bypass the intermediate 250 kVA rating entirely, moving directly from 160 kVA to 400 kVA. This would contribute even more to the ARC but was not included in this analysis.

One noteworthy point is that balancing a transformer also inherently balances the associated LV feeders. When a DT is nearing its capacity limit, at least one or more LV feeders is typically approaching its own limit as well, so there can be significant ARC implications on the feeder side. 

This effect is especially pronounced in larger units (>400 kVA) because, although the aggregate load on the DT tends to be more statistically balanced, each individual feeder serves a smaller subset of customers (usually 10–30). Smaller groups of customers are more prone to phase imbalance, amplifying the need for targeted balancing measures on those feeders.

The cost for each load reconnection can vary widely, mainly due to differences in on-site complexity. A straightforward task might take only 15 minutes, while more involved work requiring an aerial platform could last up to an hour. As a result, individual reconnection costs range from roughly €50 to €500. 

Besides the direct costs, timing is a major concern. While upgrading a transformer right away may seem ideal, the process of securing permits and completing construction can cause delays of a year or more.

To bridge this gap, load balancing offers a practical solution. It helps maintain stable power quality for users until a full upgrade becomes possible.

Real time electrical phase identification on the field using Gridphase

The customer deployed Gridphase with a global phase model to streamline single-phase load reconnections and other network operations. The geographic information system (GIS) aligns with the global phase framework (see Figure 10) and actual electrical phases. This connection allows for faster planning and field execution while reducing identification errors.

A global phase reference creates a consistent identification system across the entire distribution network—from high-voltage busbars to low-voltage loads.

By aligning transformers, feeders, and meters to the same absolute phase designation, the customer can detect and correct load imbalances using data analysis and three-phase power-flow calculations.

This system also helps mitigate single-phase faults, eliminates confusion from inconsistent local phase labels, and simplifies maintenance procedures with a clear phase reference.

The result is better planning, targeted load balancing, and lower costs and downtime across the grid.

The video shows the Gridphase using the network model in action in the secondary substation.

Why no feeder identification?

Once a feeder is correctly assigned and documented in the GIS, there is generally no need to re-verify it in the field. Its routing and connections rarely change and it is easily distinguished by its unique geographical path. In contrast, although phase data can also be stored in the GIS, field measurement remains crucial. Since color-coding and other visual markers cannot reliably separate individual phases (all of which reside in a single low-voltage cable), on-site measurement is necessary to ensure accurate phase information and correct execution of any reconnection operations.

Why are static reconnections enough?

In this low-voltage network, each single-phase customer is limited to a maximum capacity of 1 × 35 A. Consequently, the worst-case phase imbalance under static reconnections is capped at 35 A. This indicates that the most significant instances of phase imbalance arise from systematic factors, which can be addressed by physically redistributing loads among phases.

Other methods can also be used, such as static balancers (e.g., shunt zig-zag transformers) to reduce neutral current by lowering zero-sequence impedance. However, in many cases, simple static reconnections provide enough benefit to extend operations for a few more years. By alleviating near-term capacity constraints, static reconnections effectively prolong asset life, deferring expensive upgrades and giving utilities extra time for more strategic planning.

Conclusion

Static reconnections offer a practical, cost-effective strategy for improving load balance and enhancing the hosting capacity of low-voltage networks. By leveraging smart meters that grids are equipped with, and measurement data analytics, this approach not only tackles current constraints but also provides a scalable means to accommodate future load growth.

Most non-conventional methods for extending hosting capacity—including static reconnections—tend to have shorter effective lifespans compared to complete asset replacements. Nonetheless, phase load balancing remains a valuable tool in the distribution operator’s toolkit, particularly for situations where rapid, low-cost interventions can defer large-scale infrastructure upgrades. By incorporating solutions like Gridphase, operators gain accurate phase information, reduce fieldwork errors, and ensure that any changes made are seamlessly reflected in network models.

Through these proactive steps, network assets can remain in service longer, allowing DSOs to strategically plan capital investments and prioritize critical upgrades. Moreover, integrating routine load reconnections into field operations can be done with minimal added overhead. Altogether, this measured approach helps maintain power quality, manage rising demand, and ultimately deliver a more resilient, efficient power distribution system.