Understanding PCS Architectures for Utility-Scale BESSes

Utility-scale battery energy storage systems (BESSes) are increasing in power rating and operating complexity. But as projects move from small installations to plants rated at hundreds of megawatts, the power-conversion system (PCS) serves as more than an inverter between the battery and the grid.

A BESS is divided into battery racks, DC collection buses, PCS units, transformers, switchgear, and protection zones that determine how power moves through the plant and how much of the system is forced to behave as a single electrical unit.

There is no one-size-fits-all PCS architecture. There are three approaches in designing the PCS—centralized, distributed, and hybrid—each with tradeoffs across conversion losses, usable battery energy, equipment count, maintenance, and control granularity.

How different PCS architectures divide the power block

The electrical path through a utility-scale BESS is similar, irrespective of the PCS architecture. Battery cells are grouped into modules, modules into racks, and those battery groups feed a bidirectional conversion system.

What changes from one architecture to another is where those battery groups are electrically combined, where independent conversion begins, and how much equipment sits behind each common electrical boundary.

Centralized PCS architecture

In a centralized PCS architecture, a large bidirectional inverter is used for multiple battery racks connected to a common DC bus. Multiple battery racks are paralleled through DC collection equipment and feed a common high-voltage DC bus (in the 1,500-V class) before reaching a multi-megawatt PCS (Figure 1).

The centralized approach does not mean that the entire plant relies on a single inverter. In large BESS projects, the plant is divided into multiple power blocks, each with its own central PCS. These converters are rated in the multi-megavolt-ampere range, with several units operating in parallel.

The architecture concentrates conversion equipment into fewer locations. For instance, a central PCS may be mounted on a skid with associated AC switchgear and a medium-voltage transformer, while battery containers located some distance away are connected to it via DC feeders.

Because many battery racks share the same DC connection, the central PCS must handle high current before converting it to AC. Using a higher DC voltage reduces that current and the associated power losses. Even so, multi-megawatt systems still require large cables, busbars, fuses, and disconnects on the DC side.

Distributed PCS architecture

In a distributed PCS architecture, the conversion boundary is closer to the battery. The plant divides the battery racks among many smaller, independently controlled converters.

A converter may serve one rack, one battery cluster, or another subdivision within a container. Its AC output is then combined with neighboring PCS units on a local, low-voltage AC bus before a transformer raises the voltage for the medium-voltage collector. This keeps high-current DC paths short because the DC/AC conversion takes place closer to the battery. Power is then collected on the AC side and, after the local transformer, at medium voltage, where substantially less current is required for the same megawatt transfer.

One commercial implementation packages 10 or 12 200-kW PCS units into a roughly 2- to 2.4-MW skid together with low-voltage equipment, communications, and a medium-voltage transformer (Figure 2).

The tradeoff is equipment count, whereby a distributed plant uses many more converters, controllers, sensors, cooling components, and communication interfaces than a centralized design.

Hybrid PCS architecture

In this approach, battery groups are connected via bidirectional DC/DC converters that feed a common DC link and a shared DC/AC inverter.

The DC/DC stage may be isolated or non-isolated and designed for full or partial power processing. Its role is to decouple each battery group’s operating point from the common DC-link to allow individual groups to be controlled.

The power block has two levels of granularity. Battery control and specific protection functions can be partitioned at the rack or cluster level, but the AC conversion, transformer capacity, and portions of the cooling or auxiliary system remain shared.

Therefore, the architecture determines how many batteries share a DC operating point, where power can be independently controlled, how far high-current DC must travel, and how much equipment is grouped into repeatable power blocks.

How PCS architecture influences efficiency and availability

It is important to note that PCS efficiency cannot be analyzed from the peak number on the datasheet alone.

Utility-scale BESS plants operate across varying power levels, so annual performance depends on semiconductor losses, DC cabling, transformers, medium-voltage collection, auxiliaries, and standby consumption.

Part-load operation is very important for services such as frequency regulation or renewable smoothing. For example, a modular plant can improve weighted efficiency by operating fewer PCS blocks closer to their efficient loading range while placing others in low-loss standby. This is beneficial if response time and grid-support requirements allow units to be switched off.

Hybrid architectures introduce another tradeoff. Rack- or cluster-level DC/DC conversion adds another conversion stage and additional semiconductor and magnetic losses. In return, it gives battery groups current or voltage control.

This flexibility is valuable, as racks diverge in state of charge, capacity, resistance, temperature, and state of health. When several racks are tied to a common DC bus, their usable operating range can become constrained by the first group to reach a BMS limit.

Availability also requires a system-level view. More converters do not make a distributed plant less available—for example, assuming a 100-MW plant built from 4-MW PCS blocks can lose 4% of rated power after one complete PCS outage. On the other hand, a 200-kW converter failure in a more distributed architecture removes only 0.2%.

The practical metrics are capacity loss, repair time, and the ability to keep unaffected blocks operating. In addition, field-replaceable modules, on-site spares, and N+1 converter capacity can limit the loss of capacity when equipment fails. The redundancy works only if batteries, transformers, and collection equipment allow the hardware to carry the load.

The comparison is among conversion efficiency, usable battery energy, and plant availability over the system’s operating life.

Therefore, selecting the right PCS architecture for utility-scale battery storage is a system-optimization problem. The design must balance conversion efficiency, battery utilization, equipment count, fault-domain size, maintainability, and control complexity throughout the plant’s life.

Cover image: Sungrow Power

The post Understanding PCS Architectures for Utility-Scale BESSes appeared first on Power Electronics News.