Why Motor Starting Can Limit Battery Backup Systems

A battery can hold enough energy to run a pump for an extended period and still be unable to start its motor. Starting is a short, demanding interaction between the inverter, the electrical network and the driven machine. Stored kilowatt-hours describe an energy reserve; they do not determine the current or torque available during acceleration.

For a site restarting after an outage, three different outcomes matter: establishing an AC supply, accelerating the motor, and keeping it running when the process load arrives. Research on current-limited inverters shows why success at one stage does not guarantee the next. It also explains why a larger battery or a slower start is not, by itself, a complete answer.

The limiting link can be the inverter, even with energy in reserve

A grid-forming inverter can establish its own voltage and frequency reference. A grid-following inverter normally follows an existing AC waveform. That distinction matters when the supply is dead, but “grid-forming” alone says little about how much starting current a particular inverter can deliver or how it behaves when that limit is reached.

Semiconductor devices need protection from excessive current. When the inverter reaches its current limit, its controls must change the output instead of continuing to act as an unconstrained voltage source. The resulting voltage response affects the motor at the same time that the motor is demanding current. The battery’s energy reserve can remain substantial while this AC-side interaction prevents acceleration. NREL, 2025 guide, pp. 2–3.

This is also why steady running power is an incomplete starting specification. A motor drawing an acceptable load after it reaches speed may require a very different electrical supply while stationary. The starting method, the current limit and its permitted duration, voltage at the motor terminals, and mechanical load all affect that transition.

Limiting current can reduce the torque needed to escape the limit

An induction motor accelerates when its electromagnetic torque exceeds the opposing load torque. The difference supplies the acceleration; the inertia of the motor and attached equipment determines how quickly speed rises. If available torque falls below the mechanical demand, a long ramp cannot make the machine accelerate through that part of its operating range.

Lower terminal voltage reduces induction-motor torque capability during acceleration. At a fixed supply frequency and the same motor speed or slip, torque has an approximately square-law dependence on voltage. This makes voltage reduction a tradeoff: it can reduce starting current, but it also weakens the torque available to move the load. The U.S. Department of Energy’s motor sourcebook explains the resulting slower acceleration and reduced resistance to stalling. DOE, printed p. 85.

For a current-limited supply, these effects are coupled. A large starting demand can depress voltage; the lower voltage reduces accelerating torque; the motor then remains in the demanding low-speed condition for longer. Current protection can work as intended while the overall start still fails. The useful question is therefore not just how high the current rises, but whether the machine can pass through the whole acceleration period.

A simulated motor reached speed, then stalled when loaded

Jain and colleagues’ 2020 NREL study modeled a 1.5 MW induction motor supplied by a simplified, current-limited grid-forming source. The motor model used a 6.6 kV voltage base and a 1.8 MVA power base. It compared a source beside the motor, remote supply through transformers and a line, and three sources operating together. These were simulations of a large motor, not tests of a commercial storage cabinet. Study setup, printed pp. 2–3.

In the single-source cases, the current limit was initially set to 1.2 times the motor’s rated-current base. A separate comparison allowed ten times that base, so the model could supply the required transient without the same constraint. The latter is a modeling comparison, not a claim that an ordinary inverter can provide ten times rated current.

The study’s Table III gives a motor-terminal voltage of 0.29 per unit for the on-site, initially unloaded 1.2× case. Per unit expresses a value relative to its base; 0.29 means 29% of that base. Here it is the reported fundamental-frequency magnitude derived over the acceleration interval, not a single instantaneous minimum or a universal acceptable voltage. The depressed voltage helps explain why the simulated motor took much longer to accelerate. Table III, printed p. 3.

The decisive result appears after acceleration. At 50 seconds, applying the mechanical load caused the single-source motor to stall. Raising the modeled current limit from 1.2 to 1.5 times the rated-current base allowed it to sustain that delayed load in the comparison shown in Fig. 7. The lesson is that current availability affects both acceleration and the transition into loaded operation. Neither multiplier is a general sizing rule or a permissible setting for another inverter. Results, printed p. 4.

Unloading, parallel sources and voltage ramps change different constraints

An unloaded start reduces the mechanical torque the motor must overcome. That can help acceleration, but the operating process must actually permit the load to be delayed. A machine that must start against pressure or move an already loaded mechanism is a different duty from the zero-load case. Even where unloading is possible, the later application of load remains a separate electrical event, as the simulation demonstrates.

Adding sources changes the electrical supply. In the study’s three-inverter case, the initially unloaded motor reached the defined starting speed in 3.40 seconds, with each source limited to 1.2 times its own rated current. Each source retained the rating matched to the motor in the single-source case, so this increased total source capacity; it did not divide the original capacity among three smaller units. But applying mechanical load from the beginning led to a low-speed, unstable operating point instead. More sources improved one starting sequence without making all sequences successful. The model also assumed ideal current sharing; real parallel systems depend on their controls, synchronization and network. Table III, Fig. 8 and limitations, printed pp. 3–5.

A further simulation gradually increased the source voltage. The authors reported suppression of the initial inrush and successful support of the later-applied load. This is evidence that the voltage trajectory can change the outcome within their model. It is not evidence that any soft starter, ramp setting or storage inverter will reproduce the result. The ramp was part of the modeled source behavior, and the authors still identified transformer saturation and detailed controls as areas requiring more complete treatment. Fig. 9 and limitations, printed pp. 4–5.

A variable-frequency drive is another distinct mechanism. It changes both the frequency and voltage supplied to the motor, rather than simply applying a lower voltage at a fixed frequency. The DOE sourcebook describes that distinction and the different demands of variable-torque and constant-torque loads. A drive-fed motor therefore has a different starting path from a directly connected motor; its input supply and operating limits remain part of the system. The NREL cases above do not establish compatibility with a particular drive. DOE, printed p. 89.

A successful current limit is not yet a stable power system

The 2020 model deliberately simplified several difficult details. It assumed zero no-load torque, omitted motor magnetic saturation, and used linear transformer models. Actual machinery still has friction even when the process load is removed. A marginal torque surplus in the simulation may therefore be insufficient in a real start. The paper also identifies motor thermal protection and protection along the supply path as follow-on work, especially when acceleration takes longer. Limitations, printed p. 5.

Later research adds a control-system reason for caution. NREL’s 2025 guide explains that current limiting changes an inverter’s dynamic behavior and can affect synchronization, voltage support and system protection. The limiter protects the hardware, but its interaction with the rest of the controls determines whether the connected system remains stable. NREL, 2025 guide, pp. 3–4.

Taken together, the studies make current magnitude only one part of the explanation. The sequence of voltage, speed and mechanical loading matters too. A result that ends when the unloaded shaft reaches speed leaves the subsequent process-load transition unanswered.

What PowerFac-261 ratings tell a project team

For a commercial or industrial project, the HUA-X PowerFac-261 provides a concrete example of why energy, rated power and the starting duty need separate interpretation. Its published technical brief lists 125 kW rated AC power, 261 kWh nominal battery energy and a 400 V AC connection. The sheet also specifies output derating above 45 °C. PowerFac-261 technical brief.

The practical connection is a defined motor duty, rather than a conversion from motor kilowatts to battery kilowatt-hours. A direct-on-line start, a drive-fed start and a restart with the process unloaded impose different demands. Motor nameplate data, starting method, driven-load torque and inertia, and other loads sharing the supply describe that duty. The applicable inverter overload behavior and voltage response determine whether the proposed configuration can support it.

For a project involving PowerFac-261, those inputs can accompany a project enquiry. The model’s public ratings do not establish grid-forming, black-start or motor-starting performance; the relevant configuration and capabilities remain project-specific. The technical-document request provides a route to the applicable product information.

The related backup, transfer and black-start article explains how a site moves between supply states. Motor starting adds the next layer: after an AC supply exists, the machine must still accelerate and remain stable as its working load arrives.