How to Size an SVG: kVAr, Response Time and Frame Selection

Summary. Learning how to size an svg means answering three questions separately: how much reactive power the site needs, how fast the requirement changes, and how much of that duty the existing equipment can already cover. Sizing on a single power factor figure is the reason many installations end up either oversized and idle or undersized at the moments that matter. This guide sets out the calculations, the difference between steady and dynamic duty, how to read an SVG datasheet without being misled by peak figures, and how to decide between a wall unit, a rack module and a floor cabinet.

What Determines the Required SVG Rating?

An SVG rating is expressed in kVAr, but the underlying quantity is reactive current at a given voltage. The rating must cover the largest reactive requirement the site will present, whether that comes from a steady load or from a transient step, and must do so while remaining within the converter's thermal limits at the site's ambient temperature.

The first calculation is steady-state. Measure real power, apparent power and power factor at maximum demand, then compute the reactive power required to reach the target power factor. This is the same calculation used for capacitor sizing, and it defines the minimum continuous rating.

The second calculation is dynamic. Identify the largest reactive step — a chiller start, a charger connection, a welding burst — and compute the reactive current associated with that step. A converter must have enough headroom to cover the step in addition to the steady requirement, or it will saturate precisely when it is needed most.

The third factor is supply strength. The same reactive step produces different voltage changes depending on transformer impedance and fault level. On a weak supply, a given kVAr produces a larger voltage correction, so the device may be sized by the voltage requirement rather than the reactive power balance. That is why two sites with identical bills can need different ratings.

The fourth factor is existing compensation. If a capacitor bank is installed and operating, part of the reactive demand is already covered, and the converter only needs to handle the dynamic component and any shortfall. Ignoring the bank leads to a device that idles or over-compensates.

The fifth factor is environment. Derating above the reference ambient applies to converters as well as to harmonic filters. A cabinet in a hot plant room may need a larger frame than the same duty in a conditioned electrical room.

The power factor target is a commercial decision. Many utilities penalise low power factor below a threshold, while over-correction can also attract charges in some markets. Targeting slightly above the penalty threshold rather than unity avoids over-compensation at light load and reduces the rating required. Confirm the tariff band before choosing the target, because the target drives the size directly.

Why Sizing Errors Are Expensive in Both Directions

1. Undersizing appears only under peak conditions. A device sized on average demand saturates during the events that matter: charger steps, compressor starts and furnace melts. The site then measures voltage variation and concludes the technology does not work, when in fact the rating was simply too small.

2. Oversizing wastes capital and runs the device at low output. A converter sized for a peak that occurs twice a year spends most of its life at a fraction of rating. The capital is committed, the losses are paid, and the customer sees no additional benefit.

3. Sizing on kVAr alone ignores response. A device with enough kVAr but slow response cannot suppress flicker. Response time, not capacity, is often the deciding specification on welding, hoisting and charging loads.

4. Inconsistent voltage assumptions distort comparisons. Reactive current at 400 V and at 480 V for the same kVAr differs by 20 percent. Comparing quotations that assume different voltages without converting to current produces a meaningless ranking.

5. Existing equipment changes the answer. Where a bank is present, adding a converter sized for the full requirement doubles the installed reactive capacity and creates over-compensation at light load — a condition that raises voltage and can trip inverters.

6. Derating is rarely in the headline figure. A device quoted at 100 kVAr may deliver less at 45 °C ambient or at high altitude. Ask for the derating curve and apply it before comparing with a competitor's figure.

7. Staging is cheaper than one large unit in growing sites. Where load will grow, a modular approach lets capacity follow demand. A 150 kvar modular cabinet can be extended internally or paralleled, which is usually more economical than buying peak capacity on day one.

Reading a datasheet correctly. Check whether the quoted kVAr applies at nominal voltage, at which ambient, and over what duration. Some figures describe short-term overload capability rather than continuous rating. Convert everything to continuous reactive current at the site voltage before comparing, and ask for the derating curve in writing rather than inferring it from the headline number.

How to Size an SVG Step by Step

Step 1 — Measure at maximum demand. Record P, S, PF and voltage at the point where compensation will be installed, not at the utility meter. The difference matters because the local board may carry only part of the reactive demand.

Step 2 — Calculate the steady-state requirement. Reactive power needed equals P * (tan φ1 − tan φ2), where φ1 is the present angle and φ2 the target. Convert to current at the site voltage to compare with datasheet ratings.

Step 3 — Add the dynamic component. Take the largest reactive step, convert to current, and add it to the steady requirement. If the sum exceeds the next standard frame, consider whether the step can be managed by load sequencing instead of capacity.

Step 4 — Check voltage sensitivity. Estimate the voltage change the reactive current produces at the board using the source impedance. If the site has a flicker or voltage variation limit, size so that the compensated change stays inside the limit rather than merely matching the reactive balance.

Step 5 — Select the frame and format. Wall units suit small commercial boards; a rack-mount SVG from 10 to 150 kvar suits substations and phased projects; cabinets suit main boards and larger duties. Confirm clearances and cooling before fixing the layout.

Step 6 — Verify the response specification. Where the load changes quickly, choose a device whose response is fast enough to act within the assessment interval — a rapid-response low-voltage SVG rather than a general-purpose unit. Response and rating together define performance.

Step 7 — Commission with a step test and record the result. Apply the largest expected step and capture voltage and reactive current before and after. Compare with the calculation; a significant discrepancy usually indicates an assumption about source impedance, existing compensation or measurement point.

Two final checks often missing from sizing sheets. First, confirm whether the target applies at the incomer or at a local board, because compensating at the wrong point leaves the measured value unchanged. Second, consider a high-efficiency industrial SVG where the device will run continuously, since standing losses over years can exceed the difference in purchase price. The full SVG range lists frames across both criteria.

Sizing for combined duties. Where the device must also provide harmonic compensation or load balancing, the reactive rating available at the same time is reduced. Request the combined duty curve rather than separate maxima. A device that delivers its full reactive rating only when harmonic output is zero is not the same product as one that delivers both simultaneously, even though both datasheets quote the same kVAr.

Documenting the calculation. Record the measurement point, load condition, target, calculation and selected frame. That record is what allows the next site to be specified quickly and what defends the design if performance is questioned. Sizing without a written calculation is a guess with a decimal point.

Sanity checks before ordering. Confirm the calculated reactive current is plausible against the site's transformer rating — reactive current above roughly half the transformer current at maximum demand usually signals a measurement error, not an unusually poor power factor. Cross-check with the utility bill for the same period, and reconcile any large difference before selecting a frame.

Revisiting the calculation after installation. Measure again at maximum demand once the device is running. If the achieved power factor differs from target by more than about two points, the cause is usually an incorrect measurement point, an unreported capacitor bank or a voltage assumption that did not hold. Correcting the calculation preserves credibility for the next project.

When a capacitor bank is still the right answer. If the reactive requirement is steady, the distortion is low and space exists, a detuned bank supplies the same steady-state benefit at lower capital cost. Sizing work should conclude honestly in that direction when the data supports it, reserving converter-based solutions for dynamic, weak-grid or harmonic-constrained sites.

FAQ

How do I convert kVAr to amperes?

Divide reactive power by the phase-to-phase voltage and by the square root of three. A 100 kVAr device at 400 V carries about 144 A of reactive current; at 480 V the same rating carries about 120 A. Always compare devices in current at the site voltage.

Should I size for peak or average reactive demand?

Size for the steady requirement at maximum demand plus the largest dynamic step. Average demand describes the bill, not the performance requirement, and a device sized on average will saturate during the events that cause complaints.

What derating should I apply?

Use the manufacturer's curve for ambient temperature and altitude. As a guide, expect a reduction above 40 °C and a further reduction at altitude. Apply the figure before comparing with another supplier's headline rating.

Can I install a device that is too large?

It will work, but it costs more, runs at low output and may over-compensate at light load, raising voltage. Over-compensation is a real risk where the load profile has long light periods.

How does an existing capacitor bank change sizing?

It reduces the steady requirement. Measure with the bank in service, and size the converter for the dynamic component plus any shortfall. Confirm the bank is detuned correctly before relying on it.

Does response time affect the rating I need?

Yes. A slower device needs more headroom to avoid saturation during fast events, or it simply misses them. Where flicker or fast steps are the reason for the purchase, response time is as important as kVAr.

Conclusion

Sizing an SVG is a three-part calculation: steady reactive demand, the largest dynamic step, and what existing equipment already covers — adjusted for source strength and ambient derating. Convert everything to current at the site voltage, choose a frame with the right response class, and verify with a step test. Done this way, the device is neither idle nor saturated, and the performance claim survives contact with the site.

Need a sizing calculation? Send us your metering data, transformer details and the largest load steps, and we will return a frame recommendation with the derating applied and a commissioning test plan. Request an SVG sizing sheet.