Gabriele Ribichini

Saturday, April 8, 2023 · 0 min read

Economizer Testing - Do Energy Savings Through Harmonic Filters Really Work?

Electric power or energy economizers claiming to optimize the electrical current received by industrial consumers are presently marketed - and purchased - to reduce consumption and cut down the power bill. 

Claiming to save more than 10% electrical power the manufacturers calculate the return of this fairly significant investment within a few years. Using Dewesoft’s SIRIUS High-Speed DAQ modules and the DewesoftX Power Module with simple mathematics I checked out these claims.


I was intrigued by these devices and the amazing benefits promised. My curiosity spurred me on to analyze the applicable saving estimation methods and to investigate the reality of these savings. With example measurements, I discovered how they can deceive customers

Dewesoft Power Analyzer is the world's smallest high-precision power and power quality analyzer

Some time ago I came across some very interesting technologies to improve the energy efficiency of industrial systems. I like to be green and I like technologies helping to solve inefficiency and prevent energy losses. I was very interested in some of these devices since they claimed to generate a very consistent savings.

At the website of one producer I found this statement:


Energy saving is an infinite source of clean energy that can help any company reduce energy waste and the cost of the electricity bill. Besides new renewable energies, energy saving is one of the best ways to fight climate change and pollution.

Being in the field of precision measurement I was quite fascinated by this statement, especially the term “infinite source”. I decided to investigate more and see how much energy-saving such innovative technology could really generate in an existing industrial plant.

Talking to a couple of the economizer producers I realized that different solutions are certified to achieve very different performances, from almost zero to 10% or above in average saving. My own idea is that basically these systems can save only a portion of the energy wasted.

Wasted energy basically means losses in conduction due to reactive power. In modern industrial plants, nonlinear loads generate a not negligible distortion reactive power that increases the total reactive power (Q) - see figure 1.

Figure 1 - Wasted electrical power: Losses in conduction due to reactive power.

Energy economizers

An energy economizer is basically a filter - active or passive - changing current waveform shapes in order to reduce the total Reactive Power and thus power losses. One manufacturer describes the product this way:

… a passive inductive filter with hybrid functions, given by its capabilities to inject into the power flow some electromagnetic vectors in opposition of phase, utilizing some of the power derived from the incoming energy flow and so causing a voltage drop proportionate to the filtering level selected. The inductance is not constant, but it changes dynamically its filter impedance value, adapting to the power absorption of the electrical network and so maximizing its effectiveness.

It is a purely electrotechnical device. Something like a combination of a variable transformer and filter that acts according to a dedicated algorithm to optimize the Power Quality, and thereby generate power savings.

Electrical power economizer

Power Quality is a combination of voltage profile, frequency profile, harmonics contain, and reliability of power supply. It is the degree to which the power supply approaches the ideal case of stable, uninterrupted, zero distortion, and disturbance-free supply.

The device can be seen from the voltage optimization point of view, even though it is more of a Power Quality optimizer than a simple voltage optimizer since they also act on the current waveforms.

Basically, by optimizing the Power Quality, these products serve to generate power and cost savings. In this scenario, consumers get to benefit from improved power quality and the return on the investment due to power saving.

Starting this line of thought I was doubting very much the more than 10% average saving, this would gain the industrial plant. By saving only a portion of the energy wasted it would mean that the plant, in reality, would have average energy losses much greater than 10%. 

Analyzing it a bit further, it was becoming clear to me that calculating the savings is not at all a trivial matter! The producers I contacted had different approaches to these measurements and estimations. 

Several questions came to my mind, and I decided to perform some measurements in this field in an attempt to clarify my doubts. 

I had become interested in the producer claiming a saving of greater than 10%. As a measurement professional, I had the opportunity to install the Dewesoft power quality analyzer on a specimen of this product to measure the parameters describing the deviation of the voltage from its ideal sinusoidal waveform at a certain frequency. 

The Dewesoft power quality analyzer based on DewesoftX software and SIRIUS HS (High Speed) data acquisition systems can measure all the power quality parameters according to IEC 61000-4-30 Class A.

Dewesoft R1DB power quality analyzer with SIRIUS-HS type amplifiers

Estimation of savings

Yes, the saving cannot be calculated – it can only be estimated

The problem of calculating the saving is linked to the difficulty of keeping the connected load sufficiently constant. To have the facts, the power consumption of a system with and without the economizer at a specific given time has to be known. The difference between the two is the actual saving. However, it is easily understood that it is impossible simultaneously to know the consumption of one and the same system with and without the economizer.

Talking with economizer manufacturers I learned about a couple of different approaches for the estimation of total savings. A very simple one that requires high-performance power meters and a simpler one that unfortunately brings very high measurement uncertainty.

The estimate of saving on instantaneous power

One economizer, I found in the market embeds a high-power tri-phase bypass switch meaning that the device can be bypassed for just less than 20ms.

Assuming that the load does not change in such a short period of time, the measurement system could calculate saving based on the power values measured before and after actuating the bypass switch.

Having the load constant, a variation in power consumption could only be accredited to the economizer effect.

This is a fairly simple approach is based on two strict conditions:

  • The economizer integrates the bypass switch, and

  • The measuring system is able to calculate power within one voltage period. The DewesoftX data acquisition software allows this calculation enabling “periodic values” in the power module.  

Anyhow, most of the economizers do not include such a bypass switch - they cannot be bypassed at all. This kind of economizer is always ON and simulates the bypass function by outputting the same voltages measured on the grid (with some tolerances).

The transition ON-OFF (and vice versa) requires 20-30 seconds making it is impossible to consider the load as constant. In this case, a variation in measured power could be related to the variation of load or/and the economizer effect, and there’s no way to distinguish between the two.

The estimate of saving on long-term energy consumption

The determination of energy savings requires both accurate measurement and replicable methodology, known as a measurement and verification protocol. In this case, the economizer manufacturer utilizes the International Performance Measurement and Verification Protocol (IPMVP). The protocol defines standard terms and suggests best practices for quantifying the results of energy efficiency investments.

IPMVP provides four options for determining savings (A, B, C, and D). For this kind of industrial appliance, the most appropriate method seems to be described by Option C, since: 

Savings are determined by measuring energy use at the whole facility or sub-facility level [...] where the ECM is expected to affect all equipment in a facility.

The manufacturers then base their saving estimations on two energy counters used for 3-day measurements - the period determined from the result of an algorithm calibration procedure.

Basically, the calibration procedure verifies that in normal operation two counters will have the same total Energy value after a period of time - without the economizer. The switching period can be adjusted and a longer test applied. This is guaranteeing that a difference in Eon and Eoff values can be ascribed to just the economizer activity alone and not due to load variation.

During this test, the economizer is cycling between ON and OFF status with predefined intervals - set to 15min, also based on the result of the algorithm calibration procedure. One energy counter is summing all the energy consumption during all periods where the economizer was set to ON, the other when it was set to OFF. The difference between the total energies measured by the two counters is then used to estimate the total saving after the 3-day period.

This is a more complex approach but it can be applied to any economizer and does not necessarily require high-end measurement systems like Dewesoft. Two simple and low-cost energy counters will do the job.

Testing with measurements

Measurement setup

The system I had the chance to analyze showed about 11% average saving - this economizer does not include the bypass switch so the IPMVP protocol was implemented to estimate saving over a long-term energy measurement. 

Even though to measure consumed energy, it is sufficient to install energy counters before the economizer, I decided to install two Dewesoft power analyzers, one before and one after the economizer. I wanted to get the most possible information to see if and how the economizer is eventually cheating the user with saving estimation or to prove that such an economizer is really working!

To achieve this, I used two SIRIUS High-Speed data acquisition devices, the first measuring voltages and currents at the grid, and the latter measuring after the economizer. The two were obviously perfectly synchronized and calculated all power parameters.

Two Dewesoft SIRIUS High-Speed DAQ modules - one measuring grid voltages and currents and one measuring after the economizer.

The economizer was providing two dry contact relays activated respectively when the economizer status was OFF and ON. The dry contacts relay status was read by two isolated LV input channels having an integrated programmable power supply.

Power and energy were calculated by the DewesoftX power module. With simple mathematics, I had been able to totalize the energy for ON and OFF periods separately according to the standard IPMVP protocol

Measurement screen showing measurements and harmonics inside Dewesoft X data acquisition software


These systems were left for three days monitoring all power parameters while the economizer was left in TEST mode cycling between ON and OFF status every 15 min.

DS-FLEX-3000-35-HS current transducers installed at input and output of the economizer

Similar measurements were performed in parallel by the economizer-embedded control system showing results on a screen.

After three days the economizer was showing EOn 11.738,6 kWh and EOff 12.528,2 kWh which sums up an electrical power saving of 6,3%. Both the embedded economizer system and our Dewesoft systems were measuring comparable results.

Display of the embedded power economizer measurement system

This was already a surprise to me. The system was always showing an average saving of more than 10% during the last 3-4 months. This was reduced to 6,3% while I was testing the system using a high-performance power analyzer.


Analyzing the data acquired I found some very interesting results. Having a full-time history, I was able to generate a graph of the energy data required during any ON and OFF periods.

Energy from OFF periods in green, energy from ON periods in pink.

This graph clearly shows that in some periods when the economizer was in ON status the requested energy was measured close to zero.

I then zoomed in on a smaller section of this same graph to analyze the energy graph of the grid input and economizer output voltages and the measured total energy from the grid.

The energy graphs the voltages - phase 1 grid in green and phase 1 economizer output in red. At the bottom the measured total energy from the grid.

It turned out that the voltage output of the economizer is switching regularly while the total grid energy is increasing almost constantly.

There’s no obvious reason why the measured energy in those periods should be close to zero. Except, that is if the economizer relay was misactivated - or misread. This assumption was clearly confirmed by analyzing the switching time.

Switching time data shows multiple irregular activations.

The switching time indicates multiple activations - false activations not following the indicated economizer behaviour. And of course, bringing the ON period energy consumption to ZERO reduces the values of the Eon totalizer and finally increases the saving calculation.

Having all the time-domain signal Dewesoft Power Analyzer was able to re-construct the correct calculation times despite the incorrect relay operations and then the already reduced saving of 6,3% came down to an estimated saving of 1,8%.

I am surprised that this simple trick - or undesired behaviour - was able to deceive the third-party energy meter. But fortunately not the Dewesoft power meters. Thanks to the ability of offline mathematics allows any kind of recalculation starting from recorder signals.

Any Energy Manager with a standard power quality analyzer or energy meters would have measured 6.3%. Only a high-performance data acquisition system with power measurement capabilities like Dewesoft could detect this economizer malfunction – or trick.

Looking at the data some further strange behaviour of the economizer device is noticeable. 

Comparison of the phase 1 voltages from the grid (green) and the emulation from the economizer output (red).

The economizer does not integrate a bypass switch and to emulate the OFF status it should generate the voltage which is received from the grid.

Comparing the phase 1 voltages from the grid and from the economizer output it is evident that during OFF states the voltage is always a bit above that of the grid (+2,5V). In presence of linear loads sensible to voltage variations a voltage, the increase is associated with an increase in power and energy.

In other words, this could be a deliberate mechanism to increase the values of the Eoff totalizer and thus artificially and falsely increase the saving calculation.

Moreover, by analyzing signals' time histories I had the possibility to get data both in test mode and in normal operation. It seems that the system uses a different algorithm in test mode than the normal operation, applying different voltage levels. I suspect that also this behaviour could be designed to artificially increase saving results.


Unfortunately, I was not able to find an infinite source of clean energy in this device. The economizer that was sold to achieve a 10,5% average saving - with 8,5% guaranteed by contract - was far from compliance. The display showed 6,3% but in reality, it only provided a 1,8% saving to be even further reduced discounting multiple voltage tricks found when analyzing the signals.

I clearly realize that standard PQ analyzers or simple energy totalizers used to implement the IPMVP protocol for saving estimation can be easily sidetracked by the economizer behavior and will confirm the incorrect saving of 6.3%.

I also confirmed that consumers must be critical and should question manufacturers offering high savings. It is very hard to prove that they are deceiving without a high-end power analyzer. But keep in mind - only a small portion of losses are saved.