Active Harmonic Filter Reduces Flicker from Radiator Production — Belgium
| Facility | Radiator factory — 55,000 m², ベルギー. Six production lines, ~5,000 radiators/day |
| Disturbing loads | Presses, seam welding machines, spot welding machines — intermittent high-power loads |
| Flicker before | Pセント peaks reaching 1.6 — measured 2009 |
| Utility limit demanded | Pセント 95th percentile ≤ 0.7 — EN 50160 / IEC 61000-3-7 framework |
| ソリューション | Six Active Harmonic Filter (AHF) units — 2.1 MVAr total continuous reactive compensation |
| Flicker after | Pセント consistently below 0.63 — independently verified |
| Reduction achieved | Pセント reduced by more than 60% — from 1.6 to below 0.63 |
| Side effect | Stabilised production environment — voltage fluctuations reduced across all six lines simultaneously |
01 Context — Flicker from Industrial Welding
Flicker — the perceptible variation in light output caused by rapid voltage fluctuations — is one of the most neighbour-sensitive power quality problems in industrial environments. Unlike harmonics, which affect equipment directly, flicker is primarily a human perception problem: the voltage fluctuations caused by an industrial process can cause visible light modulation in the homes and offices of other customers connected to the same distribution network, even when those customers’ own equipment is entirely non-disturbing.
Welding processes are among the most prolific flicker sources in industry. Resistance spot welders and seam welders draw large, repetitive reactive current pulses — each weld pulse draws thousands of amperes for a fraction of a second, creating a voltage dip at the point of common coupling that modulates the supply voltage at a rate determined by the welding repetition rate. When the repetition rate falls in the range of 1–15 Hz — the frequency range of peak human visual sensitivity as characterised by the IEC flickermeter — the resulting light modulation can be perceptible to all customers on the same distribution transformer.
A radiator factory running six welding production lines simultaneously is not just a noise or emission problem for its immediate neighbours — it is a grid-connected disturbance source that affects every customer connected to the same MV/LV transformer. When the local community grows and new customers connect to the same transformer, the flicker margin shrinks — what was previously acceptable becomes non-compliant when the background flicker from other sources increases. This is exactly what happened here: community expansion forced the utility to tighten the flicker emission limit, making previously tolerated emissions unacceptable.
02 Problem — Pセント 1.6 Against a Limit of 0.7
The radiator factory in Belgium — a 55,000 m² facility producing approximately 5,000 radiators per day across six production lines — had a load mix that was inherently demanding from a power quality perspective. Presses, seam welding machines, and spot welding machines operated simultaneously across all six lines, each drawing large intermittent reactive current pulses that produced significant voltage drops at the feeding substation.
Field measurements in 2009 showed Pセント (short-term flicker severity) values with peaks reaching 1.6. The EN 50160 planning limit for flicker at the medium-voltage point of common coupling is typically Pセント ≤ 0.7 assessed as a 95th-percentile value over a one-week observation period. The factory was exceeding this limit by a factor of more than 2 at peak conditions — causing visible light flicker in neighbouring commercial and residential premises whenever multiple welding lines were operating simultaneously.
The challenge cited in this case — “rapidly fluctuating load and many different load patterns” — is the fundamental difficulty with welding flicker mitigation. A single welding machine produces a predictable, repetitive flicker signature. Six welding lines operating simultaneously produce a complex, stochastic combination of overlapping current pulses at different repetition rates and phases — the resulting voltage fluctuation at the substation is neither periodic nor predictable from the individual load characteristics alone. A compensation system that works for one operating scenario may be inadequate for another. This is why the AHF response time was specifically cited as a critical requirement: the system must track the actual voltage fluctuation in real time, not a predicted or averaged load profile.
03 Solution — Active Harmonic Filtering at 2.1 MVAR
Why an Active Harmonic Filter — not an SVC or passive filter
The solution chosen was six Active Harmonic Filter (AHF) units providing a total of 2.1 MVAr continuous reactive compensation. The AHF approach was selected over the alternatives — passive LC filters, thyristor-controlled SVCs, or standard power factor correction capacitors — for a specific reason: response time.
- Passive LC filters — fixed reactive compensation, tuned to specific harmonic frequencies. Cannot respond to the stochastic, multi-pattern load fluctuations of six simultaneous welding lines
- Thyristor-controlled SVC — updates its firing angle at each half-cycle (8.3 ms at 60 ヘルツ, 10 ms at 50 ヘルツ). For welding loads with pulse durations as short as a few cycles, the SVC response delay means the compensation arrives after the disturbance has already occurred — as described in the IPQDF PQ Overview article on flicker mitigation
- アクティブ高調波フィルター (AHF) — uses IGBTs switching at high frequency to inject precisely controlled reactive current on a cycle-by-cycle basis. Response time is sub-millisecond — fast enough to track the actual welding current waveform and cancel its reactive component before it can produce a measurable voltage drop at the substation bus
An Active Harmonic Filter continuously measures the current drawn by the non-linear load. A digital signal processor calculates, in real time, the reactive and harmonic current components that the load is drawing. The AHF then injects equal and opposite reactive and harmonic currents into the network — effectively making the welding machines appear as resistive loads to the supply network. The voltage at the connection point stabilises because the large reactive current pulses are now circulating within the AHF rather than being drawn from the network impedance. The result: the voltage drops that were causing flicker are eliminated at source, regardless of which combination of welding lines is operating simultaneously.
System configuration
The installation consisted of six AHF units — one per production line — each sized for the specific reactive demand of that line. The total installed compensation capacity of 2.1 MVAr continuous reflects the aggregate reactive demand of six simultaneous welding lines at full production. The system operates with fully automatic controls and passive cooling, requiring no regular maintenance and no operator intervention. It can operate completely stand-alone or integrated with the plant’s existing SCADA and monitoring systems.
04 Results — Pセント Below 0.63 in All Operating Configurations
After installing the AHF system, the plant consistently achieved Pセント values below 0.63 — regardless of how many welding lines were running simultaneously and regardless of the production mix on each line. This is the critical test: the utility’s demand was that the Pセント 95th-percentile value not exceed 0.7, and the AHF must achieve this across the full range of operating scenarios, not just under the single worst-case or best-case loading condition.
The post-installation measurements were conducted by external consultants and approved by the local utility — not measured and reported by the AHF manufacturer alone. This is an important credibility distinction: independently verified flicker measurements provide assurance that the Pセント reduction is real, reproducible, and not an artefact of measurement conditions or cherry-picked operating scenarios. The utility accepted these measurements as proof of compliance with the emission limit it had demanded.
The production stability side-effect
Beyond the compliance achievement, the plant gained an unexpected operational benefit: stabilised production voltage across all six lines simultaneously. When welding machines draw large reactive current pulses, the resulting voltage drops not only cause flicker on the external network — they also cause internal voltage variations that can affect the consistency of the welding process itself. By eliminating the reactive current pulses at source, the AHF simultaneously eliminated the internal voltage variations, improving the consistency of the weld quality and reducing the variation in energy delivered per weld cycle. This operational benefit — improved process quality — was a direct consequence of the PQ mitigation, not an intended design objective.
05 Power Quality Perspective
This case study illustrates the community dimension of industrial power quality — a dimension that is easy to overlook when PQ is framed solely as an equipment protection problem. The radiator factory’s welding machines were not malfunctioning. The factory was not experiencing internal production problems from its own flicker. The problem was entirely outward-facing: the voltage fluctuations on the shared distribution network were affecting neighbouring customers who had no connection to the factory’s production process.
From a utility distribution engineering perspective, this is one of the most common and most difficult flicker management scenarios: an existing industrial customer whose loads were acceptable when they connected, but whose flicker emissions exceed planning limits as the community grows and new customers share the same distribution infrastructure. The utility’s options in this scenario are limited — they cannot refuse supply to new customers, they cannot easily reinforce the network to eliminate the coupling between existing customers, and they cannot compel the industrial customer to reduce production. The only viable path is requiring the industrial customer to mitigate their own emissions — which is what happened here.
The specific mention of AHF response time as the key selection criterion aligns exactly with the utility-side perspective on flicker mitigation technology. A thyristor-controlled SVC — the traditional utility-grade flicker mitigation technology for arc furnaces and large welders — updates its reactive output every half-cycle. For a spot welder whose pulse duration is 3–5 cycles, the SVC is compensating the previous pulse while the next one has already started. The AHF, with sub-millisecond response, compensates the actual current in real time. The trade-off is cost and complexity: A 2.1 MVAr AHF installation is significantly more expensive than a comparable SVC, and the IGBT-based power electronics require a more controlled environment than the thyristor valves of an SVC. For a factory with multiple small welding machines producing stochastic, overlapping load patterns, the AHF’s real-time tracking capability justifies the premium. For a single large arc furnace with a more predictable load cycle, an SVC or STATCOM may be the more economical choice.
参照
- アクティブ高調波フィルタ. AHF Reduces Flicker from Radiator Production — Belgium Case Study. Active Harmonic Filters manufacturer publication. Available at IPQDF Case Study Library.
- IEC 61000-4-15:2010+AMD1:2012. Electromagnetic compatibility — Part 4-15: Testing and measurement techniques — Flickermeter — Functional and design specifications. IEC, ジュネーブ.
- IEC 61000-3-7:2008. Electromagnetic compatibility — Part 3-7: Limits — Assessment of emission limits for the connection of fluctuating installations to MV, HVおよびEHV電源システム. IEC, ジュネーブ.
- IN 50160:2010+A3:2019. Voltage characteristics of electricity supplied by public electricity networks. CENELEC, ブリュッセル.
This case study is based on a manufacturer case study published by アクティブ高調波フィルタ: AHFは、ラジエーターの生産からちらつきを軽減し. Pセント measurements cited (1.6 before, 以下 0.63 after) were independently verified by external consultants and approved by the local utility.
This case study is presented in summary and commentary form for educational purposes. The PQ Perspective section (セクション 5) and SVG diagram are original IPQDF editorial content by Denis Ruest, M.Sc. (Applied), P.Eng. (ret.). IPQDF does not claim authorship of the original case material.