FAQ: Industrial Power Supplies & Control Systems

Conclusion: constant current and constant voltage are both closed-loop control; only the feedback quantity differs. The unit samples output current or voltage, compares it with the setpoint and adjusts the thyristor firing angle. CC mode uses current as feedback and holds output current constant; CV mode uses voltage as feedback and holds output voltage constant. For selection, confirm the load characteristics and process requirements first, then choose the mode, because resistance furnaces, transformers and electrolysis must not be forced into one fixed mode.

1. What CC and CV mean

2. How the closed loop works

1. Sampling: output current or voltage is measured continuously by current transformers, shunts or voltage sensing circuits.
2. Comparison: the sampled value is compared with the setpoint to obtain the deviation.
3. Regulation: the controller adjusts the thyristor firing angle according to the deviation, changing the conduction angle and therefore the output.
4. Switching and limiting: CC and CV can run combined and switch automatically, with voltage and current ceilings set as protection boundaries.

Annex: one worked example — why the voltage ceiling must be stated

In constant-current mode the output voltage is set by the load resistance, following Ohm's law: U = I × R. Take a low-voltage high-current electrolysis duty at 270 V / 1400 A. At full load the equivalent load resistance is:

R = U ÷ I = 270 V ÷ 1400 A ≈ 0.193 Ω

At a constant 1400 A, if the equivalent resistance rises from 0.193 Ω to 0.21 Ω during heating or electrolysis (temperature rise, polarisation or a change in electrode condition can all cause this), the required output voltage becomes:

U = 1400 A × 0.21 Ω = 294 V

294 V exceeds the 270 V voltage ceiling of this rating, so the supply transfers automatically into voltage-limit mode and the actual current falls to about 1286 A (270 V ÷ 0.21 Ω); the process current is no longer under control. This is why an enquiry must state both the voltage ceiling and the current ceiling. A current figure on its own tells you nothing about how much voltage headroom remains later in service.

Note: the resistance change in this example illustrates the principle. Actual variation depends on measured load data and project conditions.

3. How to select the mode

4. Four items to confirm before enquiry

• Load type and the range over which resistance changes with temperature and ageing
• Whether constant voltage, constant current or constant power is required, and whether automatic switching is needed
• Feedback signal source and accuracy requirement, and where the sampling point is located
• Upper and lower limits for voltage and current, and the protection action

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: the core difference is the regulation method. SCR hydrogen power supplies use phase-shift firing and adjust output by changing the conduction angle, so regulation is tied to the mains period. IGBT hydrogen power supplies use high-frequency switching with PWM control and can correct the output within milliseconds. For hydrogen duty the decision rests on three things: single-unit power level, requirements on current ripple and dynamic response, and grid-side harmonic and power factor limits.

1. Why hydrogen production is sensitive to the power supply

An electrolyzer is a low-voltage, high-current load, a DCA 270V/1400A class unit for example, and current ripple affects electrode life and hydrogen efficiency directly. Where PV or wind is part of the supply, the power supply must also track a wide power range quickly. Selection therefore cannot stop at whether the power rating is sufficient; ripple, response speed and grid-side indicators have to be checked at the same time.

Annex: three worked examples that turn the table into numbers

Example 1 — stack power and number of parallel modules. Power follows P = U × I. Taking the 270 V / 1400 A rating:

P = 270 V × 1400 A = 378,000 W ≈ 378 kW

With an IGBT design built from 60 kW modules, the module count is 378 ÷ 60 ≈ 6.3, rounded up to 7 modules. Where N+1 redundancy is required so that the plant can keep running at reduced output after a single module failure, allow 8 modules. This is why a modular design needs expansion space reserved from the outset.

Example 2 — how ripple factor is calculated. Ripple factor = RMS ripple current ÷ DC average current × 100%. With a measured DC average of 1400 A and an RMS ripple of 14 A:

Ripple factor = 14 ÷ 1400 × 100% = 1%

When the technical specification is signed, the test conditions for this figure must be stated (full load or part load, and where it is measured). Without them, readings from the same unit can differ by several times.

Example 3 — why an SCR design needs multi-pulse rectification. The characteristic harmonic orders of a rectifier are h = p·k ± 1 (p is the pulse number, k = 1, 2, 3…):

• 6-pulse (p = 6): h = 5, 7, 11, 13; the 5th and 7th are the largest and are the most expensive to mitigate
• 12-pulse (p = 12): h = 11, 13, 23, 25; the lowest order is raised from the 5th to the 11th, which simplifies the filter considerably

This is the direct reason why SCR hydrogen power supplies are generally supplied with 12-pulse rectification and a smoothing reactor. The PWM rectifier figures in the table — power factor 0.95 and harmonics 7.67% — were measured under specific test conditions; engineering values must be calculated against the PCC short-circuit capacity and the target limits for each project.

2. Three decision lines

1. Above 5MW single-unit, steady load, capital-cost sensitive: prefer the SCR route, where device cost and supply maturity are advantageous at this power level.
2. Ripple-sensitive electrolyzer, tracking wind or solar fluctuation, or restricted space: prefer the IGBT route and use N+1 redundancy to raise continuous-run capability.
3. In the 1 to 5MW middle band: write ripple limits, dynamic response time and grid-connection harmonic limits as verifiable clauses and have suppliers confirm them before deciding the device route.

3. Four items to confirm before enquiry

• Output voltage and current range and the load factor band, for example 270V/1400A, stating whether long low-load operation is expected
• Current ripple limit and test conditions, assessed at both full and low load
• Harmonic and power factor limits at the connection point, and whether mitigation equipment is already installed
• Communication interface and interlock requirements with PLC, DCS or host computer

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: verify a retrofit item by item across four areas, electrical, mechanical, communication and thermal, and never buy on the old model number alone. Check at least seven items: existing cabinet dimensions, busbar rating, cable entry, control signals, communication protocol, cooling space and site interference sources, then define the replacement. Buying on the old model number alone often leads to a mismatch in rated current, cooling or PLC communication, causing rework on site and schedule delay.

1. Why the old model number cannot simply be copied

After years in service the cabinet has been drilled, the busbar has been built to the original rating, and the PLC program and communication point list are fixed. Even if the electrical parameters of the new supply are similar, mechanical dimensions, terminal definitions, signal levels and protocol details may differ. In addition, the original cabinet cooling was designed around the losses of the old equipment; after a change of equipment the loss level may differ, so the air path and opening area need to be re-checked.

Annex: two figures to calculate before a retrofit

Example 1 — can the existing busbar be reused? The current-carrying capacity of a copper busbar can be estimated from an empirical current density, commonly taken as 2 A/mm² (a rule-of-thumb reference only; the final check must account for busbar material, installation method and ambient temperature):

Required cross-section S = I ÷ 2

If the retrofitted circuit must carry 800 A: S = 800 ÷ 2 = 400 mm², equal to a 100 mm × 4 mm copper bar. If the existing cubicle has 80 mm × 4 mm = 320 mm², it can carry roughly 640 A, which is not sufficient; the bar must be remade or paralleled and widened. Skip this step and terminal overheating after energisation is only a matter of time.

Example 2 — is the cubicle cooling adequate? The required airflow follows the heat-balance equation:

Q = P_loss ÷ (ρ × c × ΔT)

where Q is airflow (m³/s), P_loss is total cubicle loss (W), ρ is air density at about 1.2 kg/m³, c is the specific heat of air at about 1005 J/(kg·K), and ΔT is the permitted temperature rise (K). With a total cubicle loss of 3000 W and a permitted rise of 10 K:

Q = 3000 ÷ (1.2 × 1005 × 10) ≈ 0.249 m³/s ≈ 896 m³/h

So about 900 m³/h is required. Fans and vent area in an older cubicle were sized for the losses of the old equipment; when the new equipment changes the loss level, this equation must be re-applied and checked at the maximum ambient temperature, not at rated conditions.

2. Four steps for a retrofit

1. Survey the site first. Photograph cabinet dimensions, busbars, cable entries and terminal definitions and build a retrofit checklist.
2. Then verify electrical parameters. Determine the rated current from the measured current and the impact curve of the existing load, not from the nameplate.
3. Then fix the interface scheme. Check signal types and communication protocol point by point; add isolation or conversion modules where needed.
4. Finally re-check cooling. Calculate air flow from total cabinet heat dissipation, select fans with a 1.3 to 1.5 margin, and check the air path and opening area.

3. Four items to confirm before enquiry

• Measured cabinet and busbar dimensions and ratings, and whether existing parts can be reused
• Measured current, impact curve and duty of the existing load
• PLC or host signal types, communication protocol and point list
• Total cabinet heat dissipation, ambient temperature and the permitted shutdown window

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: protection must be judged on three levels, device level, system level and maintainability, not on panel alarms alone. Check at least seven items: instantaneous overcurrent, semiconductor fuses, phase loss, over-temperature, cooling interlock, fault clearing logic and maintenance access, and confirm them together with the load impact curve, maximum current, cooling method and site short-circuit protection conditions.

1. Why a panel alarm is not the same as complete protection

An alarm is only an indication of the result. What decides whether the equipment survives an abnormal condition is the action logic and device coordination: which device acts first on overcurrent, whether output is cleared promptly after a fuse blows, whether output is blocked on phase loss, and whether the unit trips on cooling failure. These actions have to be coordinated within a very short time; if any link is missing, a recoverable event turns into device destruction or a furnace stoppage.

Annex: fuse I²t coordination and short-circuit magnitude

Example 1 — how to coordinate fuse and device I²t. The criterion is that the melting I²t of the fuse must be lower than the I²t the thyristor can withstand, that is:

K = I²t_fuse ÷ I²t_device < 1

Suppose a given thyristor rating has an allowable I²t of 1.5 × 10^5 A²s:

• With a fuse whose melting I²t is 1.0 × 10^5 A²s: K = 1.0 ÷ 1.5 ≈ 0.67, the fuse operates first and the device is protected
• If a fuse with a melting I²t of 2.0 × 10^5 A²s is wrongly chosen: K = 2.0 ÷ 1.5 ≈ 1.33, the device fails before the fuse operates

So increasing the fuse rating is not safer; it simply moves the failure from the fuse to the far more expensive semiconductor. Use the I²t values given in the device datasheet and the fuse catalogue.

Example 2 — how large is the short-circuit current? Short-circuit current is set by the loop impedance: I_sc ≈ I_N ÷ Z% (Z% is the transformer impedance as a percentage). With a rectifier transformer impedance of 5% and a rated output current of 2000 A:

I_sc ≈ 2000 ÷ 0.05 = 40,000 A

At the instant of a short circuit the current can therefore reach 20 times rated, about 40 kA. This is why protection operating time has to be assessed in milliseconds: however accurate the setting, a slow device cannot interrupt it. In project discussions ask for the setting range and the operating time, rather than accepting that an alarm exists.

2. Three decision lines

1. Set protection from the load impact curve. Settings must be based on measured or clearly defined impact curves, not on default values carried over from another site.
2. Investigate the duty before replacing a repeatedly blown fuse. Repeated fuse operation usually points to firing parameters, load characteristics or cooling; measure and re-tune on site rather than simply fitting a larger fuse.
3. Check cooling against the worst case. Verify at maximum ambient temperature, minimum water flow or single fan failure, not at rated conditions.

3. Four items to confirm before enquiry

• Load impact curve and maximum short-circuit current as the basis for protection settings
• Fuse rating versus device I2t, and how easily fuses can be replaced
• Cooling method, water quality and temperature or air flow, and the interlock action on cooling failure
• Fault logging and reporting, including whether fault codes can be sent to a host computer or DCS

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: for semiconductor diffusion furnaces, zone temperature stability comes first. Control-signal immunity and smoothness of output regulation come next. Diffusion processes are sensitive to temperature fluctuation, so the measure of a power solution is not how wide the regulation range is, but whether temperature fluctuation stays inside the process window during long continuous operation, and whether the unit is immune to interference in the site electromagnetic environment.

1. Why diffusion furnaces are stricter

Diffusion furnaces usually run several tubes and zones in parallel, with long process cycles, and temperature fluctuation directly affects the uniformity and repeatability of the diffused layer. At the same time the site contains temperature controllers, vacuum systems, RF equipment and variable-speed drives, and interference easily couples onto control signal lines, causing output drift or spurious operation. A power solution therefore has to answer three questions at once: is the regulation smooth and stepless, is the feedback accurate and stable, and is the control loop immune to interference.

Annex: turning a stable temperature into two verifiable figures

Example 1 — how temperature stability is calculated. Stability = fluctuation amplitude ÷ setpoint × 100%:

Stability = ΔT ÷ T_set × 100%

With a process setpoint of 800 °C and a measured fluctuation of ±1.2 °C:

Stability = 1.2 ÷ 800 × 100% = 0.15%

When the specification is agreed, state whether the requirement is ±1.2 °C or ±3 °C, where in the furnace tube the sensor sits, and how long the steady-state window is. Without test conditions, the word stable carries no meaning.

Example 2 — power and zone division with several tubes in parallel. Total power follows P_total = P_single × N. With a heating zone of 30 kW per tube and 5 tubes in parallel:

P_total = 30 kW × 5 = 150 kW

If the process is divided into 3 zones, each zone is 50 kW and is served by its own power controller with an independent closed loop. This avoids the situation where one supply feeds several zones and adjusting one zone disturbs the others, and it allows a single zone to be isolated after a fault. The number of zones and the power per zone depend on the actual tube layout and the process profile.

2. Three decision lines

1. Size by tubes and zones. Confirm tube count, zone division and per-zone power before fixing controller quantity and rating.
2. Filter models by control mode. Confirm which of constant voltage, constant current and constant power the model supports, and check current limiting and soft start.
3. Define the interface from site EMC conditions. Identify interference sources and grounding first, then fix the control signal type and the shielding and isolation scheme.

3. Four items to confirm before enquiry

• Tube count, zone division, per-zone power and element rated current
• Supply arrangement, voltage and frequency range, and permitted voltage fluctuation
• Temperature controller interface, control signal specification and communication protocol
• Site EMC conditions, grounding method and available shielding and isolation measures

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: configure power controllers zone by zone, and handle three-phase balance and staggered switching at the same time. Lithium battery material sintering furnaces have many zones and long continuous cycles, so the core question is not whether the total power is sufficient, but whether each zone can be regulated independently, whether simultaneous switching of several zones shocks the grid, and whether temperature uniformity stays stable over long runs.

1. Why zone-by-zone configuration is needed

Pusher kilns and roller hearth kilns used for lithium battery materials are normally divided into several zones, each with its own set temperature and power demand that changes with the process cycle. If several zones share one supply, any adjustment in one zone affects the others and temperature uniformity cannot be guaranteed. With zone-by-zone configuration, each zone is closed-loop controlled by its own power controller, which allows independent adjustment to the process curve as well as fault isolation and zone-level maintenance.

Annex: three-phase unbalance and line current worked example

Example 1 — converting total power to line current. For a balanced three-phase load, I_L = P × 1000 ÷ (√3 × U_L × cos φ). Take 6 zones of 40 kW each, 240 kW total, a line voltage of 380 V and a power factor taken as 1:

I_L = 240,000 ÷ (1.732 × 380 × 1) ≈ 365 A

Each phase carries 80 kW, giving a phase current of about 364 A at a phase voltage of 220 V, which agrees with the line current and confirms a balanced allocation.

Example 2 — how unbalance is calculated. Three-phase unbalance = (maximum phase current − minimum phase current) ÷ average phase current × 100%:

ε = (I_max − I_min) ÷ I_avg × 100%

Using the same 240 kW at a phase voltage of 220 V: if all 6 zones are placed on phases A and B (phase A 120 kW, phase B 120 kW, phase C 0), then I_A = I_B = 120,000 ÷ 220 ≈ 545 A and I_C = 0 A, and the average phase current is (545 + 545 + 0) ÷ 3 ≈ 363 A, giving:

ε = (545 − 0) ÷ 363 × 100% ≈ 150%

An unbalance of 150% significantly worsens both the neutral current and the phase voltage deviation. Distributing 2 zones per phase (80 kW each) instead brings all three phase currents to about 364 A and the unbalance close to 0. This is the origin of the rule that the number of zones is best taken as a multiple of 3.

Example 3 — why staggered switching matters. If 6 controllers conduct simultaneously, the instantaneous inrush is the sum of the 6 starting currents. With intelligent power allocation that staggers the conduction timing, fewer units are switched in at any one moment and the peak inrush falls. The exact stagger strategy and the reduction achieved must be confirmed by measurement against the process cycle.

2. Three decision lines

1. Zones first, then power. Build a distribution table from zone count, per-zone power and supply arrangement, then fix the model and quantity.
2. Coordinate timing when several units switch together. Simultaneous zero-crossing conduction of several controllers produces superimposed inrush current; intelligent power distribution can stagger the output timing on a fixed or variable cycle basis.
3. Check cooling against the continuous cycle. Sintering furnaces often run continuously, so cooling must be sized on total cabinet heat dissipation rather than on short-time duty.

3. Four items to confirm before enquiry

• Zone count, per-zone power, element type and rated data
• Supply arrangement, available phases and constraints on three-phase distribution
• Preferred control method, zero-crossing or phase-shift, and whether staggered switching is required
• Control signal type, communication protocol, host interface, cabinet layout and cooling conditions

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: never specify a DC power supply on rated power alone. Confirm at least eight items: input grid, output voltage and current range, continuous duty power, load fluctuation, ripple requirement, CC/CV mode, cooling conditions and communication interface. The more completely these parameters are defined, the clearer the design boundary, and the smoother installation, commissioning and long-term operation will be.

1. Why a single kilowatt figure is not enough

The same rated power can correspond to completely different output combinations: a low-voltage high-current design and a high-voltage low-current design differ greatly in rectifier topology, device selection, busbar and cooling. In addition, duty cycle, load fluctuation amplitude and ripple limits directly affect device margin and filter configuration. Given only a power figure, a supplier can quote only on common assumptions, leaving the design boundary vague and making later changes expensive.

Annex: the same 500 kW, two very different designs

Current is set by power and voltage together: I = P ÷ U. For the same 500 kW DC output, two voltage and current combinations give completely different results:

The calculation:

• Option A: I = 500,000 W ÷ 500 V = 1000 A
• Option B: I = 500,000 W ÷ 100 V = 5000 A

The power is identical but the current differs by a factor of 5, and the busbar section differs by roughly the same factor; the number of paralleled devices, the cooling method, the rectifier pulse number and the filter arrangement all change with it. So quoting only a single 500 kW figure leaves a supplier to price against its own assumptions.

The ripple limit needs to be quantified too. Ripple factor = RMS ripple current ÷ DC average current × 100%. If the requirement is ripple ≤ 1% at a DC average of 5000 A, the permitted RMS ripple current is:

ΔI ≤ 5000 A × 1% = 50 A

This figure must be quoted with its test conditions (full load or part load, and the measurement point). Ripple is usually worse at part load, so both conditions need to be assessed separately.

2. Where the design boundary differs by application

1. Electroplating and electrolysis: low ripple and long continuous duty are the core; check ripple limits and cooling water quality.
2. Hydrogen production: electrolyzers are sensitive to current ripple; with PV or wind input, tracking capability across a wide power range must also be checked.
3. Submerged arc furnaces and smelting: low voltage, high current and impact resistance are the core; calculate short-time overload and protection from the load impact curve.
4. Welding and plasma: dynamic response and arc starting and stability are the core; check response time and constant current accuracy.

3. Four items to confirm before enquiry

• Upper and lower limits of the output range, and whether long operation away from rated load is expected
• Ripple limit and test conditions, assessed at both full load and low load
• Whether the cooling method is practical: water quality and temperature for water cooling, ambient temperature and cabinet space for air cooling
• Communication interface, interlocks and fault reporting to PLC, DCS or host computer

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: decide the control method from the load characteristics first, then the current rating. Resistive loads such as resistance wire and graphite can be evaluated for phase-shift or zero-crossing control. Elements whose resistance changes significantly with temperature, such as silicon molybdenum (MoSi2) and silicon carbide (SiC), need constant power or current limiting. Transformer and induction loads require the magnetising inrush and firing method to be checked. If the control method is chosen wrongly, no amount of extra current margin will make up for it.

1. Why load characteristics come first

The output capability of a power controller depends not only on its rated current, but also on how the load resistance changes through cold start, heat-up and ageing, and on its inrush behaviour. Excessive cold-start inrush repeatedly blows semiconductor fuses at high firing angles; rising element resistance after ageing requires constant power capability to hold the process temperature. The correct selection order is therefore load characteristics, control method, voltage and current rating, interfaces and protection, not the reverse.

Annex: two calculations for cold-start inrush and ageing allowance

Example 1 — the cold-start factor. At constant voltage, current is inversely proportional to resistance, so the ratio of cold to hot current is:

I_cold ÷ I_hot = R_hot ÷ R_cold

For an element with a cold resistance of 4.0 Ω and a hot resistance of 5.2 Ω:

I_cold ÷ I_hot = 5.2 ÷ 4.0 = 1.3

For a 90 kW / 380 V three-phase resistive load, the hot line current is I_L = 90,000 ÷ (1.732 × 380 × 1) ≈ 136.7 A, so the cold starting current can reach:

136.7 A × 1.3 ≈ 177.7 A

Selection therefore cannot be based on the hot value of 136.7 A alone; it must be checked against at least 177.7 A, or a current-limited start combined with phase-shift and zero-crossing control must be used to bring the starting current down. The cold-to-hot resistance ratio must be taken from measurement on the actual element, since it varies widely between materials.

Example 2 — how much voltage allowance ageing requires. For resistive heating, P = U² ÷ R. If the element resistance rises to 1.2 times its original value after ageing, the voltage needed to maintain the original power becomes:

U' = U × √1.2 ≈ 1.095 × U

That is a voltage allowance of about 9.5%. If the supply was specified with its rated voltage already at the limit for the initial condition, there will be no way to compensate for ageing by raising the voltage, and the symptom will be insufficient power and a temperature that can no longer reach setpoint later in service. This is the quantitative basis for requiring constant-power control or a reserved voltage allowance with loads such as silicon carbide and molybdenum disilicide elements.

2. Three decision lines

1. Filter models by control capability. Confirm which of constant voltage, constant current and constant power the model supports, then check whether current limiting and soft start are available.
2. Size by voltage and current rating. A conventional main circuit can be evaluated at AC110-440V, 45-65Hz; single-phase ratings cover AC10-1500A and three-phase ratings AC25-1500A, subject to the nameplate and the proposal.
3. Configure by interface. Confirm whether the control signal is 4-20mA, 0-5V or 0-10V, and whether communication is RS485 Modbus RTU or needs PROFIBUS-DP, PROFINET or Modbus-TCP expansion.

3. Four items to confirm before enquiry

• Heating element type, rated data and the resistance change between cold and hot state
• Number of zones, per-zone power and total power, and whether several units switch simultaneously
• Supply arrangement, single or three phase, voltage and frequency range, and site short-circuit capacity
• Control signal type, communication protocol, interlock requirements and cabinet cooling conditions

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.

Conclusion: never choose a submerged arc furnace (SAF) power supply supplier on price alone. Judge on four verifiable criteria: engineering experience with high-impact loads, reference projects on the same ore type and capacity, completeness of the thyristor current-sharing and protection design, and executable commitments for on-site service and spare parts. If any one of the four can only be answered verbally, the risk of an unplanned furnace stoppage later is usually higher than the price difference.

1. Why a submerged arc furnace demands more from a supplier

A submerged arc furnace is a low-voltage, high-current continuous smelting load. Electrode movement, burden collapse and changing furnace conditions produce frequent load impacts; the secondary network impedance is low and the DC-side current can reach tens of kiloamps. When the power supply trips in this environment, the consequence is not limited to the equipment itself, but extends to the continuity of the whole heat and to the energy consumption figures. The right question when evaluating a supplier is therefore not whether the rated power can be met, but whether the unit stays stable under impact conditions and how quickly it can be brought back into service.

Annex: how to size tens of kiloamps and secondary circuit losses

Example 1 — estimating the current magnitude. For three-phase apparent power, I = S ÷ (√3 × U). Taking an 8500 kVA submerged arc furnace with a secondary line voltage of 180 V:

I = 8,500,000 ÷ (1.732 × 180) ≈ 27,270 A ≈ 27.3 kA

If the measured inrush peak reaches 1.8 times rated, the peak current is about 49 kA. Device selection, busbar section and protection settings must all be checked against this magnitude rather than the steady-state rated current — which is exactly why selecting on steady-state current is the first pitfall. The actual voltage level and inrush factor must be taken from measurement on the project.

Example 2 — why secondary circuit losses have to be calculated. Three-phase secondary circuit loss is P_loss = 3 × I² × R. With I = 27,000 A and a secondary circuit resistance of 0.5 mΩ per phase:

P_loss = 3 × 27,000² × 0.0005 ≈ 1,093,500 W ≈ 1.09 MW

So about 1.09 MW is lost in the secondary circuit alone. Reducing the secondary circuit resistance by 0.1 mΩ cuts the loss by about 219 kW:

ΔP = 3 × 27,000² × 0.0001 ≈ 218,700 W ≈ 0.22 MW

This explains why a submerged arc furnace is so sensitive to the layout and impedance of its secondary circuit, and why a supplier should be assessed on whether it has calculated those parameters, not merely on whether it can deliver the rated power.

Note: the voltage and resistance values above are used to illustrate the magnitude and the method; they are not parameters of any specific project.

2. Three decision lines

1. Define the duty before the configuration. Write the furnace type, electrode count, process cycle and load factor range into one enquiry sheet and have every supplier quote against the same sheet.
2. Check the grid side by pulse number on large projects. A 12-pulse arrangement is a common starting point; where capacity is larger or grid connection rules are stricter, evaluate equivalent 24-pulse and above together with reactive compensation and harmonic mitigation.
3. Treat verifiability as a knockout criterion. If references, protection design or service commitments cannot be put in writing, the bid should not enter price comparison.

3. Four items to confirm before enquiry

• Rated and maximum voltage, current and power, and whether the furnace runs for long periods away from rated load
• Load impact curve and short-time overload ratio, as the basis for device rating and protection settings
• Rectifier pulse number, rectifier transformer data, and harmonic and power factor limits at the point of connection
• Fault response time, wear-part list and spare lead time, plus the scope of commissioning and training

Timeliness: compiled from Sichuan Kulun Electric product specifications as of September 2026; final figures depend on the project technical proposal and site conditions.