Introduction: Modern large-scale generator sets employ complex computer monitoring and safety protection systems, requiring a stable and reliable 220V AC power supply that is uninterrupted for every second. Uninterruptible power supplies (UPS) have emerged to meet the stringent requirements of these demanding loads, eliminating the impact of transient voltage fluctuations on critical loads. Keywords: UPS power supply. Modern large-scale generator sets employ complex computer monitoring and safety protection systems, requiring a stable and reliable 220V AC power supply that is uninterrupted for every second. Uninterruptible power supplies (UPS) have emerged to meet the stringent requirements of these demanding loads, eliminating the impact of transient voltage fluctuations on critical loads. UPS power supplies are crucial for the safe and stable operation of the generator sets; a UPS outage means a shutdown of the generators and boilers. How to improve the reliability of UPS is a question that field technicians frequently consider. This article analyzes and studies some problems encountered over the years in the planning, configuration, selection, installation, commissioning, operation, and maintenance of peripheral equipment for the UPS system of the Huangpu Power Plant's Units 5 and 6 (300MW units), and exchanges ideas with colleagues. 1 System Overview The UPS system is a low-voltage, multi-terminal network with multiple power inputs. The core equipment of the network, such as inverters and static switches, is a set of electronically controlled power devices. The UPS system of the 300MW unit at Huangpu Power Plant is shown in Figure 1. Figure 1: The original design specifications of the main equipment of the UPS system for the 300MW unit at Huangpu Power Plant are as follows: a) Rectifier input parameters: AC 380V, three-phase, 50Hz, 126A; Output parameters: DC 280V. b) Inverter input parameters: DC 210～280V, 245A; Output parameters: AC 220V, 50Hz, 227A, 50kVA; c) Charger input parameters: AC 380V, three-phase, 50 Hz; Output parameters: DC 160～310 V, 250 A, 65 kW. 2 System Power Supply Planning and Configuration 2.1 Power Supply Configuration Analysis The orientation of the AC power supply is an important part of the UPS system planning and design. To explore the best solution, it is advisable to further examine the principles. The battery charger and the rectifier before the inverter are both three-phase half-controlled bridges, and their functions are similar. The rectifier handles the UPS's continuous load. The charger float-charges the battery and runs in parallel with the battery as a backup for the rectifier. The bypass power supply is a backup for the inverter. The APS connects to some thermal control secondary loads and also serves as a backup for the UPS. Based on this, the general principles for UPS system power configuration can be derived: a) The rectifier and charger power supplies should be connected to different buses; b) The bypass and rectifier power supplies should be connected to different buses; c) The APS and bypass power supplies should also be connected to different buses. The five AC power supplies in Figure 1 are only taken from three bus sections, three of which come from the safety b section. When the UPS unit fails, the inverter is under maintenance, or an accident occurs in the plant power system, and the UPS unit's static switch has been switched to bypass operation, if the safety b section loses voltage, both the UPS bus and the APS2 bus will lose power, and the WDPF and BMS control systems will be paralyzed. Analyzing the power supply wiring of the 300 MW unit at Huangpu Power Plant, both low-voltage transformers No. 2 and No. 0 are connected to the 6 kV B section. When the 6 kV A section loses voltage, transformer No. 0 can automatically switch to supply power to the working section a and the safety section a connected to transformer No. 1. The 380 V working section a and safety section a are superior to the working section b and safety section b. Therefore, with minimal modifications, the power supply configuration in Figure 1 can be further adjusted: a) The charger is reconnected to the working section b; b) APS2 is reconnected to the safety section a. The power plant unit has working and standby high-voltage transformers and multiple low-voltage transformers, as well as diesel generators or safety standby transformers. It is feasible to connect the multiple power sources of the UPS and APS to different transformers and different low-voltage busbars of the unit in a more rational and non-duplicative manner. After a unit is disconnected due to an accident, different power supply systems may have different frequencies. To prevent static switches from failing to switch due to asynchrony, or UPS and APS switching from failing due to asynchrony, these power supplies should be connected to the low-voltage system of the same network as this unit, and not to the public system or other unit systems. 2.2 The single-phase AC voltage output by the inverter of the bypass and APS power supply should be synchronized with the single-phase AC voltage of the bypass power supply to enable parallel conversion. Regardless of which phase of AC the bypass is taken from, the inverter can adjust the output voltage to be in phase and frequency with the bypass voltage. The DPU cabinet and electronic equipment such as computing, storage, and recording stations of the unit's WDPF control system are powered by dual power supplies, as shown in Figure 2. This is a three-terminal network of two inputs and one output formed by reverse parallel connection of thyristors, with UPS giving priority to power supply. When the UPS bus loses voltage or is undervoltage to a certain value, the control circuit triggers the bidirectional thyristors on the APS side, causing them to conduct alternately and turning off the thyristors on the UPS side, allowing the APS to continue supplying power. The switching at this time is parallel first and then disconnected, and the automatic reconnection after the UPS voltage returns to normal is also parallel first and then disconnected. Obviously, if the phase and frequency of the voltage are inconsistent between the UPS bypass side and the APS side, a short circuit will occur at the moment of switching or a large inrush current will be generated due to the large differential voltage, resulting in power outages and component damage within the station. Figure 2 Dual power supply In 1991, during the switching test of Unit 6 of Huangpu Power Plant in the later stage of UPS installation, many switches tripped, and the phase mismatch and switching short circuit problem was discovered for the first time. Other power plants have also had similar cases of incorrect phase connection. It must be pointed out that if the phase is incorrectly connected during installation, it will not be noticed unless there is a switching or reconnection under abnormal UPS voltage conditions. Even if many switches trip, if power is restored from the UPS side first (which is generally the case) and then from the APS side, there will be no short circuit, which still masks this great hidden danger. We hope that the design and construction departments will pay attention to this issue, mark the phase on the drawings, and connect the wires correctly to prevent similar phenomena from recurring. 2.3 The two power supplies of APS, APS1 and APS2, come from different transformers and are not allowed to be connected in parallel for a long time. As shown in Figure 1, Q1 and Q2 should generally be closed, and Q3 should be disconnected. Changing the manual operation of Q1 and Q2 to electric operation, or adding an AC contactor for interlocking automatic transfer, can further reduce the probability of power failure in the DPU and computer station. 2.4 Redundancy Configuration of UPS Units In some power plant projects, manufacturers have added an AC voltage regulator to the bypass power supply. In some technical upgrade projects, an additional UPS is added to the APS side, with one unit using two UPS units. It seems that the UPS power supply configuration scheme is questionable. The author believes: a) The purpose of using a UPS is not that the voltage quality of the power plant cannot meet the requirements of the computer system, but mainly to ensure uninterrupted power supply. One UPS unit already has redundant batteries and chargers; the bypass power supply only temporarily functions in case of inverter output failure. Sometimes, computer systems experience "station drops" and chip component damage, not due to bypass voltage fluctuations. The host hardware of each computer station has self-protection functions. The bypass power supply system should be simplified, and can be simplified. b) The inverter is the "bottleneck" of the UPS system. Based on years of operational experience, the failure rate of the inverter's control section is relatively high. For projects with sufficient resources, consider using one UPS unit with two independent inverters. A dual-inverter configuration is more reasonable, practical, and simpler than a bypass regulator or two UPS units. c) Only thermally critical loads need to be connected to the UPS bus, while the most critical loads (DPU and computer station) are automatically powered by the UPS and APS dual-side power supplies. Adding another UPS unit to the APS side is not advisable. If dual UPSs are necessary, their outputs should both be connected to the UPS bus. d) The configuration of the UPS unit should be scientifically sound and reasonable. Improving UPS reliability should not rely too heavily on adding backup equipment, but rather on maintenance and management. Excessive redundancy complicates the system, increases investment, but results in low utilization efficiency and potentially higher failure rates. 3 System Capacity Estimation and Selection 3.1 UPS Output Capacity To determine the appropriate UPS capacity, detailed load statistics and data collection from similar operating units are necessary. Load statistics require collecting data such as load simultaneity rate, power factor, recurring current, and maximum possible inrush current. The UPS inverter has overload protection; when the output current exceeds (1.2~1.25)In, it will automatically switch to bypass power. To avoid multiple loads starting simultaneously and accumulating inrush current, resulting in frequent switching and back-off, and to prevent overheating of main circuit components, it is necessary to have sufficient UPS capacity margin, but this margin should be moderate. The output capacity of the UPS for the 300 MW unit at Huangpu Power Plant was initially planned as 30 kVA, 136 A. After negotiations with Westinghouse Electric Machinery, it was changed to 50 kVA, 227 A. After commissioning, the actual load on the AC side is generally 90~100 A (approximately 125 A on the DC side), occasionally reaching 110~120 A. If 30 kVA, 136 A is selected, the load factor is 110 A/136 A = 0.73, which is slightly insufficient. If a 40 kVA, 182 A configuration is selected, the load factor is 100 A/182 A = 0.55, which is lower than the recommended value of 0.6. The margin factor is 182 A/100 A = 1.82, which is higher than the recommended value of 1.6, making this a more appropriate configuration. 3.2 The power relationship of the UPS equipment with the rectifier (charger) output capacity is shown in Figure 3. Figure 3 shows the power relationship of the UPS equipment. The input power of the inverter of the 300 MW unit of Huangpu Power Plant is P2=UDC×IDC=280 V×245 A=68.6 kW. The apparent power of the rectifier input is calculated using the formula recommended by SCI: S1=Uex×Iex×Ce／（kPF×η1）. Where S1——rectifier input apparent power, VA; Uex——rectifier output voltage, V; Iex——rectifier output current, A; Ce——rectifier overload factor (1.2～1.5); kPF——crest factor (generally 0.8); η1——rectifier efficiency (0.92～0.94). We can obtain: S1 = 210 V × 245 A × 1.2 × (0.8 × 0.93) = 82.9 kVA. The rectifier input power factor λ1 = P2 / (S1 × η1) = 68.6 / (82.9 × 0.93) = 0.889, and the UPS efficiency η = S3 / S1 = 50 / 82.9 = 0.603. Assuming the UPS load power factor λ = 0.7, then the inverter output active power P3 = S3 × λ = 50 kVA × 0.7 = 35 kW, and the inverter efficiency η2 = P3 / P2 = 35 / 68.6 = 0.51. It is evident that the overall UPS efficiency is relatively low due to losses in various stages. If the UPS capacity is too large and the actual load is too low, the utilization rate will be low, and the efficiency will be even lower. With the charger's output power selected at 65 kW, the rectifier input capacity is calculated to be 87.3 kVA using the formula above. The original design of the rectifier in the domestic charger had an input capacity of 140 kVA, which was too large. Given that the conversion of AC/DC parameters such as P, U, and I in each stage of rectification and inversion is related to the rectification and inversion method, wiring, and load, accurate calculation is quite complex. We hope that the design specifications will provide a set of basic formulas applicable to UPS systems as a basis for capacity selection and estimation. 4. Selection of the Number of UPS Dedicated Batteries 4.1 Reflections on Understanding In 1988, the Huangpu Power Plant's 300 MW unit began installation. As the client's representative, I was new to static inverter UPS systems and had little understanding of their principles and characteristics, leading to a detour in selecting the number of UPS batteries. The original design used 180 QFD-250 batteries manufactured by Yuasa Electric Co., Ltd. of Japan, each with a 250 Ah alkaline charge. The float charge voltage was 1.35 V × 180 = 243 V, and the equalization charge voltage was 1.47 V × 180 = 264.6 V. However, applying this to a typical DC system, it was deemed that the DC voltage was too high, which would shorten the lifespan of components such as relays and indicator lights. Therefore, it was suggested that the number of batteries be reduced. With the consent of the design representative and the contractor's representative, the number of batteries was changed to 171, with a float charge voltage of 231 V and an equalization charge voltage of 251 V. On September 26, 1991, two years after Unit 5 was put into operation, a UPS test was conducted. Less than 10 minutes after disconnecting the rectifier and charger, the battery voltage dropped sharply from 231 V to 210 V, causing the inverter input-side switch Q4 to trip and switch to bypass. After 10 hours of equalization charging, the test was repeated; after approximately 20 minutes of discharge, Q4 tripped again. Why does a battery with a charge capacity of Q=250 Ah, discharging at a rate of 2 h (Q/2 h = 125 A), drop to 210 V in less than 0.5 h? The author believes this is mainly due to an insufficient number of batteries. When the inverter trips at an input of 210 V, the voltage of a single battery is 210 V/171 = 1.23 V. Discharging at a current of 0.5Q/h, the allowable termination voltage is 1.05 V, meaning the effective capacity below 1.23 V is not fully utilized. Referring to the discharge voltage curve for QFD batteries with a discharge coefficient k=0.5 h⁻¹, the time it takes for the battery to drop to 1.23 V is approximately 20 minutes, consistent with the aforementioned experimental results. Therefore, in February 1992, the author drafted suggestions for improving the UPS battery, increasing the number of batteries to 186. Calculations showed that when discharging to 210 V, a single battery still has 210 V/186 = 1.13 V. Referring to the k=0.5 h⁻¹ discharge voltage curve, the discharge can continue for approximately 1.5 hours. The QFD battery discharge voltage curve is shown in Figure 4. Figure 4 Discharge voltage curves with different discharge coefficients. 4.2 Analysis and Correction of the Original Design The original design calculations, based on a battery current of 227 A, a discharge time of 0.5 h, and a charge output of 113.5 Ah (45.4% capacity), consulted the 0.908 h⁻¹ discharge voltage curve. The voltage of a single battery after 0.5 h of discharge was 1.17 V, and the number of batteries n = 210 / 1.17 = 180. Two points need correction: a) The inverter input current, i.e., the battery discharge current, should be 245 A, not 227 A; b) Considering the UPS battery discharge time of 0.5 h, the voltage of 1.17 V is not the battery's final discharge voltage. The corrected calculation is as follows: Based on the battery discharge coefficient k = IDC / Q = 245 A / 250 Ah = 0.98 h⁻¹, consulting the 0.98 h⁻¹ ≈ h⁻¹ discharge voltage curve, the final discharge voltage is 1.03 V. To prevent damage from over-discharge of individual lagging batteries, a margin of safety is allowed, and the discharge termination voltage is set to Upn = 1.03 V × 1.04 = 1.07 V. The number of batteries, n, is the ratio of the inverter's minimum input voltage, UDC, min, to the battery's discharge termination voltage, Upn, i.e., n = 210 / 1.07 = 196. The following conditions must be met: a) The voltage of a single battery after 0.5 hours of discharge should not be less than the battery's discharge termination voltage; b) The voltage of the entire battery pack after 0.5 hours of discharge should not be less than the inverter's minimum input voltage; c) When the discharge reaches the inverter's minimum input voltage, the voltage of a single battery should not be less than the battery's discharge termination voltage; d) The discharge time should not be less than 30 minutes. Verification shows that n = 196 meets the above conditions. 4.3 Discussion Since the UPS inverter has low-voltage protection, to fully utilize the battery capacity, it is desirable to have a larger number of dedicated UPS batteries so that the voltage of each individual battery is lower when the inverter trips due to low voltage, thus extending the discharge time. However, to protect the batteries, the number of batteries should be smaller so that the voltage of each individual battery does not drop too low in the later stages of discharge, preventing over-discharge. Designers should appropriately handle this contradiction. Battery discharge time is closely related to the number of batteries. Under the condition of an output current of 245 A and a selected 250 Ah battery, if 196 batteries are selected, it can discharge for 46 minutes, with each individual battery releasing 76% of its capacity (190 Ah) when the voltage reaches 1.07 V; if 186 batteries are selected, it can release 49% of its capacity (122.5 Ah) when the voltage reaches 210 V/186 = 1.13 V, and can discharge for 30 minutes; if 180 batteries are selected, it can only release 28% of its capacity (70 Ah) when the voltage reaches 1.17 V, with a duration of only 17.2 minutes, which does not reach the expected value of 30 minutes. Therefore, sufficient capacity must be paired with a sufficient number of batteries to achieve the battery's full potential. 5. Inverter Input Voltage Selection The range of the inverter's input voltage depends on the selection of the battery and charger. The inverter input terminal connects to the output terminals of the battery and charger, typically with an isolation diode in between. The inverter's normal input voltage should be higher than the overall voltage of the battery during normal float charging and equalization charging to prevent the battery from easily discharging. To prevent the battery from over-discharging and failing, the inverter must have low-voltage protection. This protection setting is the lower limit of the inverter's input voltage, which directly determines the number of batteries and the DC system voltage, and also affects the battery capacity. The upper limit of the inverter's input voltage restricts the initial charging and equalization charging of the battery. If 196 QFD type batteries are selected, the long-term float charge voltage is 1.35 V × 196 = 264.4 V, which exceeds the range of 220 V × (1 + 10%), significantly reducing component lifespan. The equalization charge voltage is 1.47 V × 196 = 288 V, and the initial charge voltage is 1.63 V × 196 = 319 V, which is higher than the inverter's input voltage limit of 280 V. Batteries must be removed during initial and equalization charges, and this also exceeds the output voltage limit of the GZKC2 charger (310 V). Furthermore, the actual rectifier output voltage is only 278 V, and sometimes fluctuates. Based on these considerations, it is more appropriate to increase the number of batteries to 186 after the UPS upgrade. 6. Conclusion a) From the perspective of optimizing power supply layout and operational needs, it is essential for the design department to draw a UPS system master diagram that includes power supply and load. Many projects lack this diagram. Manufacturer-generated general product diagrams or general principle block diagrams cannot replace actual system construction diagrams. The operations department always expects the design department to produce a separate set of confirmed drawings after digesting the manufacturer's information. After the power plant resumes normal production, a set of as-built drawings that conform to the actual site conditions should also be copied. b) UPS peripheral equipment is closely related to the plant's power supply and DC system. Electrical personnel are more familiar with power supplies, while thermal control personnel are more familiar with load distribution and UPS devices. UPS device selection and ordering are often considered in conjunction with the boiler and turbine control system. Therefore, coordination between specialties is essential for planning and designing a good solution and avoiding oversights at interfaces. Monitoring, power outages, and switching operations of UPS devices are generally performed by operations personnel familiar with electrical systems; however, switches in the UPS, DPU, and APS cabinets at the load end should preferably be operated by thermal control personnel. Only by strengthening management, clarifying responsibilities, and defining division of labor can production units effectively manage UPS systems.