The following is a case of UPS and generator compatibility issues, when an online service provider's new data center occurs during commissioning. It shows how vendors, engineers, and users discover and solve problems.
There are 3 sets of MGEUPS3000kVA systems on site, each consisting of 4 sets of 75kVA IGBT wide-band FM modules, which can be expanded to 6 sets. The module's design load factor is 65%, and the UPS module is equipped with an input isolation transformer and a maximum 5% input current harmonic filter. All modules are connected to two sets of generator parallel buses. Each group of buses has three 1600kW generators, which can be expanded to six. Each generator is equipped with an electronic regulator. The power conversion plan for each parallel bus is to wait for the two generators to be connected in parallel before the first load is connected. The first load contains one UPS and some air conditioning loads in each system.
With the incorporation of the subsequent generator, the same load as the first batch is subsequently added. In the failure mode test, the operator found that when one of the two generators with the first load had a fault, the other would have an overvoltage alarm and shut down after 2 s. But the first load is much lower than the capacity of a generator because the load on the UPS is very light. Further testing was arranged to determine the impact of the UPS on a single generator. Because the first suspect is the input link of the UPS to the regulator*, the tested UPS does not have a load, or the UPS inverter is turned off. The test set consists of a DC voltage and ammeter to directly monitor the field excitation coils, as these parameters are controlled by the regulator and immediately reflect the action of the regulator. At the same time, the power (W), current and voltage (VA), and var (var) of the load are monitored by the instrument of the generator itself.
The test is first performed with a pure resistive load to establish a baseline. It shows that the excitation current and voltage rise as the load increases, as we expected. The larger load current produces a larger voltage drop I x Z on the internal resistance Z of the generator, which must be overcome to keep the output voltage U stable. Then test the impact of the UPS on the generator, one at a time. The UPS has no load and observes the soft start process of the UPS rectifier. The test results clearly show that the action of the regulator is the opposite of that of a purely resistive load. After connecting two UPSs, the regulator is close to the edge of the allowable range, and one is added to make the generator enter the overload state after 2s.
At this point, pay attention to the load value corresponding to a single 750kVA UPS. It causes the generator to shut down substantially without real load. The capacity of each UPS is close to 230kvar, which makes the power factor zero.
A project team consisting of engineers, owners, contractors, suppliers and vendors, after considering all the possibilities, chose a solution to install a reactive reactor on each capacitive load. According to the data tested above, the manufacturer designed a 200kvar shunt reactor for each UPS and controlled by the contactor. The contractor installed it in parallel with the input filter of the UPS. The engineer designed the external control circuit to measure the generator. The load is allowed to be only allowed when the UPS is powered by the generator and the total load of the generator is below a (adjustable) set point. The project team re-tested with a modified UPS connected to a generator.
At this time, the effect of the capacitor still exists, and the reactor can only balance part of the capacitor instead of all. Therefore, as the UPS increases, the field current gradually decreases, but this does not cause a problem. Because the six UPSs have exceeded the capacity of one generator, the regulator is still normal and controls the output voltage.
