**1. Introduction**

Theoretically, the converters in the DC grid can be connected in parallel, in series, or in hybrid mode [1]. For the parallel converters, the DC voltage of each converter is roughly the same, and the power distribution can be achieved by controlling the DC current of each converter. For converters connected in series, the current flowing through each converter is the same, and the power distribution can be regulated by adjusting the DC voltage of each converter. However, the converters connected in parallel are more suitable for DC grids due to the higher reliability, economy, and flexibility. In this configuration, the voltage polarity of each converter is fixed. In the early ages, most high voltage direct current (HVDC) systems were based on the line commutated converters (LCCs), and the DC power flow direction was fixed since the current of the thyristor could not be reversed. Due to this shortcoming in LCC, DC grid has not received much attention in the first 50 years of the HVDC transmission. However, with the development of power electronics, the voltage source converter-based HVDC (VSC-HVDC) [2–5] has drawn much attention from both academia and industry due to its flexible operation capability, and the advantages of a DC grid can be fully achieved by VSC.Ever since, the DC grid has become a new expectation of the power industry.

There are three main technical bottlenecks in the development of DC grids. One is the fast detection and isolation technology of DC faults [6,7]. The second is the DC voltage transformation technology [8,9]. The third is the power flow control technology of the DC lines [10,11]. This paper mainly deals with the first technical bottleneck and studies the fault protection principle for DC grids.

The di fficulties of DC grids in fault protection lie mainly in the following aspects [12–15]:

(1) The fault current rises very quickly. Generally, the fault current could reach its steady value within 10 ms after the fault;


To this end, this paper proposes a local protection and local action strategy for HVDC grids. This strategy is of high protection selectivity and speed, which is able to solve the above technical problems effectively.

This paper is organized as follows. Section 2 gives the basic structure and operation principle of the hybrid circuit breaker. The local protection and local action strategy is introduced in Section 3. Section 4 describes the test system constructed in this paper. To verify the improved protection speed and selectivity, a performance comparison between the proposed strategy and the conventional relay protection strategy is presented in Section 5. The conclusions are drawn in Section 6.

#### **2. Basic Structure and Operation Principle of DC Circuit Breaker**

The HVDC CB adopted in this paper is the hybrid HVDC CB proposed by ABB [16,17], as shown in Figure 1. The HVDC CB is connected between one multi-level modular converter (MMC) and one DC line. When the faults occur on the DC lines, the HVDC CB starts action to isolate the fault line.

**Figure 1.** Structure of the hybrid high voltage direct current circuit breaker (HVDC CB).

The DC circuit breaker is composed of a normal current path, a main breaker branch, and an energy absorption branch. The normal current path consists of a load commutation switch (LCS) and an ultra-fast disconnector (UFD) connected in series. Multiple main breakers (MBs) connected in series constitute the main breaker branch. The basic structure of each part is as follows:


(4) Ultra-fast disconnector (UFD). It needs to quickly disconnect the circuit under zero current state. Furthermore, the breaking time is about 2 ms.

The principle of the hybrid HVDC CB is as follows [18–20]:


#### **3. Two Basic Protection Strategies for DC Grids**

For a half-bridge sub-module (HBSM) multi-level modular converters (MMCs)-based DC grid equipped with HVDC CBs, how to quickly detect and isolate the DC faults is a very challenging problem.

The conventional strategy is to follow the practice of the AC grid. The protection action sequence is shown in Figure 2. Firstly, the relay protection system detects the fault location, and then the fault line is isolated by CBs. However, this strategy puts extremely high requirements on the speed and selectivity of the relay protection system. According to the speed of fault detection in conventional two-terminal HVDC systems, the duration for fault detection is about 10 ms [21]. If the fault detection speed in DC grids is the same as that in the conventional two terminal HVDC system, the fault current to be interrupted by the DC breaker will reach a very high level. Furthermore, this will result in huge costs of CBs. In addition, this will severely limit the application of DC grids of this structure.

**Figure 2.** Protection action sequence of strategy 1.

The other strategy is based on local protection and local action. There could be two aspects to explain local protection. The first is the local protection of the converters. If the arm current reaches twice the rated IGBT current, the converter will then be automatically blocked. Moreover, there is no need for a relay protection signal. The second means the local protection of CBs. When the current flowing through CB is more than twice its rated current, LCS will be activated immediately as well as the whole CB. That is, CBs on both sides of the line independently finish fault detection and tripping, and there is no need for a coordination between CBs. Local action means that CB only operates when the fault occurs at the DC line where CB is located. Define that the positive direction of the CB current is from MMC to the DC line. Then, only when the fault current is the same as the positive direction and reaches twice the rated CB current, CB is activated. Local action guarantees the high protection selectivity. The protection action sequence is shown in Figure 3. Practice has shown that this strategy is very suitable for DC grids. Furthermore, it has high protection speed and protection selectivity, which could greatly reduce the required breaking capacity of CBs, thereby saving the cost.

**Figure 3.** Protection action sequence of strategy 2.

The two different strategies will be studied separately in the following sections. The first strategy is to deal with the faults by the conventional DC fault relay protection. The second strategy is the local protection and local action strategy.
