A Protection Device against Switching Overvoltage in DC Traction Networks

Abstract

A scheme of an automatic discharge device for protection against the switching overvoltage based on a controlled vacuum discharger was proposed. The discharge device is designed for shunting the reactor of the traction substation and inductance of the dc traction network during switching overloads accompanied by the power outage. Results of the experimental research and simulation of switching characteristics of such device are presented.

Nonlinear overvoltage suppressors (NOSs) are finding increasing application in dc traction networks. Materials based on zinс oxide (ZnO) with a nonlinear volt–ampere characteristic are used as the active element. However, in the case of overshoot suppression with high amplitudes of ensuing long-duration current, the NOS should be assembled from the large number of the parallel-series connected elements; it leads to the increase of the leakage current, complication of the structure, large dimensional envelope and, therefore, to the decrease of reliability and cost increase of NOS-based protection device. Solution of this problem is important for facilitation of the operation of the quick-operating vacuum switches in dc traction networks.

It has been proposed to use a diode discharge device (DDD) to decrease the current load on an NOS in the case of dc power outage in inductive networks [1–3]. This device shunts the circuit after outage and it contains the series-connected high-power diode discharge resistance. Under normal operating conditions, the diode is closed and connected in a nonconducting direction relative to the network voltage. In the process of disconnection, the current starts falling, while an overvoltage proportional to the current fall is observed on the inductance and the voltage on the DDD alternates in sign, which leads to the opening of the diode. Taking into account that the overloading is mainly concentrated on the inductive traction network (smoothing reactor), it was proposed to apply the special DDD, which is installed parallel to the reactor [4]. Such a device contains a series-connected high-voltage controlled resistor and high-current diode, as well as elements of the thyristor control circuit for the voltage on the reactor. The control circuit generates a command for the opening of the discharger thyristor practically at the beginning of the current fall, at the moment of changing the voltage polarity, because at this moment the voltage polarity on the reactor corresponds to the direction of the thyristor conductivity. However, operation of this device requires time for registration of the overloading and formation of the control impulse for thyristor opening. Application of the semiconductor elements decreases the emergency resistance of the device.

The application of a quick-operating protective device on the basis of the vacuum control dischargers (VCDs) is an alternative, cheaper method for solving the problem of the protection against extra voltage and current overloading [5–7]. Such a device is able to restrict the overloading with high amplitudes of the following long-duration current. The perspectives of using VCDs in a protective device are connected with its ability to operate within the wide range of the working voltages and currents without changing the switching delay time, high resistance to emergency impacts and comparatively low cost as compared with the gas-discharge and semiconductor devices [8].

The article proposes a scheme of a DDD-based automatic protection discharge device (DD) that defends against commutation overloading.

The subject of research is a DDD-based automatic protection DD that defends against impulse overloading. The device is connected parallel to inductance L₁ that includes the inductance of the traction substation reactor and dc networks (Fig. 1). Overloading activates the DD; the part of the accumulated in the L₁ energy scatters. The discharge device contains the series-connected discharge resistor R₁ and controlled vacuum discharger P₁ with the hands-off circuit (HOC); it contains capacitor C₁ and charging resistor R₂. The capacitor is connected by one lead to discharger P₁ and by the other lead to the control electrode of this discharger via peaking discharger P₂ with the stabilized breakdown voltage. The activation circuitry is connected parallel to nonlinear resistor R₃ (NOS) and vacuum switch B. Resistor R₄ is connected between the control electrode and cathode for the protection against the discharger false start.

Fig. 1.

Structural diagram of the discharge device.

The device operates in case of an overloading impulse on the inductive reactor during current outage by switch B. With an increase in voltage, capacitor C₁ starts to charge via resistor R₂ to the voltage U_C(t). After breakdown voltage U_C(t), discharger P₂ turns on and the voltage impulse is observed on control electrode P₁ that induces the breakthrough of the P₁ control interval initiating its start. The P₁ start moment is controlled by time constant τ = R₂C₁.

The experimental scheme is given in the Fig. 2. The oscillating contour contains a capacitor battery of C₂ = 50 mF for the maximum voltage of 6 kV, reactor with inductance L₁ to 6 mH and automatic quick-operating DC vacuum switch of the B type, BVV-3.3–4000/30 UXL3.

Fig. 2.

Structural diagram of the test rig.

Switch B contains commutation module Q consisting of two KVD-3-2000/10 UXL2 parallel-connected vacuum arc extinguishing chambers and a counterflow contour. The commutation module of the switch is shunted by block R₃ from the four parallel-connected NOS. NOS-TP-3.0/4-YXL3 overvoltage suppressors are used as nonlinear resistors. The counterflow contour with series-connected capacitors C₃, inductance L₂, and control vacuum discharger P₄ is connected parallel to module Q.

The switch in the closed condition was connected to the preliminarily charged capacitor battery with the help of control vacuum discharger P₃; the signal for its activation was sent after charging capacitor battery C₂ to the set voltage U₀. Discharge current i started flowing in the circuit; the amplitude was regulated from the change in charge voltage U₀.

The circuit breaker opening was carried out by setting current i_set. The signal for switch B opening and the signal for activating discharger P₄ with a time delay of Δt = 5 ms determined by the time of the switch opening at a distance of 4 mm, were sent at i = I_set. The current with the amplitude of I_p =U_C3/$\sqrt {{{L}_{2}}/{{C}_{3}}}$ > i started to flow in the counter-flow circuit after activation of discharger P₄. As a result, the total current in the vacuum chamber of the commutation module fell to zero and the current was disconnected. In this case, the voltage on the switch contacts, which were suppressed by the NOS at the level of 8 kV, was quickly restored.

The investigated DD was connected in parallel to reactor L₁. The discharge device contains the control vacuum discharger P₁ and discharge resistor R₁ (Fig. 2). Discharger P₁ was activated by HOC which contains capacitor of C₁ = 0.05 μF, inductance is L₃ = 11 μH and peaking discharger P₂ with the stabilized breakdown voltage of 6 kV (Fig. 2). Capacitor C₁ charges via resistor R₂ = 650 Ω for 30 μs after overvoltage on resistor R₃ VCD -31N control vacuum dischargers [9] were used as dischargers P₁, P₃, and P₄.

During the tests, total current I and current i_opn1 via the NOS (R₃) and current i_r via resistor R₁ were measured by current sensors T₁ and T₂ on the basis of the Hall effect; voltage U on the switch was measured by a voltage sensor (VS) on the basis of the Hall effect. Electrical signals were registered by a DPO-4034V digital oscillograph with the following processing of oscillograms on the personal computer.

A DD with discharge nickel–chromium resistance R₁ = 1.8 Ω, inductance L₁ = 6 mH, and current setting I_set = 3 kA was subjected to tests.

Measurement results in this mode at charge voltage of U₀ = 3.3 kV are given in Table 3.

After discharger P₃ was activated, current i started to flow through the switch with a rise speed of di/dt = 550 A/ms; current disconnection in module Q was observed after 10.5 ms when the current reached I_br = I_set + (di/dt)Δt = 5 kA. After disconnection of the current, voltage on the switch U increased rapidly to 8 kV; activation of the DD and division of discharge current i for current i_opn in circuit R₃ and current i in the DD was observed near the voltage maximum. At this moment, current i_opn was 2 kA (about 1/3I_br) and it falls quickly for 4 ms. At this time the deferred drop of voltage U on the NOS was observed. The other part of discharge current i_r flows in the DD; at the initial moment, it was 2.9 kA. Then, against background of the voltage deferred drop, the deferred i_r current drop was observed. After completion of the i_opn current drop, the exponential drop of current i_r, with constant time τ = L₁/R₁ ≈ 3.3 ms for 16 ms, started. During the current drop, the voltage on the switch smoothly decreases down to the residual voltage on the capacitor battery.

The main element determining the DD protection properties is discharge resistor R₁. Upon the set R₁ value the DD characteristics depend on the line voltage and the breaking current. In addition, they are mainly determined by the shunting inductance. Taking into account the wide range of parameters influencing on the choice of the discharge resistor, the numerical simulation of the transition processes in DD at different parameters of the electrical network was carried out.

Simulation was carried out in the MATLAB 6.5 package with the use of the Simulink 4.1 application and Sim Power Systems (SPS) library.

Figure 4 shows a structural diagram of the SPS model used for computation of the commutation process during the experiment (Fig. 2). Keys S₁, S₂, and S₃, which are responsible for commutation of dischargers P₁, P₂, and P₃ and key S₄, which simulates commutation module Q, were designed with the help of an ac switch block. The keys were broken by the current decreasing to zero. The sequences of operation of the model and experimental scheme are similar. Key S₂ closes at the zero moment of time, and control of keys S₁, S₃, and S₄ is carried out from control blocks CU₁, CU₃, and CU₄. To make the diagram simpler, Fig. 4 does not show the content of the blocks. Currents and voltages were registered by virtual current (current measurement) and voltage (voltage measurement) sensors, as well as by a Time Scope virtual oscillograph. The power block is used for setting the calculation parameters of the diagram. Calculation of the energy generated in the NOS and DD was performed by mathematical blocks from the Simulink library.

A surge arrester was used for NOS simulation, consisting in nonlinear resistor R₃. Its volt–ampere characteristic is approximated by the combination of three exponential functions of the following type:

$$\frac{U}{{{{U}_{{{\text{ref}}}}}}} = {{k}_{i}}{{\left( {\frac{I}{{{{I}_{{{\text{ref}}}}}}}} \right)}^{{1/{{\alpha }_{i}}}}},$$

((1))

where U and I are the NOS voltage and current, U_ref and I_ref are the protective voltages, and k_i and α_i are the parameters of the ith region of the nonlinear dependence. Then, coefficients U_ref, I_ref, k_i, and α_i were determined and the model for further application was obtained by comparison of this characteristic with the actual NOS characteristic obtained from the manufacturer.

Figure 3 shows the results of modelling in the oscillating contour for charge voltage U₀ = 3.3 kV at reactor inductance L₁ = 6 mH and discharge resistor R₁ = 1.8 Ω by hatched lines. These results demonstrate close agreement of calculation with the experiment, which demonstrates the applicability of the proposed model for selection of the DD parameters under actual conditions.

Fig. 3.

Current i in discharge contour L₁C₂, current i_opn via resistor R₃, current i_r via the DD, and voltage U on the switch at U₀ = 3.3 kV. Dashed lines are results of numerical simulation.

Fig. 4.

Structural diagram of the SPS model.

A constant-voltage source with a maximum voltage of U₀ = 4.1 kV with addition of active resistor R_sc = 0.117 Ω was used to study the efficiency of DD application in the traction network of Russian Railways in the structural scheme of the SPS model instead of a capacitor. With such circuit parameters, the expected maximum short-circuit current is I_sc = 35 kA. Computation results for R₁ = 1.8 Ω and L₁ = 6 mH are given in Fig. 5.

Fig. 5.

Change in currents and voltages at different braking currents I_br: i is the current in the network before current break; i_opn1, i_opn2, and i_opn3 are the currents via the NOS; i_r1, i_r2, and i_r3 are the currents via the DD; and U₁, U₂, and U₃ are the voltages on a switch; I_br = (1) 10, (2) 7.5, and (3) 5 kA.

It is seen from Fig. 5 that, at a circuit break of I_br = 10 kA, the DD connects and current i splits divides into current i_opn1 via the NOS (i_opn1/I_br = 77%) and current i_r1 via discharge resistor R₁ (i_r1/I_br = 23%). With a decrease of current I_br to 5 kA, the share of current via the NOS i₃/I_br decreases to 52%.

The duration of the decrease in current in the NOS at I_br = 5 kA increased as compared with the duration of the current decrease in the oscillation contour from 3 to 5 ms (Fig. 3), possibly because of the inflow from the supply source [1]. Time t_br of complete current break increased as well; it consists of the current fall time in the NOS and the current fall time in discharge resistor R₁.

Figure 6 shows the dependences of time t_br of complete current break on the resistance of discharge resistor R₁ for different inductance L₁ at U₀ = 4.1 kV. As a result of simulation, it is shown that time t_br increases with an increase of inductance L₁ and decrease of R₁. The conditions of equality of time t_br of the complete current break of the BVV vacuum switch and VAB electromagnetic switch was proposed as the resistance selection criterion for the R₁ resistor. For the VAB-206 switch, time t_br is greatest at a network inductance of 15 mH, with a value of 50 ms [10]; this condition implies an ensuing resistance of R₁ = 2.7 Ω.

Fig. 6.

Dependence of time of complete current break on resistance of the discharge resistor at L₁ = (1) 15, (2) 10, and (3) 5 mH.

Energy W_opn = W_L + W₀ scatters in the DD-free network in the case of a current break in the NOS. It consists of energy W_L = L₁$I_{{{\text{br}}}}^{2}$/2 accumulated in the network inductance and energy W₀ coming from the source. In the network with the DD, energy W_L + W₀ is spent on energy W_opn1 scattered in the NOS and energy W_R scattered in discharge resistor R₁. Current i_opn1 in the NOS with a DD and its duration are smaller than for the DD-free situation. As a consequence, the share of energy W_opn1 coming from the source to the NOS decreases. Energy W_opn1 depends on resistance of the discharge resistor R₁, network voltage U₀, breaking current I_br, and network inductance L₁. The computation dependences of energies W_opn, W_opn1, and W_R on network inductance L₁ for selected resistance R₁ = 2.7 Ω in the case of current break I_br = 10 kA and voltage U₀ = 4.1 kV are given in Fig. 7.

Fig. 7.

Dependence of energy W scattered in an NOS and discharge resistor on inductance L₁ of the network at current break I_br = 10 kA and voltage U₀ = 4.1 kV: (1) W_opn in an NOS without a DD, (2) W_opn1 in an NOS with a DD; energy W_R scattered in resistance of the discharge resistor R₁.

It is seen from the above results that, with a decrease in inductance, energies W_opn and W_opn1 scattered in the NOS decrease proportionally:

$$\begin{gathered} \frac{{{{W}_{{{\text{opn}}}}}\left( {{\text{mH}}} \right) - {{W}_{{{\text{opn1}}}}}\left( {{\text{15}}\,{\text{mH}}} \right)}}{{{{W}_{{{\text{opn}}}}}\left( {{\text{15}}\,{\text{mH}}} \right)}} \\ = \frac{{{{W}_{{{\text{opn}}}}}\left( {{\text{6}}\,{\text{mH}}} \right) - {{W}_{{{\text{opn}}1}}}\left( {{\text{6}}\,{\text{mH}}} \right)}}{{{{W}_{{{\text{opn}}}}}\left( {{\text{6}}\,{\text{mH}}} \right)}} = 27\% . \\ \end{gathered} $$

((2))

In this case, energies W_opn1 and W_R scattered in the NOS and resistance of the discharge resistor R₁ decrease by 60% with a decrease in inductance L₁ from 15 to 6 mH.

It follows from the computation results at U₀ = 4.1 kV, L₁ = 15 mH, R₁ = 2.7 Ω, and different breaking currents that, with a decrease in the current to I_br = 5 kA, the share of energy W_opn1 energy scattered in the NOS with the use of a DD decrease as follows relative to the DD-free energy scattered in an NOS:

$$\frac{{{{W}_{{{\text{opn}}}}}\left( {5\,{\text{kA}}} \right) - {{W}_{{{\text{opn}}}}}\left( {5\,{\text{kA}}} \right)}}{{{{W}_{{{\text{opn}}}}}\left( {5\,{\text{kA}}} \right)}} = 47\% .$$

((3))

Thus, the results of experiments and simulation showed that the proposed device for protecting against commutation overloading on the basis of a control vacuum discharger intended for shunting a traction substation reactor and inductance of a dc traction network contributes to decrease of energy scattered in an NOS and facilitates vacuum switch operation.