Open-Circuit Fault Mitigation for Inverter-Driven Induction Motor Based on Closed-Loop Volt-per-Hertz

Abstract

Presented in this paper is a mitigation technique for an open-circuit fault (OCF) in a closed-loop Volt-per-Hertz controlled three-phase induction motor. Conventional proportional–integral (PI) controllers have been found inadequate for maintaining stable motor performance during the fault and exhibit significant transient issues when transitioning from fault to normal operation. To address these limitations, a proportional–resonant (PR) control method and a proportional–integral–resonant (PIR) control method are proposed. The PIR controller enhances the traditional PI controller by integrating a resonant component, enabling effective performance during the fault and improving transient responses during pre-fault conditions. Experimental validation using a dSPACE DS1104 platform demonstrates that the PR and PIR control methods significantly improve motor performance compared to the PI method. The proposed approaches eliminate the need for fault detection, offering a simpler and cost-effective alternative for maintaining motor reliability and efficiency under fault conditions. These results underscore the potential of the proposed method as a robust solution for the fault scenarios in industrial applications.

1. Introduction

Three-phase induction motors are extensively used across various industrial applications because of their high reliability, high efficiency, robust power output, and minimal maintenance requirements [1]. Over the years, significant advancements have been made in induction motor control techniques, broadly categorized into scalar and vector control methods, as illustrated in Figure 1. Scalar control, also known as Volt-per-Hertz (V/f) control, regulates motor speed by adjusting the voltage and frequency while maintaining a constant V/f ratio. On the other hand, vector-based control techniques, including Field-Oriented Control (FOC) and Direct Torque Control (DTC), offer advanced dynamic performance by directly controlling motor torque and flux [2].

Sensorless scalar control methods have been explored to improve cost efficiency and control precision. For instance, an artificial neural network-based sensorless control method has been proposed to estimate motor speed using current measurements and space vector pulse-width modulation (SVPWM). This approach achieved accurate speed estimation and reduced system noise while lowering overall system cost [3]. Additionally, a stability analysis of induction motors under light and heavy loads using open-loop V/f control has been studied, highlighting limitations in transient response [4]. Similarly, comparative analyses of open- and closed-loop V/f control methods revealed that, while closed-loop systems improve steady-state error, they often exhibit slower transient response times [5].

The application of V/f control in automotive systems showcases its efficiency and affordability, particularly for electric vehicle (EV) applications, where dynamic load and speed conditions are critical. The use of Texas Instruments’ F28069M LaunchPad has demonstrated effective motor control with minimal costs, making induction motors an appealing alternative to DC drives in automotive applications [6].

Despite the advancements in vector control techniques, scalar control remains a vital solution for applications requiring simplicity and cost efficiency. FOC and DTC offer superior dynamic performance but come with increased system complexity and cost. Induction motor control using the FOC algorithm is discussed in [7,8,9,10], while the DTC algorithm for induction motor control is detailed in [11,12,13,14]. Additionally, a comparative analysis of FOC and DTC performance is provided in [15,16].

Induction motors, while reliable, are susceptible to various faults that can affect their performance, reliability, and lifespan. Common fault types, including OCFs, are depicted in Figure 2 [17]. Under such fault conditions, the motor experiences unbalanced currents, reduced torque production, decreased efficiency, and increased vibration, which can lead to severe mechanical and electrical damage [18,19].

To overcome these challenges, fault-tolerant control (FTC) systems have been designed. FTC systems use advanced algorithms to detect faults rapidly, adjust operational parameters, or switch to alternate control modes to ensure continued motor operation, albeit with reduced performance.

Extensive research has focused on FTC operation under open-phase conditions in multiphase induction machines. For instance, studies in [20,21,22] developed FTC algorithms that integrate the nonlinear characteristics of both the machine and the converter when modeling OCF drive systems. In [23], an FTC strategy for a five-phase induction motor was proposed, enabling four-phase and three-phase mode operation under variable voltage and frequency control. A remedial approach involves compensating for the faulty phase using a modified reference signal, which can be efficiently implemented via carrier-based pulse-width modulation (PWM) techniques.

Further advancements include the model predictive current control (MPCC) method for five-phase induction motors presented in [24]. This approach offers rapid response and ripple-free torque performance under fault conditions. However, the proposed method requires large computational resources to perform an optimization problem to mitigate OCF. Similarly, an FTC scheme based on DTC was proposed in [25] for five-phase induction motors under OCF conditions. Experimental results demonstrated the ability of the scheme to maintain speed, torque, and flux references post-fault, provided that the applied load torque does not exceed the maximum capacity of the post-fault control strategy.

Additionally, ref. [26] introduced an FTC system for five-phase induction motors with trapezoidal back electromotive force, utilizing both fundamental and third-harmonic current components. Meanwhile, ref. [27] presented an enhanced FTC scheme for PWM inverter-fed induction motors, specifically tailored for EV applications. This method improves the system’s fault-tolerant capabilities, ensuring reliable performance even under challenging conditions.

However, FTC systems often introduce additional complexity and higher implementation costs, posing challenges in resource-constrained environments. Therefore, this paper focuses on a cost-effective and efficient solution: a closed-loop V/f control technique for three-phase induction motors under OCF conditions. The proposed method enhances motor performance during OCF operation without requiring FTC implementation, reducing system cost and complexity.

This study contributes to the field of induction motor control by proposing and experimentally validating a PR and PIR controller. These controllers improve the motor’s performance under OCF conditions and transient response during pre-fault operations.

The key contributions of this paper are outlined as follows:

• Demonstration of the limitations of open-loop V/f control under OCF conditions;
• The PR controller is analyzed to maintain continuous motor operation during fault and post-fault operation as the PI controller fails to achieve such performance;
• The PI control method fails to maintain continuous motor operation during the OCF due to the double sinusoidal term;
• A new novel controller using the PIR controller is introduced. The resonant controller (RC) is combined with a PI controller, resulting in a PIR controller, which makes the new modified controller work effectively under OCF conditions;
• The PIR control exhibits a better dynamic performance under OCF with minimal torque ripple compared to the PI and PR controller behaviors;
• Elimination of the need for fault detection systems, offering a simpler and cost-effective solution for OCF operation.

The rest of the paper is structured as follows. Section 2 outlines the mathematical model of the induction motor. Section 3 presents the proposed control algorithms under OCF conditions. Section 4 discusses the experimental setup and case studies. Lastly, Section 5 concludes the paper.

2. System Model Equations

The three-phase model of the three-phase induction motor is simplified into a two-phase model, known as the $dq$ model, using Park’s Transformation. The stator and rotor voltages in the $dq$ model are shown in Equations (1)–(4) [28].

$$v_{ds}=R_s{i}_{ds}+L_s{\displaystyle \frac{d{i}_{ds}}{dt}}+L_m{\displaystyle \frac{d{i}_{d}r}{dt}}$$

(1)

$$v_{qs}=R_s{i}_{qs}+L_s{\displaystyle \frac{d{i}_{qs}}{dt}}+L_m{\displaystyle \frac{d{i}_{q}r}{dt}}$$

(2)

$$v_{dr}=R_r{i}_{dr}+L_m{\displaystyle \frac{d{i}_{ds}}{dt}}+L_r{\displaystyle \frac{d{i}_{dr}}{dt}}+\omega L_m{i}_{qs}+\omega L_r{i}_{qr}$$

(3)

$$v_{qr}=R_r{i}_{qr}+L_m{\displaystyle \frac{d{i}_{qs}}{dt}}+L_r{\displaystyle \frac{d{i}_{qr}}{dt}}-\omega L_m{i}_{ds}-\omega L_r{i}_{dr}$$

(4)

where $R_s$ and $R_r$ are the stator and rotor resistance, respectively; $L_s$ and $L_r$ are stator and rotor inductance, respectively; $L_m$ is the mutual inductance; $v_{ds}$ and $v_{qs}$ are stator voltages in the $dq$ reference frame; $v_{dr}$ and $v_{qr}$ are rotor voltages in the $dq$ reference frame; $i_{ds}$ and $i_{qs}$ are stator currents in the $dq$ reference frame; $i_{dr}$ and $i_{qr}$ are rotor currents in the $dq$ reference frame; and $\omega$ is the reference frame speed.

The electromagnetic torque developed by the motor is expressed in the $dq$ reference frame model, as in Equation (5).

$$T_e={\displaystyle \frac{p}{2}}{\displaystyle \frac{3}{2}}{L}_m[i_{qs}{i}_{dr}-i_{ds}{i}_{qr}]$$

(5)

where p is the number of poles.

3. Closed-Loop V/f Control Techniques Under OCF

The closed-loop V/f control technique is widely used in regulating the speed of induction motors. This technique involves maintaining a constant ratio between the motor’s voltage and frequency, which is essential for achieving stable and efficient motor operation. Different control methods are proposed to achieve closed-loop control and regulate the induction motor speed during healthy, OCF, and post-fault operations.

The speed control of an induction motor is achieved by maintaining a constant voltage-to-frequency ratio. In this method, the stator voltage and frequency increase proportionally until they reach their maximum values at the base speed. For frequencies above the rated value, the stator voltage must be kept within its rated limit to prevent insulation breakdown. The rotor speed, input voltage frequency, and number of poles are interrelated, as expressed by Equation (6).

$$n={\displaystyle \frac{120f}{p}}$$

(6)

where f is the fundamental frequency of the input voltage, P is the number of poles, and n is the speed in rpm.

The closed loop V/f control technique for the induction motor control under OCF was investigated using three different control strategies: PI control, PR control, and PIR control. The Simulink model of closed-loop V/f control technique is presented in Figure 3.

When the IM is working in a healthy situation, the three-phase stator currents are as shown in Equation (7).

$$\{\begin{array}{c}{i}_a=I_{m}cos({\omega }_{s}t+\varphi )\\ i_b=I_{m}cos({\omega }_{s}t-{\displaystyle \frac{2\pi }{3}}+\varphi )\\ i_c=I_{m}cos({\omega }_{s}t+{\displaystyle \frac{2\pi }{3}}+\varphi )\end{array}$$

(7)

where ${\omega }_s$ is the phase current frequency, $I_m$ is the amplitude of each phase current in healthy conditions, and $\varphi$ is the initial phase angle in healthy conditions.

The rotor flux position ${\theta }_s$ is calculated as in Equation (8).

$${\theta }_s=\int {\omega }_{s}dt$$

(8)

In healthy operation, the dq currents can be expressed as in Equation (9).

$$\{\begin{array}{c}{i}_d=I_{m}cos\left(\varphi \right)\\ i_q=I_{m}sin\left(\varphi \right)\end{array}$$

(9)

When OCF occurs in phase A, the current in phase A ($i_a$) will drop to zero, while the current in the remaining phases will increase and become distorted, as in Equation (10). This leads to increased speed and torque ripples, causing the motor to produce excessive vibration and noise. As a result, severe mechanical and electrical damage may occur, ultimately compromising the motor’s operation, longevity, and reliability. Therefore, the control method was modified to effectively manage the challenges posed by such faults.

$$\{\begin{array}{c}{i}_a=0\\ i_b=-i_c=I_{mf}cos({\omega }_{s}t+{\varphi }_f)\end{array}$$

(10)

In OCF operation, the dq currents can be expressed as in Equation (11):

$$\{\begin{array}{c}{i}_d=\underset{\mathrm{Double}\mathrm{Sinusoidal}\mathrm{term}}{\underbrace{-{\displaystyle \frac{\sqrt{3}}{3}}{I}_{mf}sin(2\theta +\varphi )}}+\underset{\mathrm{DC}\mathrm{Value}}{\underbrace{{\displaystyle \frac{\sqrt{3}}{3}}{I}_{mf}sin\left({\varphi }_f\right)}}\\ i_q=\underset{\mathrm{Double}\mathrm{Sinusoidal}\mathrm{term}}{\underbrace{-{\displaystyle \frac{\sqrt{3}}{3}}{I}_{mf}cos(2\theta +\varphi )}}+\underset{\mathrm{DC}\mathrm{Value}}{\underbrace{{\displaystyle \frac{\sqrt{3}}{3}}{I}_{mf}cos\left({\varphi }_f\right)}}\end{array}$$

(11)

3.1. Closed-Loop V/f Control Using PI Control Method

The PI is a simple control method and widely used in induction motor control. The PI controller is employed in the outer speed control loop, with its transfer function defined by Equation (12). The gains of the PI controller are chosen based on the Ziegler-–Nichols (ZN) method, as described in [29,30].

$$G_{PI}=K_p+{\displaystyle \frac{K_i}{s}}$$

(12)

where $K_p$ and $K_i$ are the proportional and integral gains, respectively.

The block diagram of induction motor control based on a V/f technique using a PI control method is presented in Figure 4.

The PI control strategy is a fundamental method and widely used in induction motor control. However, under OCF conditions, the PI controller may fail to effectively manage the motor operation because the double sinusoidal term exists in $I_d$ and $I_q$. Therefore, this study proposes the closed-loop V/f control for three-phase induction motors, employing PR and PIR control methods.

3.2. Closed-Loop V/f Control Using PR Control Method

The PR controller is proposed for the closed-loop V/f induction motor control under OCF. The block diagram for this method is illustrated in Figure 5.

The transfer function of the PR controller is described by Equation (13).

$$G_{PR}=K_p+{\displaystyle \frac{K_{r}s}{s^2+{\omega }_r^2}}$$

(13)

where $K_p$, $K_r$, and ${\omega }_r$ are the proportional gain, resonant gain, and resonant frequency, respectively.

The PR controller’s gains are selected following a structured tuning approach, similar to the well-established Ziegler–Nichols method, which does not require Bode plots or stability margin analysis [18]. The tuning process is carefully performed as follows:

• The proportional gain ($K_p$) is set to a small value to ensure system stability while minimizing overshoot in the system’s performance;
• The resonant gain ($K_r$) is tuned to enhance disturbance rejection and improve fault tolerance. However, excessive $K_r$ values may introduce system instability, so a balance must be maintained;
• The resonant frequency (${\omega }_r$) is determined directly from the fundamental frequency of the motor currents under OCF conditions, ensuring accurate tuning. It is calculated as

  $${\omega }_r=2{\omega }_s=4\pi f_s=4\pi \left(n_s{\displaystyle \frac{P}{120}}\right)=n_s{\displaystyle \frac{\pi P}{30}}$$

  (14)

  $$f_s=n_s{\displaystyle \frac{P}{120}}$$

  (15)

  where P is the number of poles, $n_s$ is the desired reference speed in rpm, and $f_s$ is the frequency at the desired speed. This method ensures that ${\omega }_r$ is accurately synchronized with the motor dynamics without requiring iterative tuning techniques;
• The PR controller gains are selected through careful tuning, striking a balance between improved control performance and maintaining system stability, similar to classical tuning methodologies.

3.3. Closed-Loop V/f Control Using PIR Control Method

Because a PI controller is inadequate for closed-loop V/f induction motor control under OCF operation, it can be modified by incorporating a resonant controller, resulting in a PIR controller.

The PIR controller combines proportional, integral, and resonant control to improve the overall performance of the induction motor under OCF operation and during the pre-fault operation. The block diagram of closed-loop V/f induction motor control based on a PIR control method is shown in Figure 6.

Equation (16) presents the transfer function of the PIR controller. The gains of the PIR controller are chosen based on combing the ZN and PR control procedures, as mentioned in Section 3.1 and Section 3.2.

$$G_{PIR}=K_p+{\displaystyle \frac{K_i}{s}}+{\displaystyle \frac{K_{r}s}{s^2+{\omega }_r^2}}$$

(16)

The PIR controller’s gains are selected using a structured tuning approach, similar to the well-established Ziegler–Nichols method, which does not require Bode plots or stability margin analysis. The tuning process is carefully performed as follows:

• The proportional gain ($K_p$) is set to a small value to ensure system stability while minimizing overshoot in the system’s performance. It can be tuned in the same manner as Z–N method;
• The integral gain ($K_i$) is tuned to balance steady-state error reduction and dynamic response. While increasing $K_i$ reduces steady-state error, excessive values may introduce overshoot and oscillations, so careful tuning is essential. It can be tuned in the same manner as Z–N method;
• The resonant gain ($K_r$) enhances disturbance rejection and improves performance under OCF conditions. However, excessive $K_r$ values may compromise system stability, necessitating a well-balanced selection. It can be tuned in the same manner as in the PR controller tunning method;
• The resonant frequency (${\omega }_r$) is determined directly from the fundamental frequency of the motor currents under OCF conditions, which can be tuned in the same manner as in the PR controller tunning method. It is calculated as

  $${\omega }_r=n_s{\displaystyle \frac{\pi P}{30}}$$

  (17)

  where P is the number of poles and $n_s$ is the desired reference speed in rpm. This ensures that ${\omega }_r$ is accurately synchronized with the motor dynamics without requiring iterative tuning techniques;
• The PIR controller gains are selected through careful tuning, ensuring an optimal balance between improved control performance and maintaining system stability, similar to classical tuning methodologies.

4. Experimental Case Studies

The closed-loop V/f control of a three-phase induction motor using PI, PR, and PIR control methods has been experimentally verified. The test setup consists of a three-phase induction motor powered by a 42 V source through an inverter. The motor control system is implemented using the dSPACE DS1104 platform (dSPACE GmbH, Paderborn, Germany), with real-time angular position and speed data provided by an encoder. The switching frequency is set to 10 kHz. Figure 7 illustrates the experimental setup, and the motor parameters are detailed in

Table 1. Three-phase induction motor parameters.

Parameters	Value	Unit
Rated.Power	120	W
Rated.Volts	30	$V_{AC}$
Max.Speed	4000	RPM
Rated.Amps	6	$A_{rms}$
Resistance.(L-L)	0.7	Ohms
Inductance	2.27	mH

.

4.1. PI Case Study

The experimental results for an induction motor operating under healthy, OCF, and post-fault conditions are presented in this section.

Initially, the motor runs in a healthy mode at 500 rpm reference speed. An OCF is introduced in phase A at $t=13.9\mathrm{s}$, and the fault is cleared at $t=17.8\mathrm{s}$. The results are divided into three distinct regions: Zoom 1 captures healthy operation, Zoom 2 highlights the faulted operation, and Zoom 3 depicts the post-fault operation.

In Zoom 1, the speed controller successfully tracks the reference speed, demonstrating the precise and stable performance of the PI controller under normal operating conditions.

In Zoom 2, during the OCF, the motor speed drops to zero, resulting in a complete cessation of rotation. Concurrently, the current in phase A falls to zero, while the currents in the remaining two phases exhibit distortions, accompanied by pronounced torque ripples.

In Zoom 3, representing the post-fault operation, significant transients in speed and torque responses are observed. These transients cause the motor to take additional time to stabilize and reach steady-state conditions.

The experimental results for the closed-loop V/f induction motor control employing the PI control method under healthy, OCF, and post-fault conditions with their transients for speed, three-phase currents, dq currents, and torque are illustrated in Figure 8, Figure 9, Figure 10 and Figure 11.

Under OCF, because of the double sinusoidal term in $I_d$ and $I_q$, the PI controller will fail to track a sinusoidal reference signal because it is not fast enough. Therefore, using a PI controller results in a complete speed drop to zero during an OCF. This would have severe consequences, particularly for applications like EVs, where a sudden halt in motor speed could cause the EV to stop abruptly in the middle of the road—a dangerous and unacceptable outcome during the fault region.

When the fault is cleared, the PI controller leads to a significant overshoot in speed at the startup of the post-fault region, with the motor accelerating to a dangerously high value before stabilizing back to the desired speed. For an EV, this behavior is highly undesirable, as it could jeopardize safety during normal operation.

The results demonstrate that the PI control method is not effective for controlling induction motors under OCF conditions. Additionally, it leads to significant transients in speed, current, and torque during pre-fault operation. As a result, it is essential to modify the PI control method to make it suitable for induction motor control under OCF conditions. Therefore, advanced control techniques have been proposed to improve closed-loop V/f induction motor control during OCF operation.

4.2. PR Case Study

This section details the implementation of the PR controller within a closed-loop V/f control system during OCF conditions. Initially, the motor operates under healthy conditions with a reference speed of 500 rpm. An OCF occurs at phase A at $t=5.59$ s and is resolved at $t=11.5$ s. Experimental results are captured during healthy, OCF, and post-fault conditions with their transients for speed, three-phase currents, dq currents, and torque illustrated in Figure 12, Figure 13, Figure 14 and Figure 15.

During OCF operation, the current ($i_a$) drops to zero, while the remaining currents ($i_b$ and $i_c$) become distorted, leading to increased ripples in both speed and torque responses. Despite these distortions, the PR controller effectively tracks the reference values, exhibiting only a minor transient drop with minimal torque ripples during the OCF interval, as highlighted in the Zoom 1 region.

Since the PR controller has the resonant term, the resonant frequency can be taken from the double sinusoidal term. Therefore, using a PR controller reduces the severity of the fault’s impact. While speed drops and fluctuates between 400 rpm and 500 rpm (with a pre-fault speed of 500 rpm) in the fault region, the EV remains operational, albeit with some instability in speed. This outcome is safer compared to the PI controller, but the fluctuations can still impact the driving experience.

In the post-fault operation, illustrated in the Zoom 2 interval, the controller demonstrates superior transient response and achieves steady-state performance quickly. This highlights the controller’s robustness in maintaining performance during normal operation.

Overall, the proposed PR control strategy ensures continuous motor operation under OCF conditions, maintaining low speed and torque ripples during both fault and post-fault scenarios. This contributes to enhanced system reliability and performance, as observed in the experimental results presented in the Zoom 1 and Zoom 2 intervals.

4.3. PIR Case Study

Since traditional PI controllers are inadequate for closed-loop V/f induction motor control under OCF conditions, their performance can be enhanced by incorporating a resonant controller, resulting in a PIR controller.

In the experiments, the motor operated under healthy conditions with a reference speed of 500 rpm. An OCF occurred in phase A at $t=6.04\mathrm{s}$ and was cleared at $t=11.05\mathrm{s}$. During the OCF interval, $i_a=0$, causing distortion in the currents of the remaining phases. This led to a slight increase in ripples within the speed and torque responses.

The experimental results demonstrate that the integration of the PI and resonant controllers enables the PI controller to effectively manage closed-loop V/f induction motor control during OCF conditions. Additionally, the proposed PIR controller provides improved transient performance during the post-fault period. The results of the PIR control method for speed, three-phase currents, dq currents, and torque are illustrated in Figure 16, Figure 17, Figure 18 and Figure 19.

The PIR controller further improves the response during fault conditions. The speed fluctuates minimally, between 450 rpm and 500 rpm, indicating much better fault tolerance during the fault region. The EV remains operational with significantly reduced fluctuations, ensuring a more controllable and reliable performance under OCF conditions.

A comparative analysis of PI and PR control methods for V/f induction motor control under OCF conditions highlights the superior performance of the proposed PR method. Experimental results confirm that the PR controller ensures continuous motor operation during OCF conditions and significantly enhances speed and torque response during post-fault operation. Conversely, the PI controller fails to sustain motor control under OCF conditions, leading to pronounced transients in speed and torque responses upon fault clearance.

To address this limitation, the PI controller can be augmented with a resonant controller, resulting in the PIR control approach. This enhancement enables the PI controller to perform effectively in managing induction motor control under OCF conditions. Experimental findings further validate the PIR control method’s reliability in maintaining stable and efficient induction motor operation in both OCF and post-fault scenarios.

5. Conclusions

This study advances the field of induction motor control by introducing and experimentally validating PR and PIR controllers to enhance performance under OCF conditions. The findings highlight the limitations of the conventional PI controller, which fails to sustain motor operation during OCF, leading to a complete speed drop, distorted phase currents, and severe torque ripples. In contrast, the PR controller maintains continuous motor operation by mitigating the impact of the fault, though it introduces speed and torque fluctuations due to the resonant term. To address these challenges, a novel PIR controller is proposed, integrating an RC with a PI controller, resulting in superior fault-tolerant performance.

The experimental results demonstrate that the PIR controller effectively minimizes speed fluctuations and torque ripples, ensuring a more stable operation under OCF conditions. Unlike the PI controller, which causes a total loss of rotational motion, and the PR controller, which introduces moderate oscillations, the PIR controller maintains speed within a narrower range, significantly improving the transient response. Furthermore, the proposed PIR controller eliminates the need for complex fault detection mechanisms, offering a simpler and cost-effective solution for ensuring reliable motor operation under fault conditions.

Despite these advancements, certain limitations remain. This study primarily relies on graphical representations to analyze system performance, without incorporating detailed spectral analysis or THD metrics to quantify harmonic reduction. Additionally, while the proposed controllers are tested under a single-phase OCF condition, further research is needed to evaluate their performance in multi-fault scenarios and short-circuit fault. Computational efficiency in real-time embedded systems also warrants further investigation to optimize implementation for industrial and EV applications. Addressing these challenges in future research will further strengthen the applicability of the proposed controllers, enhancing their robustness and adaptability in critical high-reliability applications.