Monitor medium-voltage switchgear in refineries
When medium-voltage switchgear (MVS) in a refinery fails, it can force a process unit or even an entire refinery to shut down.
When medium-voltage switchgear (MVS) in a refinery fails, it can force a process unit or even an entire refinery to shut down. Costs can climb into the millions of dollars as a result of lost production, environmental issues and litigation arising from injuries or fatalities, and the cost of repairing and/or replacing the damaged equipment.
Fortunately, modern sensing technology makes it possible for a refinery to continuously monitor the health of MVS (FIG. 1) and related equipment, and inform the maintenance department when, or even before, problems arise. This article describes the potential causes and problems of MVS failures, as well as the various types of sensing systems that can be used to detect and predict problems.
Fig. 1. Typical installation of MVS in a refinery.
Continuous monitoring of temperature, partial discharge and humidity in switchgear can uncover small problems before they become large issues that can shut down an entire refinery.
Scope of the problem
Refinery power outages have been in the news over the last few years. In January, a sitewide power outage at ExxonMobil’s refinery in Beaumont, Texas required operators to empty systems by burning product, wasting 365 Mbpd of crude oil. The burning created large flames and plumes of black smoke that drifted over nearby neighborhoods.1
According to the Billings Gazette, the city of Billings, Montana has been pulled into an ongoing lawsuit between ExxonMobil and NorthWestern Energy that was filed by the oil corporation after two power outages disrupted its Billings refinery.2 In January, ExxonMobil sued NorthWestern for refinery outages in 2014 and 2016 that led to excess flaring. ExxonMobil claims the outages cost the company millions of dollars, and it is seeking undisclosed damages in the lawsuit.
In February, a California federal jury found Pacific Gas and Electric (PG&E) negligent and partially to blame for a power outage at Tesoro’s Martinez, California refinery, and ordered PG&E to pay $3.5 MM in damages.3
Power outages at refineries are a serious problem, resulting in millions of dollars in damages, environmental lawsuits, flaring, smoke, fires, lost production, bad public relations and threats to employee safety. While not all of these outages are caused by MVS failures, the impacts of these types of failures can be similar.
Refinery power concerns
Studies by ARC Advisory Group, Hydrocarbon Publishing Co., and Harris and Williams assert that 82% of power outages in refineries can be attributed to random failures. The majority of refinery power systems are more than 25 years old, with many running beyond design life.
Electrical problems accounted for 20% of all refinery disruptions between 2009 and 2013, according to the US Department of Energy. While power failures can be caused by utility outages, snowstorms, hurricanes and other unforeseen events, these studies claim that nearly 20% of refinery power disruptions are the result of electrical power equipment failures. In many cases, these failures are caused by MVS.
A typical refinery power system has generators, generator circuit breakers, transformers, MVS, bus ducts, low-voltage switchgear, motor control centers and other equipment needed to distribute power throughout the refinery. Power substations are located at the sources of incoming power and at various smaller distribution stations throughout the refinery.
Diagnosing and detecting all potential problems in this myriad of equipment (such as generator vibration or transformer oil deterioration) is beyond the scope of this work. Instead, it will examine how to monitor three of the main sources of electrical failures in MVS: overheating of conductors, insulation breakdown and high-moisture environments. Note: While this article concentrates on MVS, much of the discussion also applies to the other electrical assets.
MVS issues
MVS is subject to overheating due to excessive loads, normal wear and tear, and challenging environmental conditions. Left unattended, these conditions can lead to failures that result in costly damage to switchgear and surrounding equipment, power production loss and, in extreme cases, severe injury or death. Common failure modes include excessive temperature, partial discharge and high humidity.
Excessive temperature. Circuit breaker, bus bar and cable connections tend to loosen and/or corrode over time, resulting in thermal failure of the connection and nearby cable insulation.
Fig. 2. Partial discharge and high humidity caused voltage stresses in this switchgear, leading to an explosion.
Partial discharge. As insulators age, weak spots and defects evolve. Under certain load conditions, a dielectric breakdown will initiate across the defect, causing a partial arc between conductors at different potentials. This effect is known as a partial discharge. The breakdown causes a small but sudden rise in current accompanied by a current pulse, as well as electromagnetic (radio or light), acoustic and ozone emissions. Left unattended, this condition can cause the switchgear to explode (FIG. 2).
High humidity. Moisture in switchgear can create shorts or be absorbed by the insulators, leading to insulation breakdown. Humidity also causes metallic corrosion, which can lead to elevated heating, partial discharge, surface tracking and the potential for shorts and flashover.
Manual inspections
Switchgear monitoring has been in practice for some time, and is often carried out through periodic manual inspections while the switchgear is powered down. Such inspections look for obvious problems, such as physical damage, frayed connectors, degraded insulation and evidence of overheated components. Inspectors confirm proper alignment of primary and secondary interlocks, tightness of bolted connections and correct phasing of bus bars.
Electrical measurements can also be conducted while the power is off. Applying voltage with calibrated AC and DC high-potential test sets checks insulation resistance in the panel enclosure, bus bars, circuit breakers and other components. It also assesses contact resistance to confirm bus bar joints are connected properly.
Manual inspections with infrared (IR) equipment can be conducted while the power is on. Periodic IR monitoring techniques require a glass window to be installed in the switchgear, a relatively expensive IR camera and a trained technician. One significant limitation of this type of inspection is that personnel cannot perform monitoring procedures behind bus insulators and cable shrouds because line-of-sight is required with IR technology.
All manual inspections require trained technicians and specialized test equipment, often provided by an outside service company. Depending on the testing service provided (e.g., thermal and/or partial discharge), typical costs for an onsite visit by a service provider could total up to $30,000. Any electrical problems occurring after the inspection can go undetected until the next inspection, which could be a year or more. During that time, small problems can become large ones, potentially leading to a complete failure of the asset(s) and a power shutdown.
A better solution is to employ continuous monitoring of switchgear, which affords refineries the ability to collect data generated during the switchgear’s normal operating conditions, thereby providing awareness to problems in real time.
Real-time trending during full load of electrical stresses, vibration, insulation breakdown and environmental influences provides insight into the health of the switchgear. When performing continuous monitoring, it is not always critical to identify the exact location of degradation, but rather to understand the trend of the defect over time. Monitoring and trending the most common failure modes allows for planned and proactive maintenance, instead of running the MVS to catastrophic failure.
Thermal monitoring
Temperature monitoring is a primary method for detecting switchgear problems, although some facilities employ only partial discharge monitoring. One challenge in implementing continuous temperature monitoring of critical connection points (circuit breaker, bus bar and cable connections) inside MVS is that the sensors must maintain the impulse-withstand voltage, also known as basic impulse level (BIL). Consequently, conductors at different potentials must have a minimal distance between one another to prevent breakdown and ensure the impulse rating. Another key challenge is powering the sensors so as to avoid the requirement for regular maintenance of these devices.
Specific to the BIL concern, Section 5.2 of the Institute of Electrical and Electronics Engineers Standards Association’s C37.20.3 standard, “Metal-enclosed interrupter switchgear,” states that switchgear rated with a maximum voltage of 15 kV must have an impulse voltage of 95 kV, relating to a distance (in air) of approximately 160 mm.
This requirement eliminates the most common types of direct contact temperature-monitoring systems, such as thermocouples and RTDs, because they are susceptible to electromagnetic interference/radio-frequency interference in the high-voltage environment. This leaves only non-invasive systems, such as fiber optics, continuous IR sensing and wireless direct contact sensors, as viable options for switchgear thermal monitoring.
Fiber optic systems. Fiber optic temperature sensors can be routed directly to critical switchgear-monitoring points. The sensors are rigidly attached to hot-spot locations, and are completely immune to electromagnetic interference and noise bursts caused by high-voltage switching. These systems require direct surface contact. They are considered non-invasive, but when operating under the dusty and humid environments often found in refineries, fiber optic cables can create a conductive ground path and compromise the aforementioned withstand voltages.
Continuous IR sensing. IR sensors can detect “hot spots” in switchgear, cables and components. Monitoring switchgear with handheld IR equipment during manual inspections is common, but IR sensors can also be permanently installed inside switchgear for continuous monitoring. The sensors capture a thermal image and then send data back to a control or SCADA system, where specialized thermal mapping software is required to detect problems.
These systems are often expensive and difficult to install, and have several measurement limitations, such as adjacent surfaces having different emissivity or reflections causing false readings. These sensors also cannot measure the bus connections behind bus insulators.
Wireless battery-powered direct-contact sensors. Wireless temperature sensors measure the temperature of strategically important points on metal or insulating surfaces of current-carrying parts. The sensors are attached to components with a high-temperature adhesive. They measure temperatures from 0°C to 150°C and send data to a receiver, which then transmits the data to a control system.
Fig. 3. Three SAW temperature sensors (orange sensors in middle) mounted in MVS.
At a first glance, these systems seem to provide an adequate solution. Unfortunately, the batteries have a limited life span that is usually reduced in the high-temperature environments found inside MVS. The associated cost of taking the switchgear out of service to replace batteries is untenable from an operational perspective, and also elevates the actual cost of system ownership to an unacceptable level.
Wireless passive direct-contact sensors. Wireless passive sensor systems provide real-time continuous monitoring via direct connection to critical measurement points. These systems are easy to install, require no maintenance and have a life expectancy comparable to the switchgear itself. These sensors employ surface acoustic wave (SAW) technology (FIG. 3).
SAW technology utilizes a piezoelectric substrate (a material that changes electrical charges due to mechanical stresses), an interdigitated transducer (IDT) resonator and an antenna. One of the most common piezoelectric materials used is quartz crystal, onto which a metalized IDT is fabricated. SAW transducers, when activated by a radio frequency (RF) wave transmitted by a system control device, reflect back a surface RF wave that changes frequency linearly with temperature.
Compared to other sensors, SAW temperature sensors have longer reading distances. They also do not require line-of-sight. These characteristics maximize basic impulse levels, while using lower transmission power to minimize control system electromagnetic compatibility issues.
Partial discharge monitoring
The most commonly used partial discharge detection instruments directly measure the current and voltage spikes in high-frequency current transformers or high-voltage capacitive couplers, as outlined in the International Electrotechnical Commission’s IEC 60270 standard, “Partial discharge measurements.”
This method has several strengths, including the ability to analyze pulse shapes and assemble a graph of the discharge events relative to the phase of the power line waveform. These systems are very expensive and require trained technicians to analyze the data, and do not lend themselves to permanent, continuous monitoring installations in switchgear.
Utilities are evaluating additional partial discharge detection methods with IEC 62478, a prospective standard for acoustic and electromagnetic PD measurements. These methods use instruments to make indirect analytical measurements and obtain a relative signature of partial discharge pulses that testers can use for system trending. A number of testing methods exist, as outlined in the following sections.
Transient earth voltage (TEV). Refinery or outside service personnel use TEV test equipment to check switchgear externally by measuring the electromagnetic emissions conducted to ground. Although testers can use TEVs for continuous monitoring, these voltages cannot always monitor faults between phases.
High frequency (HF) and very-high frequency (VHF). HF/VHF systems operate between 3 MHz and 300 MHz and use large antennas, high-frequency current transformers or coupled sensors to detect partial discharge spikes. Large antennas are not compatible with measurements inside switchgear, and direct-coupled sensors impair safety.
Fig. 4. A monitoring device installed in a safe position (black box on inside panel) within an MVS compartment monitors partial discharge.
Acoustics. Testers use a microphone or similar acoustic sensor to monitor frequencies between 10 Hz and 300 kHz. Refinery personnel can use this wireless method for continuous monitoring, but it has a limited detection range due to sound damping within dielectric objects.
Ultra-high frequency (UHF). This broadband (300 MHz–3 GHz) detection method monitors transient electromagnetic waves through an antenna. Traditional UHF methods are susceptible to noise from cell phones, radios and other transmitters. However, newer instruments use selective, banded UHF monitoring to detect partial discharges while rejecting noise sources. UHF provides the safest, most non-intrusive continuous partial discharge monitoring system (FIG. 4).
UHF partial discharge detection
Effective UHF partial discharge detection for continuous monitoring requires the distillation of an overwhelming amount of complex data down to a concise piece of information, all without the intervention of a highly trained operator.
Band-pass filtered UHF partial discharge detection methods are capable of avoiding strong interfering signals at close proximity—while advanced digital filtering methods are capable of detecting the presence of partial discharge—even with residual, unfiltered noise.
To provide an autonomous approach for PD monitoring, advanced system algorithms must be implemented to present data that can be easily processed for health assessment and long-term system trending. UHF emissions signals can be broken into three major categories: noise, and asymmetric and symmetric discharges.
Noise. Noise denotes UHF energy in the selected frequency band(s) not in close correlation with the power line frequency. External radio interference is a reliable classification of noise; however, weak and erratic partial discharge, which occurs early in the evolution of a defect, is also a noise classification.
Fig. 5. Continuous trending of partial discharge events captures periodic spikes due to condensing humidity that would not have been recognized with non-continuous detection methods.
Asymmetric discharge. Events occur primarily on the negative half cycle of the power waveform, where electrons emitted from the metal ionize the air. Asymmetric discharge (also called corona or surface discharge) presents significant results at both odd and even harmonics of the power line frequency.
Symmetric discharge. Discharge events that occur in the bulk of the material, often referred to as internal or symmetric partial discharge, happen at the positive and negative polarity portions of the power cycle and are often represented as even harmonics of the power line frequency.
One of the causes of partial discharge is excessive humidity, which condenses onto cables and connectors. Continuous UHF monitoring can detect these voltage spikes (FIG. 5).
Continuous monitoring
Complete switchgear continuous-monitoring systems are available with temperature, humidity and partial discharge sensing capabilities. In addition to the wireless temperature and partial discharge sensors outlined above, wired sensors are often installed on the chassis of the switchgear cabinet to provide ambient temperature and humidity readings. Ambient temperature readings are important because the critical issue is temperature rise of hot spots above ambient, as opposed to absolute temperature.
Fig. 6. Switchgear monitoring system with temperature, humidity and partial discharge sensors.
A typical continuous-monitoring system for MVS would include a monitoring unit connecting temperature, humidity and partial discharge sensors, as shown in FIG. 6.
Each temperature and partial discharge air interface device shown in FIG. 6 uses banded UHF technology to sense partial discharge directly. Each air interface device can also wirelessly link to three or more SAW temperature sensors.
Up to four air interfaces can be wired to the monitoring unit via low-loss coaxial cables. The monitoring unit can also accept up to eight conventional wired humidity or ambient temperature sensors, which are well suited for taking measurements of these variables in bus ducts.
The monitoring unit can be a full-featured human-machine interface (HMI) with monitoring capabilities, or a reader that provides remote monitoring. The reader provides all the necessary wireless interrogation signals for the SAW sensors through the air interface device, internally implements the partial discharge detection algorithms and communicates directly with the humidity and ambient temperature sensors. All reader data is accessible through industry-standard Modbus remote terminal unit (RTU) (RS485) digital communication protocols, affording ease of integration into a plant’s existing supervisory control and data acquisition system, distributed control system or historian.
The HMI unit provides the same monitoring capabilities as a reader, but adds functionalities including a local touchscreen for real-time display, data storage, alarming and multi-unit functionality. Multi-unit functionality provides an input device port for connecting up to seven readers via the Modbus RTU protocol, as well as extending the HMI display, alarm and communication capabilities to all connected readers. This option allows for easy integration into switchgear lineups. The HMI unit can be connected to an existing plant SCADA system or historian via Modbus transmission control protocol (TCP), distributed network protocol 3 (DNP3) or IEC-61850 digital communications.
Takeaway
Continuous condition-based monitoring systems provide clear, early detection of evolving issues with MVS, thereby helping avoid unplanned downtime and outages. Data provided by a condition-based monitoring system facilitates historical trending capabilities, as well as the ability to detect failures before they occur, giving refinery operators the information they need to perform proactive maintenance.
A continuous thermal and partial discharge monitoring system that focuses on main switchgear failure modes will help lower maintenance costs, reduce downtime, provide necessary early warning information and ensure the highest level of operational safety. HP
LITERATURE CITED
- Kang, Y. P., “Jury dings PG&E $3.5M for Tesoro refinery outage,” Law360, February 2, 2016, online: http://www.law360.com/articles/754290/jury-dings-pg-e-3-5m-for-tesoro-refinery-outage/
- Besson, E. and B. Scott, “Smoke at Exxon refinery in Beaumont due to power outage,” FuelFix, January 21, 2016, online: http://fuelfix.com/blog/2016/01/21/flames-smoke-seen-at-exxon-refinery-in-beaumont/
- Hudson, M., “NorthWestern: Fault began with city employees in Exxon refinery power outages in Billings,” Billings Gazette, April 5, 2016, online: http://billingsgazette.com/news/local/northwestern-fault-began-with-city-employees-in-exxon-refinery-power/article_c30dd4ca-0dc4-5ba9-b809-2f9a6fd83bd0.html
The Author
Murray, J. - Emerson Process Management, Boston, Massachusetts
Jonathan Murray is director of IntelliSAW products at Emerson, focusing on new business strategies, product development and marketing of the IntelliSAW Critical Asset Monitoring (CAM) Platform. Prior to joining IntelliSAW, Mr. Murray focused on automation and control systems for smart grid, energy storage and electric vehicles, holding positions such as business unit manager, business development manager, project manager and senior engineer. He holds a BS degree in electrical engineering from the University of Massachusetts.
Related Articles
From the Archive