Despite the large push for green energy currently being touted, the demand for oil and gas continues to increase because of its availability here and now. Natural gas in particular is currently seen as the bridge to green energy, allowing us to maintain economic expansion while renewable markets come of age.
To keep up with this growing petroleum demand there is a constant need for continued exploration and for pushing existing reserves to further production. Both increase bottom line costs, causing a similar increase in oil and gas prices, and a decrease in profit. Reducing operating and maintenance (O&M) costs can be a contributing factor to maintaining or even reducing investment costs, allowing for increased profit instead. With low-cost, online condition monitoring systems that predict failures and maintenance requirements, you can better forecast and schedule maintenance activities to lower these O&M costs.
The Need for Condition Monitoring
In the oil and gas industry, production downtime can equate to hundreds of thousands of lost dollars per day. Ensuring that production machinery and other critical equipment downtime are minimized not only reduces operating and maintenance costs but also increases production revenue.
There are three main approaches to asset equipment monitoring: reactive maintenance, scheduled maintenance, and predictive maintenance. Using a reactive maintenance strategy, also called run-to-failure, an asset is run until it no longer functions and is then repaired or replaced. In addition to the increased cost of equipment replacement, this approach can often result in a catastrophic machine failure, causing harm to related assets or personnel. This is similar to not repairing the brakes on your car, you may save money in the short run, but if they stop working while driving down a steep hill is it really worth it?
A scheduled maintenance approach calls for repairs and maintenance to be performed at regularly scheduled intervals, usually dictated by the manufacturer. This philosophy is the most expensive to follow because components could be overhauled that have no need for it. Additionally, because recommended repair cycles are based on averages and statistics, an asset can unpredictably fail between repairs. Going back to our car analogy, a scheduled maintenance approach is similar to changing your oil every 3000 miles. The oil gets changed because that's what you were told to do, whether or not it is actually contaminated.
In a predictive maintenance strategy, assets are monitored at regular intervals for signs of degradation. This is where condition monitoring comes in, as part of a predictive maintenance approach. From these regularly monitored parameters a complete picture of the machine can be obtained as well as historical trends to predict future behavior. Maintenance can then be most effectively scheduled for known downtimes before a failure actually occurs. This would be similar to the check engine light on your car. Various parameters are constantly being measured and assessed over time by an onboard computer. If one of them trends too far a warning is given, signaling that repair is needed. You can then schedule a repair time when convenient for you (within reason), rather than instantly having your engine seize up.
Figure 1. Predictive maintenance allow machines to run a maximum amount of time, performing repairs when the machine is on the cusp of its acceptable operating levels, eliminating unnecessary repairs while keeping the machine running.
The Electrical Power Research Institute has calculated comparative maintenance costs in dollars per horsepower (HP) for each maintenance philosophy. They determined that a scheduled maintenance strategy is the most expensive to run at $24.00 per HP. A reactive maintenance strategy is second most costly at $17.00 per HP but has the additional costs of safety being compromised. A predictive maintenance strategy is the most cost effective, coming in at only $9.00 per HP, and all but eliminating the risks of secondary damage from catastrophic failures.
Predictive maintenance clearly presents a significant cost savings over the other two techniques, but only for large, critical machinery. It certainly wouldn't make sense to run a preventive maintenance program for light bulbs in your home - the cost of the program would outweigh the maintenance savings. However, in the oil and gas industry, where assets are typically in the millions of dollars each, a preventive maintenance strategy can amount to huge cost savings.
By taking a look at the unique maintenance challenges that are presented by the oil and gas industry, the case for a predictive maintenance strategy is made even stronger.
Oil and gas assets, specifically upstream, are often spread about in difficult to reach or dangerous locations, making payment to a repair technician to travel to them expensive, unfeasible, or even impossible. Petroleum also has one of the highest plant shutdown costs of any industry. If a critical asset fails unexpectedly, restart time can be up to several weeks. With regulatory safety checks and weeks of lost production revenue, unexpected shutdowns can cost millions of dollars.
Both of these factors mean that an unexpected failure translates into huge cost overruns in either lost production or the cost of technicians, transportation, and replacement parts. With condition monitoring and a predictive maintenance plan both of these costs can be minimized by reducing unnecessary repairs and scheduling efficient maintenance at the most opportune (least costly) times, usually during an already planned shutdown.
The Benefits of Distributed Monitoring
A predictive maintenance approach can come in two different varieties: route based and distributed monitoring. A route-based approach relies on a technician who regularly visits assets and records their condition with a hand held device. With this system you can conduct non-routine tests on the fly and gather anecdotal information that may not otherwise have been available. A distributed monitoring system on the other hand, uses a remote data acquisition device to collect physical signals from permanently mounted sensors and transmits them back to a site data server for analysis at a central location.
In the oil and gas industry, a distributed online monitoring system often makes the most sense because of how spread out assets can be. With a distributed online monitoring strategy, a central office can monitor all assets across the world and deploy timely maintenance teams when necessary. This has the added advantage of centralizing expertise. Rather than employing a number of monitoring and maintenance experts across the globe, a smaller number can be brought on in one location.
Figure 2. A distributed monitoring system allows the monitoring of remote assets from a local workstation. Data can be transferred from remote locations to a central data server for further analysis.
Emerging Trends in Oil and Gas Condition Monitoring
Because of the high cost of downtime in the oil and gas industry, it is often at the forefront of condition monitoring technology. New innovations are constantly being introduced in both hardware and software that can benefit condition monitoring, allowing faults to be detected sooner and a machine's future state to be better predicted.
As wireless technology continues to advance, the cost of implementing it becomes cheaper and cheaper. With such low implementation costs, route based monitoring is increasingly being replaced with low cost Wireless Sensor Nodes. These nodes can perform basic measurements and then transmit the data wirelessly within a defined space such as a plant, reducing the need for a walk-around technician. These nodes will often transmit both operating parameters of the machine, such as vibration RMS levels, but will also periodically send time waveforms for storage and further analysis. Though there are still advantages to having a person at the machine, the cost effectiveness of such nodes is making them an increasingly attractive option.
Traditional condition monitoring systems largely rely on accelerometers and vibration data for extracting machine parameters. Vibration is slowly being augmented however by stress wave analysis, which is the newest trend in condition monitoring hardware. Developed by SWANtech out of Florida, stress wave analysis works by detecting ultrasonic stress waves produced by tiny defects in machine components. Using this technique you can detect faults way before any significant vibration change is detected by more traditional means. The sensors used are similar to accelerometers but measure much higher frequency components and filter out the lower-frequency vibration components. This allows the stress waves to be seen through the noise of standard machine vibration where they otherwise would go unnoticed.
In software, large strides are being made in smart technology to create algorithms that can better predict the future state of a machine based on its current condition. Using advanced techniques such as neural network mapping and statistical regression, you can compare a machine's current state to either empirical data or known models to compute the most likely future causes of failure and how long until those occur.
Control System Integration
Thus far we have discussed how a condition monitoring system can calculate critical parameters and report these back for further analysis and prediction of future failures, but such a system can do much more than that. By integrating your condition monitoring system with the control system of the machine being monitored you can help them run more efficiently and safely.
Gas turbines for instance often use dynamic pressure sensors on their cylinders to monitor efficient firing. When the frequency or peak level of pressure in the cylinders begins to shift the feedback to the control system this can help alter the fuel mixture to retune the machine to its optimum performance.
Additionally, feedback to the control system can help avoid catastrophic failures. If the condition monitoring system detects parameters outside of safe operating conditions it can signal a machine shutdown, avoiding further damage to the machine. A system like this usually depends on redundancy being built into the monitoring system, as you would not want an errant sensor reading to unexpectedly shutdown your machine. Often a voting system is used, so if two out of three sensors were indicating a machine shutdown the command would then be sent.
The control system for the machines being monitoring is generally part of a larger plant SCADA or DSC system. The condition monitoring system can communicate with these systems through a variety of protocols. Most recently OLE for Process Control (OPC) is being implemented in facilities, but Modbus, Profibus and other low bandwidth protocols still widely remain in use.
Taking it to the Real World
Supreme Electrical Services is a great example of a successful implementation of an oil and gas condition monitoring system. Providing electrical and monitoring services for the oil industry, Supreme Electrical set out to design a condition monitoring system for high pressure fracturing pumps that could survive the often-harsh environments found in the oil field.
Each fracturing unit traditionally comes equipped with a control system and sensors for monitoring basic parameters such as discharge pressure, RPM, and oil pressure and temperature. The designed system uses the National Instruments Singleboard-RIO and LabVIEW to interface with this control system and read in these parameters through the SAE J 1939 communication protocol as well as incorporate additional instrumentation. All of this data is then transmitted back to a central control console for further analysis.
Once the data is back to the central server it is checked to see if it is within the normal operating parameters of the system. With this information, operators are better equipped to determine if they should continue or discontinue operation of the fracturing pump.
Figure 3. Using National Instruments RIO architectures with LabVIEW, Supreme Electrical Services produced a system to monitor hydraulic fracturing pumps.
A condition monitoring system as part of a predictive maintenance strategy can help to reduce operating and maintenance costs. In the increasingly competitive and safety conscious world of the oil and gas industry implementing such a system can free up money for investment in further exploration activities to keep production at pace with market demand.
About the Author
Doug Farrell is the product manager for machine condition monitoring solutions at National Instruments. He began his career at National Instruments as an application engineer and was a founding member of Waterloo Labs, a group of engineers at NI who specialize in DIY projects. He holds a bachelor's degree in Mechanical Engineering from Georgia Tech.