Shining a light on filters

Particulate build-up in gas turbines is potentially a very expensive problem for power generators. Cameron Stathers of ETR-Unidata introduces a novel and superior approach to filter monitoring

In any gas turbine power plant, poor inlet air filtration results in contaminated inlet air. In turn, this will cause fouling, corrosion and erosion of the turbine compressor blades, reducing the efficiency of the plant and creating additional service requirements. In extreme situations, such as the Middle East, where reserved pulse-air jet cleaning is deployed to manage high levels of ambient dust, pulsing can cause individual filters to be dislodged. In this situation unless a frequent visual inspection of the inlet filter house is undertaken then damaged or dislodged filters will result in significant levels of dust entering the inlet filter system and ultimately cause turbine blade damage.

While the overall performance of filter banks can be monitored, determining how the individual filters within a filter bank are performing has, to date, not been possible. Identification of the precise cause of any degradation requires an offline inspection of each individual filter. Currently the only instrument used is a pressure drop monitor, which simply enables alarm conditions to be identified and provides no information on which filter has failed.

The new Epsilon System from ETR-Unidata is able to monitor sub-ambient particle concentrations at the levels associated with inlet air downstream of a filter bank, and does so over the entire filter bank. The Epsilon’s ground-breaking approach provides a high degree of spatial resolution, or particle mapping, as well as quantifying how much dust is getting through the filter bank and can identify which section of the inlet filter is failing.

Epsilon background information

Gas turbine power plant operators seek to minimise the amount of dust passing (or hitting) the turbine blades by installing large, high-performance filter banks on the plant inlet side.

There are two reasons for this:
• Without filter banks, dust particles would travel through the turbine, colliding with the compressor blades at very high velocity, eroding and shortening the life of the compressor blades. In extreme cases, catastrophic breakdown of the compressor blades can occur.
• While many factors determine the day-to-day operating efficiency of a gas turbine plant, all experience a gradual decline in their efficiency over time caused by the build-up of particulate matter on the compressor blades. The efficiency drop means that the plant has to work harder to produce the same power output, consuming more fuel. Filter banks slow the particulate build-up and the associated efficiency drop.

Figure 1: Typical Epsilon laser and single detector configuration.

The inlet duct on a typical large gas turbine is commonly nearly 350m² in area and contains over 700 pairs of filters. If a potential ingress of dust is identified, for example by pressure differential instruments, each filter in the filter bank must be inspected to locate the problem, a time-consuming task.

Worse, while the visual inspection will normally identify the location of a “catastrophic” failure of a filter, it may well miss the simple degradation which occurs over time and which precedes the “catastrophic” failure. Inlet filters operate at varying degrees of efficiency, gradually deteriorating over time at a rate which is not always linear. A visual inspection off-line cannot provide as good a measure of the condition of individual filters, nor be as predictive of their future performance, as good continuous online monitoring for particles actually breaching the filter bank.

The costs associated with unrecognised particulate presence in the inlet air are very significant. The cost is not only that of the filters replaced earlier than necessary, nor that of any failed turbine blades, nor indeed that of the extra fuel consumed as the plant efficiency declines, but also the opportunity cost of extended downtime of the turbine, if a failure occurs.

The result? Operators build an insurance policy into their maintenance scheduling, replace filters earlier than necessary, and probably have more and longer outages than would be necessary if a reliable means to monitor the performance of the individual filters and the build-up of particulates on the compressor blades was available.

Extending filter life and an improved ability to monitor plant efficiency results in lower CO₂ emissions and reduced costs.

Figure 2: Inlet duct Optical Window with
blanking cover removed.

Several years ago a major filter manufacturer approached ETR-Unidata seeking to discuss this situation. ETR-Unidata, with its 20 years of experience, reviewed the current technology and realised that a new and radical approach was necessary to solve the problems and provide the gas turbine industry with the non-intrusive monitoring solution it required.

To provide a spatial map of the duct it was clear that a new type of imaging and digital processing system was required. Various new approaches were considered, including electrostatic techniques and microwaves. Eventually, the design settled on a laser scanning system paired with a camera-based digital imaging system.

In combining these two technologies, ETR-Unidata developed a breakthough instrument which is fully covered by worldwide patents. The Epsilon became the first product able to perform real time particulate monitoring at sub ambient levels and provides:
• Full duct coverage even in very large ducts such as the GE Frame 9A.
• Spatial information on particulate distribution within the duct.

Epsilon technology and system configuration

The Epsilon system has four components:
1. A laser – the power output depends on the minimum size of particle to be monitored.
2. A computer-controlled projection system comprised of a small-angled mirror and a precision stepper motor.
3. A number of digital cameras (depending on the geometric and physical constraints of the inlet duct to be monitored).
4. Computer hardware and software for data analysis and user graphical interface.
Importantly, none of the components are intrusive, and the system therefore poses no threat to the integrity of the plant.

Figure 3: An illustration of the multiple image
combination technique to remove unwanted reflected
background noise within the duct.

To monitor in-filter performance, the Epsilon system is located as close as possible to, and on the clean side of the inlet filters and mounted on the outside of the inlet duct wall. The laser is projected into the duct through the wall via a small critically-located optical window. The digital imaging is similarly achieved by the installation of windows in a slightly different position and no hardware is located within the duct itself. The key data required for filter performance measurement is the trend in the quantitative count of particles, and their spatial distribution.

Figure 4: Isolated scattered particulate centres using
Hough Transformation technique to detect those particles
that have been illuminated by the laser beam.

The exact location of all the optical windows is installation-dependent, as each gas turbine plant generally has its own unique physical geometry. The windows are normally installed during an outage period. System set-up during an outage is preferred but is not critical.

No air purging system is required to keep the windows clean and free from dust or particulate build-up, because the windows are located on the clean side of the inlet filter. Even in the event of dust build-up, no interference with the monitoring would occur as the digital camera is focused well beyond the window, into the duct.

Full inlet duct coverage is provided by projecting the laser onto a small, angled mirror which is controlled by a stepper motor. As the mirror is rotated through its range the angle at which the laser beam is reflected changes, and the system sequentially projects a fan of light in a single plain across the whole duct. As the dust particles pass through the laser beam, its intense light is scattered. This refracted light is detected by the digital camera receiver unit, creating 20 images of particulate presence per second. Imaging software then integrates the data from up to 18,000 images per full duct scan to:
• Determine the particle count at thousands of points across the duct.
• Count individual particles (generating extremely accurate continuous read-outs.
• Map the particles for proximity/distribution profiling.

Epsilon image processing

To create a meaningful image of particulate distribution from the captured images a number of steps are required:
1. As wide-angled lenses are used on the cameras, the image has to be corrected for fish eye distortion to create the correct rectangular shape of the duct.
2. Each individual laser scan has to be located within the image.
3. Random and background noise has to be removed from each frame to ensure that only the light scattered by the individual particles is “counted”.

Fish-eye distortion is removed during set-up, when the field of view of each camera is corrected by the software to ensure that the image mapped is rectangular in shape. This adjustment is similar to, but a much more sophisticated application of, the technique that is used to remove the keystone effect in LCD projectors. It only has to be made once for each camera used in the installation.

Unwanted random background noise from the image, such as reflected light from duct walls or internal framework, is removed by the advanced digital processing software. Multiple high-speed images of each laser position are taken. As the noise is random in nature, it can be cancelled out by combining the individual images. This also increases the signal (light) from the particulate, which is random in nature.

After the background noise has been removed, a technique known as a “Hough Transformation” is used to detect the light generated by individual particles as they interact with the laser. This signal algorithm relies on the parameterisation of the centres of the laser-scattered light.

The information extracted from the digital image is plotted and the Hough Transformation algorithm applied to determine whether the scattered centres are located on a straight line or not (see Figure 4). The technique distinguishes between random background scatter and scatter (on a straight line) caused by the laser. This sequence is repeated for each position of the laser to build-up an image of the particulate distribution within the duct itself.

Epsilon user interface

Epsilon’s graphical interface is located in the control room and consists of a high resolution grid display (see Figure 5) that relates to the inlet filters. If a filter breakthrough happens, an individual cell would change colour in a traffic light type approach as an alarm condition occurs. The operator would then be aware which filter or group of filters had started to fail, allowing appropriate action to be taken and eliminating the drawn-out process of investigating each filter individually.

Figure 5: Epsilon user display with real-time
particle distribution map.

The user interface also has a zoom facility which enables the operator to focus in more closely on the area where the breakthrough has happened and identify more precisely the location of the problem. The magnitude of the particulate breakthrough can then be assessed and the appropriate actions taken.

With the Epsilon system, operators are able to base their decisions on filter condition using real-time information on actual particulate levels within the duct itself and without having to rely on an indicative and less informative pressure drop measurement.

Epsilon user benefits

Epsilon’s ability to locate and focus on areas of filter deterioration and provide information on the on-going particulate level in the duct creates the potential for huge savings in plant maintenance through reduced downtime and better filter management. As the Epsilon system is extremely sensitive it will quickly identify and locate any potential filter or seal failure and hence prevent significant levels of particulate from entering the inlet duct and damaging the turbine blades.

It has also been suggested that for operators who “wash” their compressor blades online to manage the rate of particulate build-up, the system offers a much improved ability to determine when the cost of washing falls below the costs associated with the efficiency drop, allowing the plant to run, over time, at higher levels of efficiency.

For operators generally the Epsilon offers an improved ability to monitor the efficiency drop associated with particulate build-up on the compressor blades and better optimise their maintenance cycles, especially if “washing” the compressor blades is a major determinant in the outage cycle. Again, if the plant can run at a higher level of efficiency on average over time, substantial fuel savings will result and the plant’s carbon emissions will be reduced, which given the move towards carbon trading could also reduce expenses.

Cameron Stathers, ETR-Unidata’s sales manager, can be contacted at:
cs@etr-unidata.com

A 3D animation of the Epsilon System can be seen at:
http://www.etr-unidata.com