Index: 1. Application Examples 1.1 Jet Engine 1.2 Wind Tunnel 1.3 Ground Vibration Testing 2. Test Preparation: Setup, Transducers, and Cabling 2.1 Documentation 2.2 Calibration Checks 2.3 Charge Transducers 2.4 ICP Transducers 2.5 Strain Gauges 2.6 Data Acquisition Hardware 3. Test Measurement: Recording, Monitoring, and Alarming 3.1 File Sizes 3.2 Configuration 3.3 Redundancy 3.4 Monitoring and Analysis 3.5 Alarming 4. Test Reporting: Processing and Reports 4.1 Data Selection 4.2 Processing 4.3 Reports 4.4 Active Pictures
1. Overview of Applications
Certain products require days of testing in specialized facilities before they can be commissioned and released. A few examples of these types are tests are explained in this section.
1.1 Jet Engine
Part of the commissioning of a new jet engine requires measuring dynamic pressures, vibrations, and strains. This type of test is done under multiple conditions over many days in specialized facilities as shown in Figure 2.
Figure 2: A jet engine (right side, white) being tested for dynamic performance. It is hung from a pylon (grey) which provides fuel, commands, and contains thousands of channels of instrumentation.
During this commissioning test, the information from the dynamic strain, pressure, and vibration sensors are converted from analog signals to digital signals as outlined in the diagram in Figure 3.
Figure 3: Configuration of a high channel count, dynamic data acquisition system for jet engine (top left, white) commissioning. Analog data (blue) from dynamic transducers are converted into digital data (black) and stored and archived (top, right). Engineering groups can monitor the test in progress from remote locations (bottom, right).
This digital data is handled by computers in a network configuration. Sample rates exceed 100,000 samples per second per channel, and typically hundreds of channels are collected.
Multiple computers are often used to distribute the workload. Computers are paired with analog to digital converters (for example the Simcenter SCADAS) to acquire the dynamic data. Other computers are used for dedicated tasks like allowing engineering groups (in different physical locations) to monitor the test in progress. Additional computers can be used to issue warnings if critical vibration limits are hit.
All of this is typically controlled from a master computer (and operator) that overlooks the entire process from a single screen as shown in Figure 4.
Figure 4: In Simcenter Testlab Turbine Testing, the control system master computer screen shows the status of all systems being used for dynamic data recording.
General purpose computers, with separate analog signal conditioning and digitalization (i.e., Simcenter SCADAS), have several advantages for this type of applications. In the past, specialized integrated min-computer/signal conditioners were used. When computer performance improved, these systems were not modular – the computer and signal conditioning hardware were not separate. Upgrading required both the signal conditioners and corresponding mini-computer to be decommissioned together.
1.2 Aero-Acoustic Wind Tunnel Array
Aero-acoustic wind tunnels are used to measure the wind noise performance of passenger vehicles. Occupants of today’s vehicles expect quiet interiors, and with the coming of electric vehicles (which eliminated noise from combustion engines) the wind noise is a critical element in the perceived comfort of the vehicle.
In order to pinpoint where the wind noise originates, the wind tunnels are often outfitted with an array of acoustic microphones on the top and sides of the vehicle as shown in Figure 5.
Figure 5: An array of microphones (white frame on right, black dots are microphones) are used to determine where wind noise is generated on a passenger vehicle (left).
Each array consists of hundreds of microphones that are measured up to 50 kHz sampling rate each. The goal is to triangulate from the measured sound waves at the array back to the vehicle surfaces to determine the wind induced noise levels.
Each hour of wind tunnel operation is carefully tracked. The data must be processed immediately and viewed to determine if it is acceptable to move to the next operating point. In a wind tunnel application, colors are superimposed over an image of the vehicle to show where the highest noise levels are located as shown in Figure 6.
Figure 6: After each measurement in the aeroacoustic wind tunnel, the operator is shown an image of the sound levels overlaid on a picture of the vehicle which indicates the noise levels.
Before a new aircraft is permitted to fly, a Ground Vibration Test is performed to measure the dynamics of the airframe. The natural frequencies and mode shapes are checked against analytical models to make sure they are as expected before the first flight can take place.
These tests need to take place in a short time frame. The first flight worthy aircraft of a development program cannot be left idled for long.
Hundreds of accelerometer channels must be acquired and analyzed in a short time frame. An example test setup is shown in Figure 7.
Figure 7: Ground vibration test on a F-16 fighter aircraft. Not only is the first aircraft tested, any variants are during the life of the aircraft are also checked.
During a Ground Vibration Test, a Design Engineering Representative (DER) will decide if the aircraft meets the requirements for first flight. The representative needs continuous access to the data to make important decisions and sign off on test results.
Preparation and planning for a test often takes longer than the actual execution. To have a successful test, instrumentation must be installed and properly configured. All critical aspects of running the test must be documented and planned properly. Some aspects of test preparation are documented in the following sections.
Ideally, all information relevant to a test will be documented, and if possible, stored in a single central location.
Documentation can cover a lot of areas: photos of transducer installations, test request forms, procedure documents, test conditions, etc. An example of the Simcenter Testlab documentation worksheet is shown in Figure 8.
Figure 8: The Simcenter Testlab Documentation worksheet can store relevant test information like photos, Excel files, and Word documents.
Documentation is very important, especially during the processing and analysis of the data post testing. Questions can arise about the exact test conditions, how a transducer was installed, or what settings were used. Only with proper documentation can these questions be answered.
Making sure the transducers are installed and working properly before the test starts is critical. Using calibrators, the sensors can be checked for the correct output. Not only does calibration check for correct installation and sensitivity values, but wiring problems and other issues can also be discovered. Precheck options vary depending on the type of transducer. For example, an ICP fault (damaged wire or sensor) can be found by checking the current draw on individual data channels.
For strain gauges, a shunt check with a known resistance can be performed as shown in Figure 9.
Figure 9: Shunt calibration of a strain gauge also checks for proper operation and installation.
When using long cables in a high vibration environment, it is important that they are tied down properly to minimize their movement. Some transducers, like charge-based accelerometers, can generate local changes in capacitance due to cable movement alone. This might show as large spikes in the vibration data that are not real. This is called the triboelectric effect but is also sometimes referred to as “cable whip”.
To avoid this problem, tying down the cable in a way that it cannot move excessively is helpful. This is illustrated in Figure 10.
Figure 10: Top – Cable is tied to reduce motion, Bottom – Cable vibrates freely increasing the chance of spurious vibration readings.
Not only should the motion of the cable be minimized, it is also important to eliminate relative motion between the cable and accelerometer.
2.2 ICP Transducers: Cable Length and Capacitance Affects
Long cables can act as low pass filters on data signals that come from Integrated Circuit-Piezoelectric (ICP) transducers (also called IEPE). The longer the cable, the higher the chance that frequency content in the signal is reduced or eliminated.
Given the length of the cable, the cable capacitance, the voltage level of the signal, and the current from the signal conditioner, the frequency content of the signal can be determined using the nomograph in Figure 11.
Figure 11: Nomograph of cable length capacitance effect on frequency range.
The frequency content and signal level determine how much the data is affected. For example, it is possible that a certain voltage level in a signal can pass at low frequencies but would be attenuated at high frequencies. Increasing the supply current to the ICP transducer can help alleviate this issue.
2.3 Strain Gauges: Electromagnetic Interference and Voltage Supply
Strain gauges are used to measure small deformations in a test object. The strains result in low level voltages signals. Low level voltage signals are more susceptible to interference from external electrical noise.
The strain gauge wiring can act like an antenna, picking up electrical noise from the surrounding environment. The longer a cable, the more likely it is to pick up electrical interference as shown in Figure 12.
Figure 12: Electrostatic interference is more pronounced with longer signal wires.
In addition to electrical interference, long strain gauge wires also affect the power supplied to the gauge.
Because long wires have more resistance, the actual supplied voltage that reaches the strain gauge can differ from what was intended. To ensure the proper voltage is supplied, data acquisition hardware like the Simcenter SCADAS support sense lines. Sense lines are used to measure the supply voltage at the gauge itself and then adjust the supply accordingly.
Transducers are plugged into data acquisition hardware. Data acquisition hardware provides signal conditioning for the transducers and performs conversion from the analog signals to digital data.
Some potentially helpful aspects of data acquisition hardware are as follows:
Separate Computer: To support upgrades over time, the computer and data acquisition hardware should be separate.
Modular: Hardware system should allow individual components to be swapped in and out. For example, if Ethernet is used as the interface to the control computer, ideally the interface component should be modular. This would allow future interfaces that are faster in speed (for example: a future USB 5 versus today’s ethernet) to be swapped in independently of the rest of the hardware to facilitate easy upgrades.
Flexible Signal Conditioning: Transducers of any type (strain, pressure, accelerometers, etc.) should be able to be conditioned by the hardware.
True Synchronization between Channels and Frames: Data from different channels and different acquisition units should be able to be truly synchronized. In other words, all channels should be sampled simultaneously. This is especially important with dynamic phenomenon where the phase relationship between different locations on the test structure need to be well understood. Hardware that does multiplexing or “muxing” – taking samples sequentially between different channels – should be avoided. Data acquired via a multiplexing process is not synchronized.>
Anti-aliasing Protection: An anti-aliasing filter should be used when gathering the data to avoid misidentification of frequency content.
The Simcenter SCADAS data acquisition hardware, which meets these requirements, is shown in Figure 13.
Figure 13: Simcenter SCADAS hardware is modular. Cards can be mixed and matched to support any application.
The Simcenter SCADAS hardware allows cards to be mixed and matched, and modular components to be replaced.
3. Test Measurement: File Sizes, Recording, Monitoring, and Alarming
When performing a commissioning test as described above, it is usually not enough to simply press “Go” on the measurement system and then wait until the test is done. During the test, other considerations include: • Disk space is rapidly filled up due to the high volumes of data being acquired. It needs to be managed. • Because this test can last days, various stakeholders need to look at the incoming data in real-time to make sure everything is as expected. • Data may need to be analyzed while the test is still in progress. It is not possible to wait until the test is finished before analysis is started.
These considerations are explained in the next sections.
3.1 File Sizes and Data Management
The data file size generated during a test acquisition is shown in Equation 1.
Equation 1: The size of a data file is dependent on sampling rate, time duration, and number of channels.
For an acquisition of:
100,000 samples per second
60 minutes duration
The file size is 108,000,000,000 bytes. This is 108 Gigabytes for one hour of data collection. It is important to make sure that enough storage is available.
The file size shown in the equation is for a Siemens Simcenter time file. This does not necessarily apply to other file formats from other acquisition systems. The file size generated by a data acquisition system depends on the number of actual digits used to store each sample. The higher the number of digits, the greater the precision, but the higher amount of storage that is required.
To acquire and handle large amounts of data, different system configurations are possible as shown in Figure 14.
Figure 14: Left – A master-master configuration has one computer and multiple Simcenter SCADAS systems. Right – A distributed configuration with multiple nodes that are backed up to a server.
Configurations could consist of (arranged from lowest data throughput to highest data throughput):
Single Computer: A single Simcenter SCADAS (signal conditioning and analog-digital converter) and a single computer.
Master-Master: Multiple Simcenter SCADAS and a single computer.
Distributed: Multiple Simcenter SCADAS, multiple acquisition computers, and a single control computer.
The selection of a configuration depends on the size and rate of the dynamic data being acquired.
For smaller amounts of data, a single computer and a Simcenter SCADAS can be used.
When the amount of data exceeds what can be transferred through a single ethernet cable, then a multi-master configuration can be used. This allows multiple Simcenter SCADAS units to transfer data to a single computer through an ethernet switch.
Even higher amounts of data can be handled in a distributed configuration. In a distributed configuration, multiple computers and multiple SCADAS pairs are used. The data from each pair is continuously transferred to a server while the acquisition is in progress.
The data, when stored on the network server, will be time synchronized and combined into a single file. This can be done by using a common time stamp protocol. For example, the Institute of Electrical and Electronics Engineers (IEEE) 1588 standard defines a precision clock synchronization for networked acquisition systems. Alternatively, an IRIG-B (Inter-Range Instrumentation Group timecode) or Global Positioning Satellite (GPS) signals could also be used to synchronize the data.
When a critical test of a one of a kind prototype is being performed, it is often desirable to record the data redundantly. Redundant recording of data means that the data is acquired and stored in two places at once.
Some options for redundant data acquisition are shown in Figure 15.
Figure 15: Left - Simcenter SCADAS can record in simultaneously on an internal compact flash and the host computer. Right – Simcenter SCADAS VCF4 charge card equipped with an analog output for each analog input.
A common way of doing a redundant data acquisition is to output the incoming signals to a separate data acquisition system. For example, a Simcenter SCADAS can be outfitted with analog outputs for each input channel. The data is conditioned in the SCADAS card, and the incoming signal is converted to an equivalent analog voltage output. This output can be recorded by a secondary acquisition system.
A second option, available in Simcenter Testxpress, is to record on internal media in the SCADAS (a compact flash card) and in parallel to the host computer. Two files are created this way. If the connection between the SCADAS and the host computer was to be lost, recording will continue to the internal media.
High throughput test campaigns can last many days. While the test is in progress, there are typically many stakeholders who need to view the data. For example, engineering groups might check that levels are within prescribed limits, or at predicted levels.
A monitoring station is used for to allow non test-cell operators to view the data at remote locations. A monitoring station is shown in Figure 16.
Figure 16: While the test is in progress, test stakeholders need to be able to monitor the data from anywhere in the world.
If implemented, monitoring should be available anywhere in the world over a computer network. The test operators should be able to select which channels of information are viewable by remote groups.
While monitoring, the dynamic data should also be viewable not just in time format, but as other data types as well. Other types include spectra, orders, colormaps (Figure 17).
Figure 17: Colormaps and orders can be specified from a monitoring station.
Monitoring is different than data analysis. The test in progress is not stopped while monitoring takes place.
In some test campaigns, testing may be halted while analysis of wind tunnel acoustic hotspots or modal results are performed. This analysis needs to be done quickly, but properly, to decide whether to move to the next test point.
Additional alarming stations can be used to monitor levels to ensure they do not exceed critical limits during a test. The alarming stations need to react quickly if levels exceed expectations. For example, the vibration level on key components should be constantly checked. If the vibrations become too large, the alarming signal would inform the test controller so that a shutdown can be initiated before any damage occurs to the test article.
4. Test Reporting: Data Selection, Processing and Reports
The results of the test need to be published. Analysis software that contains certain features can help make creating reports from terabytes of data an easier process.
Some of these features are described in the next sections.
4.1 Data Selection
Easily finding a specific event in the terabytes of data from a large multi-hour/day, multi-channel test is of paramount importance.
For example, data acquired while using IEEE-1588, IRIG-B, or GPS (mentioned in previous sections) can be viewed in absolute time. Absolute time gives time in year, month, day, hour, second, millisecond.
In Simcenter Testlab, absolute time can be shown by right clicking on the X-axis and choosing “Absolute Time” as shown in Figure 18.
Figure 18: In Simcenter Testlab, data can be viewed versus absolute time by right clicking on the X-axis. This allows time segments to be selected down to the millisecond for a given day, month, and year.
If test logs were kept to note when specific events were initiated, this allows an easy selection of the event.
The speed of visualization is important as well. Here, some “tricks” can be used to allow quick visualizations of time histories. Even if a time history has one million data points, a computer screen can usually show only about 2000 data points, or pixels. This creates an opportunity for an illusion of speed!
File formats like the Simcenter Testlab LDSF file (*.ldsf) create a reduced number of points overview which is stored in the time data file. This reduced point reduction is made in an intelligent way, with the proper minimum and maximum amplitudes.
When initially viewing the data, the overview is shown. As a user zooms into the data, eventually the individual data points are shown. This can be done transparently for the user, so that data selection is fast, responsive, and accurate.
After an event is selected, it can be processed. This is described next.
To fully analyze and process the data collected, multiple steps might be involved. An example of these steps might include:
Conditioning: Preparing time domain data. For example, calculating rpm from pulses, applying filters, integrating, subtracting channels, etc.
Analysis: Performing frequency analysis including order extraction.
An example of preparing time data for analysis is shown in Figure 19 below.
Figure 19: The Simcenter Testlab Time Signal Calculator is used to calculate the relative displacement between two measurement locations by converting from acceleration to displacement and subtracting.
The Simcenter Testlab Time Signal Calculator allows time domain data to be prepared for further analysis. Operations are done as equations.
In the report itself, Siemens Simcenter Testlab allows graphs to be stored in “Active Picture” format.
An active picture allows a graph to be activated in a Microsoft Office report (Word, PowerPoint, etc.). Without a license of the software, a user can: change limits, add cursors, determine peaks, integrate, etc. It is like a “living” graph.
Active pictures also support end-to-end traceability of data as shown in Figure 22.
Figure 22: After activating an Active Picture, right click and choose “Data Properties” to view key information. In this case, viewing the data properties shows that the low levels of the red curve are due to an improperly entered sensitivity value on one axis of a triaxial accelerometer.
In an active picture, the data properties can also be viewed. This includes information like transducer serial number and sensitivity, as well as the serial number and channel of the Simcenter SCADAS used for acquiring the data.