Emanuele Burgognoni

Thursday, May 23, 2024 · 0 min read

The Effects of Different Accelerometer Adhesives for Vibration Measurement on Steel Structures

Fixing the accelerometer sensors is sometimes tricky and complicated when doing modal analysis of steel structures. Mechanical anchors such as screws and plugs are impossible to use. At the same time, the installation represents a crucial step to obtain reliable data on the structure's dynamic behavior. Using two-component and single-component adhesives plays a fundamental role since it directly affects the accuracy of vibrations' transmission and modal measurements.

What is the impact of varying the two types of glue on the data acquisition phase? How does the response vary over transmitted frequencies, and what are the practical implications of structural analysis?

Fixing sensors on structures

Depending on the structures to monitor, we can use a variety of mechanical anchors. Screws and plugs are easy to apply and quick to use. Various types depend on the material you work on (reinforced concrete, plasterboard, wood, etc.). They have high durability, especially if galvanized, and provide a firm grip for each element anchored through them.

Figure 1. IOLITEiw-3xMEMS-ACC-8g-T module attached to the soffit of a bridge deck.

Screws and anchors provide a solid mechanical anchor. It may be preferable in situations that require greater tensile strength or lateral load resistance. Strong mounting is critical in applications where the structure is subject to intense vibrations or significant movements.

The use of screws and anchors can lead to perforation or damage to the structure. Such damage is a risk to consider, especially when dealing with fragile or delicate materials, such as in the case of some types of steel or historic structures. Their installation requires drilling into the structures, which may be impractical or undesirable in some situations, such as particularly hard or inaccessible surfaces.

At this point, glue takes over. It may be the preferred way to avoid structural damage or working with materials you cannot simply drill.

Single-component and bi-component adhesives

Two-component epoxy adhesives constitute a versatile solution for fixing equipment. It presents an innovative approach through two distinct compounds during the installation phase. In this formula, the first component houses the resin, while the second contains the hardening agent essential for triggering the polymerization process.

The effectiveness of these glues depends significantly on the mixture of the two components, which must remain isolated until used. Once you have carefully combined the two compounds, the remaining operating time may be partly affected by environmental conditions, as highlighted by the example in Figure 2.

Figure 2. Polymerization times of the glue as the temperature varies.

Mixing the components of a two-component glue results in an additional delay in the application process. Single-component glues can be used immediately from the container. Furthermore, the drying time may vary depending on ambient temperature and humidity and the correct proportion of the two mixed components. These factors can influence the speed and effectiveness of the polymerization reaction, thus influencing the time required for the two-component glue to harden fully.

The single-component epoxy glue is distinguished from the previously mentioned products by a peculiar characteristic: both reagents necessary for polymerization are already present inside the cartridge. Contrary to what you might expect, the polymerization reaction doesn’t occur right after you remove the glue from the cartridge.

This delay is because a chemical reagent does not activate the polymerization in this type of glue but rather by exposure to temperatures higher than ambient temperatures. In other words, the polymerization reaction is thermo-activated. For single-component epoxy glue to cure fully, you must expose it to high temperatures, generally above 120°C.

This heat-activated characteristic of one-component epoxy glue offers several advantages. The benefits include greater flexibility in managing working time and improved resistance to high temperatures once fully cured. Although its flexibility is advantageous in anchoring objects, it is also a disadvantage in transmitting vibrations.

Adhesive installation times

Adhesives offer a perforation-free solution that can be more discreet and aesthetically pleasing. The strength of the fixing depends on the quality of the adhesive and surface preparation. These qualities translate into longer application times to ensure secure adhesion and long-lasting.

The glue fitting is an essential step in installing sensors in the field. On the one hand, we need to apply the correct quantity of glue, but on the other hand, excess glue requires more time to make so that it hardens.

Figure 3. We used an elevating work platform (PLE) to assemble the instrumentation at height.

Drying introduces significant temporal variations in the assembly of the instrumentation. If a two-component glue requires an average of 10 minutes of drying and multiply this number by 80 sensors, we obtain a time dilation in the installation process of more than 13 hours. 

Such a period means extending the use of personnel and machinery for another three days. In economic terms, this means adding an expense of thousands of euros to the cost of completing the work. 

Not to mention the difficulty of requesting additional traffic-blocking permits to carry out work along the route instrumented. In short, from many points of view, using glue is only recommended if there is no other way. What can be done in this case to optimize the installation process?

The need to carry out in-depth analyses of the types of glue arises. Given its greater flexibility after the drying phase, single-component glue is less suitable for transmitting vibrations. However, it is also quicker to install as it has shorter stiffening times thanks to its thermo-active response. For this reason, we investigated the effect of the single-component glue used as a transmission (and anchoring) means for accelerometer sensors.

We used the following hardware and software for these tests:

Dewesoft Artemis OMA is specialized testing software for performing OMA (Operation Modal Analysis), EMA (Experimental Modal Analysis), ODS (Operating Deflection Shapes), and structural health monitoring analysis.

IOLITEi 3xMEMS-ACC is a family of data acquisition devices with an embedded triaxial MEMS accelerometer, analog-to-digital conversion (ADC), and EtherCAT interface. It is tailor-made to monitor the structural health of large structures such as bridges.

First test: mechanical vs. chemical anchoring

The first test aimed to compare the differences in the signal provided by the two modules fixed with and without glue. The sensor body, fixed without glue, was anchored to the structure using self-tapping screws. The object of the analysis was the leaf of an industrial metal door.

Although it does not resemble a scientific test, comparing the signals coming from the two nearby sensors should reasonably represent the same vibration net of the spatial and temporal errors, which, in this case, are both considered null.

  • We considered the temporal error zero as we used two IOLITEiw (waterproof version) modules designed for synchronous distributed acquisition (synchronization accuracy is more than one µS.

  • We considered the spatial error null as we positioned the two sensors a few centimeters apart on the leaf of a 3m high industrial door subject to the same oscillation.

Figure 4. We installed IOLITEiw-3xMEMS-ACC-8g-T modules to perform the first adhesive test.

We paid attention to what effects the glue introduced on the acquired vibrations. Strictly speaking, any material placed between a sensor and a body to be analyzed behaves like a mechanical filter, imposing variations in the module and phase of the signal. The questions to answer at this point are: what are these effects, and to what extent do they manifest themselves?

Figure 5. The layer of glue used for fixing is visible on the module on the right.

The analysis showed the glue's apparent elastic behavior, which shifts the signal along the axis perpendicular to the plane of the door leaf (Y axis). The IOLITEiw-3xMEMS-ACC-8g modules have a self-alignment function of the relative axes with the absolute axes. 

This functionality means installing them in level is unnecessary, thus speeding up the assembly processes. In this case, you don’t have to give weight to the direction of the axes shown on the module body since the software does the realignment. See the correct orientation of the axes in Figure 4.

The signals shown in Figure 6 result from a 30 Hz low-pass filter that eliminates all vibrations with too high a frequency compared to those of interest for the structure analysis.

Figure 6. Time domain signals. In red, we anchored the module with glue, while in green, with self-tapping screws.

From this first test, we concluded that the glue's elastic effect imposes a significant phase shift in the acquired accelerometer signal. We proceeded with tests to obtain more information on this phase shift.

Figure 7. The team of SITE technicians carried out the tests with the support of Dewesoft Italy.

Second test: phase shift effects vs. quantity of glue

In this second test, we chose to make the following changes:

  • Reducing the quantity of glue applied during installation;

  • Adding a sensor with a hybrid fixing logic (glue + screws);

  • Carrying out analyses on a more flexible structure.

We introduced the first two points to study the phase variations as the conditions for fixing the modules with the glue vary. The more flexible structure provides more significant oscillations that facilitate the comparison of the signals generated between modules in the time domain.

Figure 8. To perform the second adhesive test, we installed IOLITEiw-3xMEMS-ACC-8g modules.

For practical reasons, installing the modules at the same measurement point was impossible. Therefore, to confine the analysis of the phase shift only to the contribution imposed by the glue, a 26.6Hz bandpass filter corresponds to the structure's first mode of vibration. 

This resonance frequency is the first and only one to present oscillations in phase for all measurement points along the bar - see Figure 10.

Figure 9. The first three mode shapes of a vertical steel bar.
Figure 10. Frequency spectrum with a peak at the resonance frequency.

The green signal represents the module fixed to the structure with screws, the red one the glued module, and the blue one the module fixed with glue and screws.

The proposed spectrum highlights a frequency peak at 26.6 Hz common to all three sensors. This peak is characteristic of a resonant frequency. From the frequency domain analysis, we then moved on to the time domain analysis to analyze the phase shift imposed by the glue once again.

Figure 11. Time diagram. 
Figure 12. Time diagram with zoom.

The timeline returns phase-shifted signals comparing those acquired with the screwed module (in green) and those acquired with the glued modules (in red and blue).

In short, the second test's results confirm the presence of a time lag in the signal acquired with the modules fixed using the glue.

Third test: glue effects on the modal analysis

Although carried out in non-optimal conditions, this last test aimed to obtain the first modal shapes of the bar under analysis. It was possible to prove that you can carry out the modal analysis of a structure by mounting the sensors with single-component glue.

Figure 13. A picture of the sensors installed to carry out the modal test analysis of the steel bar.
Figure 14. The installation of the modules on the steel bar

We found the most significant difficulties in the spectral density noise during this last step. We are talking about non-optimal conditions since the analyzed body links to other metallic components with resonance frequencies within the band of our interest.

Noise manifests as a series of irregular peaks and valleys in the frequency spectrum. These peaks and valleys can be more pronounced at specific points on the spectrum and less noticeable at others, creating a visually chaotic appearance. The cause of this noise lies in the complex interaction between the main analyzed body and the other structural parts connected to it.

When the main body vibrates, other connected structural parts can resonate at specific frequencies, amplifying and modifying the transmitted vibrations. This phenomenon generates a variety of frequencies that overlap in the spectrum, creating the observed noise. These additional frequencies can make identifying and interpreting the main body's vibration modes more challenging.

Despite the apparent chaos in the frequency spectrum, you can apply advanced analysis techniques to separate and identify main body vibration modes from other structural interferences. These techniques may involve filtering algorithms and data analysis to isolate and characterize different vibration components.

Figure 15. We reduced the glue layer to a minimum to verify the reduction of the effects on vibration.
Figure 16. Modal analysis signals in the time domain. In this application, we fixed all sensors with glue.
Figure 17. Modal analysis signals in the frequency domain. In this application, we fixed all sensors with glue.

But what do we expect from modal analysis? The theory requires us to obtain a dynamic behavior of this kind - see Figure 18. We see that the first flexural mode shapes reflect the same behavior as a beam constrained at the ends. So it was, the mode shapes obtained were in line with expectations.

Figure 18. Bar theoretical mode shapes.

By carrying out the modal analysis with Dewesoft Artemis OMA, we obtained a spectral density diagram with a very high noise level. It made it possible to identify only the first vibration modes.

Figure 19. A spectral density plot in Dewesoft Artemis OMA.
Figure 20. First mode shape of the steel bar.


Figures 20 and 21 show the reconstructions of the first flexural vibration modes. The areas with the warmest colors identify the parts subject to more significant deformation.


Using glue to fix the accelerometer modules on metal structures can significantly influence the measurement. Therefore, it requires careful and targeted analysis, considering the different types of glues available.

On the one hand, we have two-component glues characterized by rigid drying due to the presence of the hardening component. These adhesives are excellent at transmitting vibrations through the structure, ensuring accurate detection of vibration modes. 

It is essential to consider that using two-component glues can entail some disadvantages. First, they can be expensive, as they require purchasing two separate components. Additionally, the drying process is often slow, leading to delays in setup and measurements. Finally, their rigidity may make them impractical in certain application situations where flexibility is essential.

On the other hand, single-component adhesives stand out for their quick drying. These adhesives offer a more time- and cost-effective alternative, allowing for more efficient sensor installation. However, the greater elasticity of single-component adhesives can lead to some compromises. In particular, they can cause attenuations and phase shifts of the vibrations perceived by the sensors, affecting the precision of the measurements.

Ultimately, the choice between two-component and single-component glues depends on the specific needs of the glue, the application, and the characteristics of the structure in question. It is essential to carefully evaluate the advantages and disadvantages of each option to ensure accurate and reliable measurements of the structure's vibration modes, especially if the ultimate goal is to perform a modal analysis.

Following our experimental experiences in the field, we consider a scientific, in-depth study on the topic beneficial, as it can shed light on the effect of the most common glues in processing accelerometer signals.

Figure 22. The SITE technical team is preparing the material to fix the modules. From left to right Daniele Nisticò, Salvatore De Rinaldis, and Simone Di Marco.


For the availability and outstanding professionalism that distinguishes them in all our collaborations, we thank: