MEMS Vibrational Power Generator for Bridge Slab and Pier Health Monitoring

Round 1

Reviewer 1 Report

the paper presents a theoretical study on the possibility of applying a new type of sensor in the health monitoring of concrete bridges, which if of great relevance to the society considering the aging infrastructure. The work really demonstrates the possibility of this new sensor technology. Nevertheless, looking at the paper itself, I would advise the authors to carefully revise the paper again to improve it further. Hopefully my comments below could help:

1. Line 38 what does the authors mean by stating > 2m? Bridges with length larger than 2 m? Why is such statement important.
2. Line 48-49: Pier souring is the major issue for railway bridges, does the authors have source on this conclusion? Why only railway bridges?
3. Line 50: The authors claim that for concrete bridges the main damage types is fatigue. I strongly disagree with this. I believe you meant steel concrete composite decks.
4. Equation 1: Adjust the fond size
5. Equation 4: It would be even more clear if the authors can show the relationship between acceleration/displacement and amplitude directly.
6. Line 145: ‘It is known that the dominant vibration frequency of a structure generally decreases when the bridge deteriorated.’ This is not a physically valid argument. What can happen is the reduce of frequency is related to reduce of stiffness.
7. Equation 5: I believe at its current form, Eq5 is incorrect. x looks like the displacement of matrix for me. An second derivation over time is needed on the right hand side. Please do check this again.
8. Line 163: Is it common to have a T shape steel profile in Japan? For structural applications, I shape profile is most widely applied.
9. Line 167-168: It is very unclear in the description what kind of damage are you simulating and which material properties are you turning down. When I look at the table 2, it seems that you have turned down the concrete elastic modules from 30 GPa to 2 GPa. This is an extremely low value even for damaged concrete. Do you have evidence that in real work concrete with such low equivalent elastic modules is possible?
10. Figure 3: Please provide a clear description on how the concrete panels are connected to the steel footing? how does the concrete panels connects to each other? These are essential conditions for the modelling.
11. Figure 3: the authors also mentioned about reinforcement, although I don’t think it affects the simulation, but would you indicate how are the panels reinforced?
12. Line 182: Please always use either GPa or kN/mm2 for elastic modules. Don’t mix them.
13. Section 2.4.2: when the authors introduced the modeling of bridge piers, it looks like that the pier is not connected to the bridge. For the pier, I think the boundary condition on top is not correct. Because the mass of the bridge deck and the stiffness of the bridge deck will have a clear impact to the eigen frequency of the pier. Neglecting that will lead to wrong interpretation of results.
14. Section 3.1. I have the feeling that section 3.1 belong to the introduction section.
15. line 270: the authors mentioned that the response frequencies of the sensors can be adjusted in accordance to the nature frequency of bridge. How did the authors simulate that? I believe the authors obtain displacement data from the simulation. But no detail on the conversion from FEM results and energy from MENS is given.
16. Equation 6: If I understand it correctly, it shows the deviation of the average energy harvest per sensor. But then you can always find the maximum, besides, how do you know if there are two panels being damaged?
17. General comment: In real life, how will you execute this? I have the feeling that you need to have always the same excitation at the same location to be able to compare the response of the structure. This is probably note so applicable.
18. General comment: how will you collect the electric signal? you will need additional energy to send the signal or wire to send the signal right?

Author Response

1. Line 38 what does the authors mean by stating > 2m? Bridges with length larger than 2 m? Why is such statement important?

In Japanese bridge inspection standard executed by Ministry of Land, Infrastructure, Transport and Tourism, all the bridges which their length is more than 2 m are regularly targeted to implement the present inspection protocol once-in-5-year for preventive maintenance.

2. Line 48-49: Pier souring is the major issue for railway bridges, does the authors have source on this conclusion? Why only railway bridges?

It is about bridges over waterways, not only about railway bridge. (Line 49-50)

3. Line 50: The authors claim that for concrete bridges the main damage types is fatigue. I strongly disagree with this. I believe you meant steel concrete composite decks.

Since it was misleading the introduction of fatigue damage in RC slab, Line 51 says that the often-occurred damage type is fatigue damage due to numerously repeating traffic overloads in steel-concrete composite decks.

4. Equation 1: Adjust the fond size

It has been adjusted.

5. Equation 4: It would be even more clear if the authors can show the relationship between acceleration/displacement and amplitude directly.

In order to show the electrical performance of MEMS vibrational power generator, it is more important to derive P (power generation amount) and I (circuit current). Therefore, the equations (1) to (4) are described to express them. Also, in Line 136 to 138, it is noted that acceleration is proportional to vibration amplitude and frequency squared.

6. Line 145: ‘It is known that the dominant vibration frequency of a structure generally decreases when the bridge deteriorated.’ This is not a physically valid argument. What can happen is the reduce of frequency is related to reduce of stiffness.

I totally agree with your comment. The dominant vibration frequency of a structure generally decreases when the stiffness of bridge components is reduced with a sort of material and structural deterioration. (Line 164-165)

7. Equation 5: I believe at its current form, Eq. 5 is incorrect. x looks like the displacement of matrix for me. An second derivation over time is needed on the right hand side. Please do check this again.

Equation 5 is correct. This is an eigen-equation rather than an equation of motion. In the equation of motion, yes, a second derivation over time presents, but when it is converted to eigen-equation (by a separation of time and spatial variables), the spatial dependent eigen-equation may hold no time derivatives. For this reason, no revision was made on Equation 5.

8. Line 163: Is it common to have a T shape steel profile in Japan? For structural applications, I shape profile is most widely applied.

“T-shaped” was incorrect description. I shape is common in Japan as well. Therefore, it was modeled with I-shaped girder (Line 183). Additionally, the explanations in Table 2 are modified.

9. Line 167-168: It is very unclear in the description what kind of damage are you simulating and which material properties are you turning down. When I look at the table 2, it seems that you have turned down the concrete elastic modules from 30 GPa to 2 GPa. This is an extremely low value even for damaged concrete. Do you have evidence that in real work concrete with such low equivalent elastic modules is possible?

2.01 is an error in writing. In Table 2, Young’s modulus (E) for “deteriorated concrete” is modified to be 7.5 GPa (severely damaged condition), where it is assumed that ultrasonic velocity (C) obtained by non-destructive inspection testing falls to one-half after deterioration. Since E decreases as the inverse square of C in physics, E of deteriorated concrete is 7.5 GPa when that of intact concrete is 30 GPa. (Line 203)

10. Figure 3: Please provide a clear description on how the concrete panels are connected to the steel footing? how does the concrete panels connects to each other? These are essential conditions for the modelling.

The connections between the panels and between panels and girders are all rigid connections. To address this concern, the sentence in Line 184 to185 was revised as follows. They were all modelled with solid elements, in which the panels as well as the panels and girders are rigidly connected.

11. Figure 3: the authors also mentioned about reinforcement, although I don’t think it affects the simulation, but would you indicate how are the panels reinforced?

It is added that there is no steel reinforcement to simplify the model in Line 183-184. The writing is modified that the model consists of 6 concrete, not RC, panels in Line 182.

12. Line 182: Please always use either GPa or kN/mm2 for elastic modules. Don’t mix them.

The notation is unified as “GPa”. (Line 203)

13. Section 2.4.2: when the authors introduced the modeling of bridge piers, it looks like that the pier is not connected to the bridge. For the pier, I think the boundary condition on top is not correct. Because the mass of the bridge deck and the stiffness of the bridge deck will have a clear impact to the eigen frequency of the pier. Neglecting that will lead to wrong interpretation of results.

Yes, you are right. The superstructure may affect the eigen-frequency of the pier to some extent, but to include it into a full superstructure-substructure model may greatly increase the computation cost. A simple way to model the superstructure is added mass. Our major focus is to detect scour with a change in the pier bottom boundary condition. Known that pier is a cantilever-like structure, no matter whether an extra mass is added on its top or not, the change in its bottom boundary condition would change its vibration properties in a similar way. For this reason, the superstructure is omitted in this study with little loss of generality. The additional explanation is mentioned in Line 231 to 234.

14. Section 3.1: I have the feeling that section 3.1 belong to the introduction section.

In accordance with moving Section 3.1 to the introduction section at the end of Section 1.3, Figure numbers (from Figure 3 to Figure 10) and Section numbers in Section 3 are changed in the revised draft.

15. line 270: the authors mentioned that the response frequencies of the sensors can be adjusted in accordance to the nature frequency of bridge. How did the authors simulate that? I believe the authors obtain displacement data from the simulation. But no detail on the conversion from FEM results and energy from MENS is given.

As for adjusting the frequency response of MEMS vibrational device to the natural vibration frequency of bridge, there is no simulation but the device itself can be designed to achieve the targeted frequency, which is explained in Section 1.3 from Line 90 to 92. It is assumed in this study that the computation on conversion from FEM results to generated energy from vibrational MEMS device, using Equations (1) to (4), is followed by expressing sinusoidal vibrational waveform as the relationship between displacement and time in accordance to relative vibration amplitude and frequency provided from FEM analysis at each node of interest. The above-described explanations are added at the end of Section 3.1. (Line 268-271)

16. Equation 6: If I understand it correctly, it shows the deviation of the average energy harvest per sensor. But then you can always find the maximum, besides, how do you know if there are two panels being damaged?

That is very good suggestion. In the present paper, only single panel is supposed to be damage. (Line 186) Actually, we have considered the cases with two panels being damaged. Using MEMS vibrational power generator to detect those damage locations, there is different algorithm applied in the case of multiple-panels damaged situation. We would like to introduce the further results, which is under investigation right now, in another paper soon. Although the improved algorithm capable for multiple-panels damage detection is a bit complicated, the damage locations can be detected with high accuracy.

17. General comment: In real life, how will you execute this? I have the feeling that you need to have always the same excitation at the same location to be able to compare the response of the structure. This is probably note so applicable.

The proposed system only monitors the power generation potential or behavior of MEMS vibrational devices at observed point, so that vibration frequency and amplitude at each point targeted are detected without using conventional accelerometers or power grid cables. This is the point of the present research work. We do not have to the same excitation of vibration. The MEMS devices, in environmental vibration, installed at specified (fixed) observation points on the targeted structure detect the relative change in values of power generated after a sort of damage is occurred.

18. General comment: how will you collect the electric signal? you will need additional energy to send the signal or wire to send the signal right?

In our opinion to install this system for existing structures, the electrical signal from MEMS device is not needed but it is necessary to figure out power generation from the devices. In the future achievement of much higher efficiency of power generation from MEMS vibrational device than its current technical performance, the signals would be wirelessly transmitted to the receiver on using its own power energy. For an easy-to-use example of the application for now, the generated power can be visualized with flashing LED lights in proportion as vibration frequency and amplitude. I believe that there might be other options to utilize this innovative device which enable to detect environmental vibration for structural health monitoring of civil infrastructure.

Reviewer 2 Report

A self-powered, wireless sensor is attractive for monitoring bridge vibration to detect damage.  A previous study  (ref. [22]) has shown that a micro energy harvester device can generate enough energy to power itself using typical bridge vibrations. Two diagrams from that paper are shown in present paper, and are referenced. This paper uses finite element modelling of two idealized bridge structures, associated with smaller bridges, to show that signals indicating the presence of damage in concrete can be obtained, and the location of damage may also be found, but with some uncertainty.  This is a small step to implementation, as it has not been tested in a laboratory, let alone in practice.  As well, the modelling simplifications may exaggerate the ability to distinguish changes in signals, and in practice there are many different types of bridges.

The text is well written, with occasional minor errors.  eg Page 12 line 365 seems to have a phrase by itself that should have been deleted and line 369 has 'propose' when it should read 'proposed'.

Author Response

Thank you so much for the great and positive comments.

As for some writing errors, I have revised the phrases in Line 370-371 and Line 375.

To step further as you kindly suggested, we have already proceeded to some lab experiments on “validity verification” of our proposed MEMS power generator utilized as vibration sensor and structural monitoring device. We would like to report in the next paper in near future.

Round 2

Reviewer 1 Report

The authors have addressed most of my comments. In principle I agree to publish it.

There are two points still needs to address:

1. Eq 5 might be correct numerically, but physically not as the units can not be consistent. Please do clarify it.
2. Regarding the boundary condition of the bridge pier FE model, it is also nice to clarify in the text of the paper that simplification has been made from the reality as points of improvement in the future.

Author Response

1. Eq 5 might be correct numerically, but physically not as the units can not be consistent. Please do clarify it.

The units are also consistent. Equation 5 reads Kx=(lambda)Mx.

On the left hand side of the equation, the SI units in K matrix are N/m (equivalent to kg/s^2) for translation degrees of freedom and N-m (equivalent to kg-m^2/s^2) for rotation degrees of freedom. On the right hand side of the equation, the units in M are kg for translation degrees of freedom and kg-m^2 for rotation degrees of freedom. The unit of lambda is always 1/s^2 no matter the degree of freedom is translation or rotation (it makes sense as physically lambda is frequency squared).

2. Regarding the boundary condition of the bridge pier FE model, it is also nice to clarify in the text of the paper that simplification has been made from the reality as points of improvement in the future.

In accordance with your kind suggestion, a sentence from Line 234 to 236 has been additional to explain the further investigation on the FE model of pier in real.