Demonstration with optical fibres by Smart Fibres Ltd. Task 15

Transcription

1 Demonstration with optical fibres by Smart Fibres Ltd. Task 15 Dutch Offshore Wind Energy Converter project DOWEC rev1 Name: Signature: Date: Written by: J.F. Kooij (LMGH) version Date No of pages New document in PDF DOWEC rev1 Page: 1 of

2 Summary: One of the techniques to do load registration is the use of optical fibres in a blade. Smart Fibres Ltd. in England has developed an optical fibre strain sensing system that can measure strain in a composite structure. To give a feeling about the capabilities of this system a demonstration has been held on an existing structure. This structure was the sectional blade as tested at TU Delft. This document gives an outline of this demonstration. The demonstration gave promising results: The differences between strain gauges and optical fibre sensors are severe for lower load levels; The differences for higher load levels give a constant difference (5 %); The results for a single and a multiplexed optical sensor are comparable; In the dynamic testing the results in the low strain region differ more than 10 %. For the high strain region the differences are comparable to the differences seen at the static test (3-5 %). Contents 1 Introduction Mounting of optical fibres Test blade Alignment of fibres Lamination specification for demonstration Mounting strain gauges Testing Static testing Comparison between strain gauges and optical fibre sensor Comparison between single and multiplexed fibre Dynamic testing Conclusions & recommendations Conclusions Recommendations Appendix A: Strain gauges vs. Optical fibres (pos. strains) DOWEC rev1 Page: 2 of

3 1 Introduction In the DOWEC project a task described in the knowledge development is the load registration and condition monitoring for an offshore rotor. One of the techniques to do load registration is the use of optical fibres in a blade. Smart Fibres Ltd. in England has made a commercial system to implement this. They have developed an optical fibre strain sensing system that can measure strain in a composite structure. To get a feeling about the capabilities of this system a demonstration has been held on an existing structure. This structure was the sectional blade as tested at TU Delft. This document gives an outline of this demonstration. The demonstration contained the following: Mounting of optical fibres on an existing structure (handling, etc); Comparison of single grating in fibre and a few gratings in series (multiplexed); Comparison between optical fibres and strain gauges (conventional method). Test performed in co-ordination with TU Delft and ECN. In chapter 2 an outline of the mounting of the optical fibres (SF) is given and in chapter 3 the mounting of the electric strain gauges (ESG). The comparison between strain gauges and between the single and multiplexed fibres is given in chapter 4. Finally the conclusions are drawn in chapter 5. DOWEC rev1 Page: 3 of

4 2 Mounting of optical fibres 2.1 Test blade The structure on which the demonstration was performed was the 23m sectional blade developed in a European project (JOR , Design, structural testing and cost effectiveness of sectional wind turbine blades ). The blade is divided into two parts and they are mounted together with an Ikea (T-bolt) connection. This blade was already mounted in a test rig at TU Delft and the testing was finished when the demonstration was held. 2.2 Alignment of fibres To do a good comparison, two locations were selected at which measurements were performed. These locations are a location where low strain levels occur and one where high strain levels occur. The low strains occurred near the trailing and leading edge of a cross section if loaded in flapwise direction. As the trailing edge was easier to use for mounting this side was used to mount the optical fibres. High strain levels occurred at a cross section located at L = 5300 mm. A layout of the locations where the optical fibres sensors were mounted is shown in Figure 2.1. Figure 2.1: Locations for mounting optical fibres At each location two gratings were mounted, one in line with another grating in the same fibre and one single grating in a fibre. This is useful to compare the measurements between a single grating and one in line with another grating. The optical fibres, and of course the gratings, are located on the outside of the blade. The gratings are mounted parallel to the neutral axis of the blade. This all lead to 3 optical fibres with in total 4 gratings (2 gratings in line, and two in a single fibre). 2.3 Lamination specification for demonstration The laminating specifications for the post embedment of fibre optical strain sensors are used as recommended by Smart Fibres. Unidirectional 450g/m 2 E-glass tape is used to laminate optical fibres onto the blade surface using the following lay-up (see Figure 2.2). The method of applying resin is hand lay-up. DOWEC rev1 Page: 4 of

5 Figure 2.2: Lay-up for mounting optical fibres The 1 st layer was 100 mm wide UD-tape, then the optical fibres are embedded into place, followed by a second layer of 60 mm wide UD-tape. E-glass tape (0-90º) is placed on top of these two layers to have a better protection if the strain gauges are mounted and finally a layer of peel ply is placed to suck out the redundant resin. The distance between the sensors was 10mm. Polyester resin has been used and a slow hardener to give plenty of working time before setting of the laminate by heating it up. Special care had to be taken of the location where the optical fibre runs out of the laminate. During lamination resin with added silica had to be added to the surroundings of this vulnerable area. To make sure the protected part of the fibre (blue in figures) did not stick to the resin it was rubbed with a release wax. The peel ply had to be between the laminate and the optical cable (see Figure 2.3). Figure 2.3: Ending of optical fibre in laminate A complete patch with optical fibres mounted in a laminate is shown in Figure 2.4. DOWEC rev1 Page: 5 of

6 Figure 2.4: Complete patch 3 Mounting strain gauges When the resin of the laminate was fully hardened, strain gauges were mounted on the locations where the gratings are present. Four strain gauges were enough to give a good comparison (one strain gauge on top of each grating). Now it became clear that the last layer of glass (0-90º tape) was necessary as the surface had to be sanded to give a good adhesion between the strain gauges and the laminate. A detail is given in Figure 3.1. A systematic sketch of layout is shown in Figure 3.2 and the complete layout is shown in Figure 3.3. In the sketch the two different measurement systems are separated to give a better overview, but in reality they are located on top of each other. DOWEC rev1 Page: 6 of

7 Figure 3.1: Strain gauges on surface Figure 3.2: Layout of test set-up DOWEC rev1 Page: 7 of

8 Figure 3.3: Complete test set-up on the blade 4 Testing Two types of tests were performed: static and dynamic. The static test was done with a few different steps to have a good comparison for different loading levels (and thus strain levels). The dynamic test took place for 5 minutes as the data storage of the optical fibre strain sensing system only had a limited capability. These few minutes were also sufficient to give enough data to do a comparison. 4.1 Static testing The static testing was done on six load levels: four levels with positive strain (F = -15 kn, -30 kn, -45 kn & -60 kn) and 2 levels with negative strain (F = +30 kn & +60 kn). In this line-up a positive sign of the force results in negative strains (compression on measured side of blade) and a negative force thus results in positive strains (tension on measured side of blade). The force applied in six different steps and after each load level the force was reduced to zero. The time range was 20 seconds building-up of force, 20 (or 30) seconds of continuous force and 20 seconds building-down of force Comparison between strain gauges and optical fibre sensor The results of the strain gauges are listed in table 4.1 and the results of the optical fibre sensors in table 4.2 and the graphs in appendix A and B. Force kn kn kn kn kn kn DOWEC rev1 Page: 8 of

9 Nr 1. c * * ** Nr ** Nr * Nr ** * ** Table 4.1: Results of strain gauges c: corrected with an offset of µε *: fluctuations between 0 and µε **: fluctuations between 0 and µε On the two locations (high and low strain) the results of the strain gauges are comparable (difference is not more than 10 µε). This is not surprising as the distance is only 10 mm apart. In the static test the measurements do not fluctuate heavily. Only the most occurring value is listed in the table. If a fluctuation is present this result is marked with a *. Strain gauge number 1 gave a strain of µε if the blade was unloaded. Therefore all the results of this strain gauge were corrected with this value. Force kn kn kn kn kn kn Nr Nr Nr Nr Table 4.2: Results of optical fibre sensors *: the standard deviation was approx. ± 10 µε, with maximums reaching ± 20 µε. The results of the optical fibre sensors give similar behaviour as the strain gauges; only the values are slightly higher. The listed values are the average values during the static test. The standard deviation differed between ±2 and ±20 µε with an average of ±10 µε. The differences between the optical fibre sensors and the strain gauges are listed in table 4.3. Force kn kn kn kn kn kn Nr 1. 6 % (19µε) 2 %(3µε) -1 %(1µε) 3 %(5µε) 5 %(12µε) 3 %(11µε) Nr 2. 7 %(22µε) 1 %(3µε) 8 %(7µε) 6 %(9µε) 5 %(13µε) 7 %(24µε) Nr 3. 6 %(68µε) 5 %(30µε) -9 %(25µε) 1 %(5µε) 3 %(28µε) 3 %(30µε) Nr 4. 6 %(73µε) 6 %(37µε) 3 %(8µε) 4 %(26µε) 5 %(40µε) 4 %(52µε) Table 4.3: Differences between optical fibre sensors and strain gauges (percentage and absolute) For the lower load levels (F = kn, kn & kn) the percentage differences between strain gauges and optical fibre sensors are severe, especially for F = kn, but the absolute values are low (except for nr 3.). The results for higher load levels give a more constant difference (5 %). Three reasons can be found for this difference: 1. The strain transfer between the optical fibres is different than for strain gauges. The optical fibres are more or less embedded into the structure and the strain gauges are mounted on top by an adhesive. These different types of mounting are likely to give different results. 2. Due to the embedment of the optical fibres they can also sense a portion of e.g. transverse strains. This is likely to occur in highly curved structures. Strain gauges only measure in one direction and in this case in the direction of the neutral axis. 3. The optical fibres are calibrated by a factor of 2.66 on the raw data. This is used by Smart Fibres as calibration for the raw data in their experience. For this structure or embedment the factor can be adjusted to fine-tune the results. DOWEC rev1 Page: 9 of

10 4.1.2 Comparison between single and multiplexed fibre As mentioned earlier at one location two types of optical fibre sensors are present: a fibre with a single sensor and a fibre with multiplexed sensors. In this section the results of the two sensors are compared. In table 4.4 the mean, maximum and minimum values of the difference between the optical fibre sensors are shown as well as the difference between the strain gauges at one location. In appendix C the results are shown graphically. Force 57.4 kn 29.1 kn kn kn kn kn Average Difference between nr 1 and 2 Mean 6 % 6 % 9 % 2 % 4 % 7 % 6 % Min 9 % 10 % 10 % 3 % 3 % 6 % 7 % Max 3 % 5 % 5 % -3 % 5 % 7 % 4 % ESG 6 % 6 % 0 % 0 % 4 % 3 % 3 % Difference between nr 3 and 4 Mean 1 % 3 % 12 % 4 % 1 % 3 % 4 % Min 1 % 3 % 10 % 4 % 1 % 2 % 4 % Max 1 % 2 % 14 % 2 % 2 % 3 % 4 % ESG 1 % 2 % 0 % 0 % 1 % 0 % 1 % Table 4.4: Comparison between single and multiplexed fibre The table shows clearly that the results of the optical fibre sensors in the low strain area (nr 1. and 2.) differ more than at the high strain area (nr 3. and 4.). As the locations of the sensors differ 10 mm the results are likely to be different, especially in the tail area. This can also be concluded out of the results of the strain gauges. The difference between the two optical fibre sensors is approximately 3 % higher than the difference of the strain gauges at the same location. For the lower loading levels (especially in the low strain area) the differences are the greatest (10-12 %). The specifications of the Smart Fibres system list a resolution of ±10µε so it will not be surprisingly that for a low strain level the same absolute magnitude of fluctuation gives higher deviations in percentage than for higher strain levels. Therefore it is also necessary to look at the absolute differences (see table 4.5). This shows that at the low strain location the absolute difference is less than at the high stress location. Force 57.4 kn 29.1 kn kn kn kn kn Average Difference between nr 1 and 2 SF 22 µε 9 µε 8 µε 4 µε 11 µε 22 µε 13 µε ESG 18 µε 9 µε 0 µε 0 µε 9 µε 14 µε 8 µε Difference between nr 3 and 4 SF 14 µε 17 µε 33 µε 21 µε 12 µε 31 µε 22 µε ESG 9 µε 9 µε 0 µε 0 µε 0 µε 9 µε 5 µε Table 4.5: Comparison between single and multiplexed fibre DOWEC rev1 Page: 10 of

11 4.2 Dynamic testing In the dynamic testing two load amplitudes were used: -5 to -30 kn and -5 to -60 kn. The results of the optical fibre sensors were stored in the "black-box" of the optical fibre strain sensing system. Due to the internal capacity of this system the disk was full after 7 minutes (86 Hz of measurement frequency). Therefore the dynamic tests only ran for 5 minutes to avoid overwriting of old data. In the second dynamic test the initial load was already -5 kn and therefore the strain gauge already gave results. Due to the downloading of old results obtained by the optical fibre strain sensing system, the optical fibre sensors gave zero strain at the start of this dynamic test. Therefore the results have to be corrected to obtain results that can be compared. Comparing the data for the dynamic testing was very comprehensive as the measuring frequencies differed (88 Hz vs. 20 Hz). To do a good comparison not the absolute maximums have to be compared as this, but the average values of all the measured maximums. The maximum values may be true for only one cycle. As there was a large amount of results only the first few cycles were compared, which may conclude in larger deviations than expected. These results are listed in table 4.6. F = -30 kn F = -60 kn Optical Strain Diff. Optical Strain Diff. fibres gauges fibres gauges Nr % % Nr % % Nr % % Nr % % Table 4.6: Average maximum values of first few cycles in dynamic testing The results show clearly that in the low stress region (nr 1. and 2.) the results differ mostly more than 10 %. For the high stress region (nr 3. and 4.) the differences are comparable to the differences seen at the static test (3-5 %). Some reasons for the large difference are the following: The influence of the geometry has a larger impact at the low region (tail) than at the high stress region (main spar). Due to the different embedment of the optical fibres they can also sense a portion of e.g. transverse strains. This is likely to occur in highly curved structures (tail). DOWEC rev1 Page: 11 of

12 5 Conclusions & recommendations 5.1 Conclusions On the two locations (high and low strain) the results of the strain gauges are comparable (difference is not more than 10 µε). The results of the optical fibre sensors give a similar behaviour as the strain gauges; only the values are slightly higher. The standard deviation differed between ±2 and ±20 µε with an average of ±10 µε. For lower load levels the differences between strain gauges and optical fibre sensors are severe, especially for the lowest load level. The results for higher load levels give a more constant difference (5 %). Three reasons can be found for this difference: The strain transfer between the optical fibres is different than for strain gauges. The optical fibres are more or less embedded into the structure and the strain gauges are mounted on top by an adhesive. These different types of mounting are likely to give different results. Due to the embedment of the optical fibres they can also sense a portion of e.g. transverse strains. This is likely to occur in highly curves structures. The optical fibres are calibrated by a factor of 2.66 on the raw data. This is used by Smart Fibres as calibration for the raw data in their experience. For this structure or embedment the factor can be adjusted to fine-tune the results. The difference between the single and multiplexed fibre in the low strain area (nr 1. and 2.) is more than at the high strain area (nr 3. and 4.) but still comparable. This is cause by the fact that the sensors are located 10 mm apart, which is also seen at the results of the strain gauges. The difference however between the two optical fibre sensors is approximately 3 % higher than the difference of the strain gauges at the same location. For the lower loading levels (especially in the low strain area) the differences are the greatest (10-12 %). The specifications of the Smart Fibres system list a resolution of ±10µε so it will not be surprisingly that for a low strain level the same absolute magnitude of fluctuation gives higher deviations in percentage than for higher strain levels. Looking at the absolute values it can be seen that the difference at the low stress region is less than at the high stress location. In the dynamic testing the results in the low stress region (nr 1. and 2.) differ more than 10 %. For the high stress region (nr 3. and 4.) the differences are comparable to the differences seen at the static test (3-5 %). Some reasons for the large difference are the following: The influence of the geometry has a larger impact at the low region (tail) than at the high stress region (main spar). Due to the different embedment of the optical fibres they can also sense a portion of e.g. transverse strains. This is likely to occur in highly curves structures (tail). 5.2 Recommendations This demonstration shows that optical fibres give comparable results for strain gauges and optical fibre sensors. Therefore recommendations will be given for further research if they are going to be used for load registration in a rotor blade. Some of those recommendations are the following: Find out the accuracy of both types of measurements. How much influence has the structure on the measurements of the optical fibres and do transverse strains have an influence also; How is the difference when the fibre is completely embedded in a structure and not almost at the top (internal strains); DOWEC rev1 Page: 12 of

13 How has embedding to be done using vacuum infusion and how can the location be fixed; How can optical fibre sensors be used in a test; Can optical fibres be used as production monitoring? These are only a few recommendations but more will certainly exist. For each partner in the DOWEC project however the priority will be different e.g. a blade manufacturer is more concerned about the embedding process, but a research institute will be more interested in the signal reading of the optical fibres and test facilities are willing to know if they are reliable enough for test results. DOWEC rev1 Page: 13 of

14 Appendix A: Strain gauges vs. Optical fibres (pos. strains) DOWEC rev1 Page: 14 of

15 DOWEC rev1 Page: 15 of

16 Appendix B: Strain gauges vs. Optical fibres (neg. strains) DOWEC rev1 Page: 16 of

17 DOWEC rev1 Page: 17 of

18 Appendix C: Single vs. multiplexed optical fibres DOWEC rev1 Page: 18 of