Real time stress change estimation using strain measurements Stepan Kodeda, Frans Ritala, Topias Siren & Lauri Uotinen Aalto University School of Engineering Department of Civil and Environmental Engineering FINLAND pp. 1071-1076
Outline Team Introduction Kodeda et al. 2015 Real time stress change estimation using strain measurements Outline Team Introduction Concept Synthetic testing In-situ testing Results Conclusions Discussion Stepan Kodeda, Master’s Thesis worker: method & code development Frans Ritala, Master’s Thesis worker: in-situ experiment, data analysis Topias Siren, Doctoral Student Thesis Instructor (Ritala) Lauri Uotinen, Doctoral Student Thesis Instructor (Kodeda), coordinator
Kodeda et al. 2015 Real time stress change estimation using strain measurements Introduction Our goal was to develop a method suitable for real time analysis using strain measurements. Inverse calculation of the stress change is carried out using: unit stress responses superposition principle multiple linear regression CHILE rock conditions are assumed for the rock mass Numerical modelling was used to generate the stress responses
2014 – 2015 DynaMine C – Real time stress change estimation Kodeda et al. 2015 Real time stress change estimation using strain measurements Research Roadmap 2011 – 2016 TEKES Green Mining – Mining Efficiency & Social License to Operate 2014 – 2015 DynaMine – Dynamic Control of Underground Mining Operations 2014 – 2015 DynaMine C – Real time stress change estimation 2014 Kodeda Theory 2015 Ritala Practice This presentation
Kodeda et al. 2015 Real time stress change estimation using strain measurements Concept The actual tunnel geometry and actual sensor locations and orientations are modelled numerically. For each stress component, the unit stress responses is calculated and stored for later use. The actual strain changes are measured over time. The corresponding stress field changes are solved as a multiple linear regression calculation using the stored unit responses.
Kodeda et al. 2015 Real time stress change estimation using strain measurements Unit stress response Unit stress response is the numerically measured response for a load state containing “1” for the relevant stress component. For 2D analysis, three unit responses are needed: 𝜎 𝑥 , 𝜎 𝑦 , 𝜏 𝑥𝑦 𝜎 𝑥 = 1 0 0 0 0 0 0 0 0 , 𝜎 𝑦 = 0 0 0 0 1 0 0 0 0 , 𝜏 𝑥𝑦 = 0 1 0 1 0 0 0 0 0 =
Superposition principle Kodeda et al. 2015 Real time stress change estimation using strain measurements Superposition principle Any 2D linear elastic stress state is a linear combination of the unit stresses: 𝜎=𝐴∗ 𝜎 𝑥 +𝐵∗ 𝜎 𝑦 +𝐶∗ 𝜏 𝑥𝑦 . Multiple linear regression is used to obtain the multipliers 𝐴, 𝐵 and 𝐶. This results in best fit stress tensor. The obtained generic 2D stress state can be converted to the principal stress state 𝜎 1 , 𝜎 3 , 𝛼 where 𝛼 denotes the angle.
Kodeda et al. 2015 Real time stress change estimation using strain measurements Design assumptions The current method requires homogenous continuum. Especially any joints crossing through a sensor hole must be included in the model as they concentrate movements. Linear elasticity is required. Plasticity can be added later, but it breaks the mathematical commute property and introduces sequencing of the combination. Isotropy was used. The method is compatible with anisotropy if included in the modelling.
Synthetic benchmarking Kodeda et al. 2015 Real time stress change estimation using strain measurements Synthetic benchmarking Four benchmarking cycles were used: 1. blind 2D, 2. open 2D, 3. blind 3D, 4. open 3D. The 2D blind tests show that the method is fairly tolerant of uniform random noise or missing data. Corrupted data or crossing joints can easily be detected from erratic results. The 3D blind test show slightly less tolerance against random noise, but cope equally well with missing data. Again, corrupted data can be detected. The complementary open tests contained additional cases to study the observations in more detail.
Synthetic benchmarking 2D blind round results Kodeda et al. 2015 Real time stress change estimation using strain measurements Synthetic benchmarking 2D blind round results Case Description Estimation result 1-3 different stress state clear data 1: ~100 %, 2: ~100 %, 3: 99.7 % 4-6 noise 1 %, 5 % and 10 % 4: 98.4 %, 5: 92.4 %, 6: 96.0 % 7 clear data, three data points missing ~100 % 8 clear data, six data points missing 9 clear data, one instrument missing (five data points) 99.9 % 10 unknown joint crossing all instruments 24.2 % 11 parallel joint set, not intersecting instruments 74.1 % 12 two instruments swapped 0 %
Selected site: Boliden Kylylahti Kodeda et al. 2015 Real time stress change estimation using strain measurements Selected site: Boliden Kylylahti Established in 2012. Operating depth 600 m. 220 employees. Production 2014: 172 ktons (2546 ton Copper, 335 ton Zinc and 82 kg of Gold). Mining method: open stoping with delayed backfilling. Boliden purchased the research instruments (14 multipoint extensometers) for this study.
Mine installation – Kylylahti Kodeda et al. 2015 Real time stress change estimation using strain measurements Mine installation – Kylylahti Two sites were selected: -322 m 2D site at access tunnel 30 m away from the stope (2 extensometers) -410 m 3D site at close proximity to mining operations (12 extensometers) Only -322 m data was analyzed in pilot stage. Picture: Boliden
Kylylahti 2D site -322 m & 3D site -410 m Kodeda et al. 2015 Real time stress change estimation using strain measurements Kylylahti 2D site -322 m & 3D site -410 m 2D site -322 m 3D site -410 m
Kylylahti 2D site -322 m 25 m Before excavation After excavation Kodeda et al. 2015 Real time stress change estimation using strain measurements Kylylahti 2D site -322 m Before excavation After excavation 25 m
Kylylahti 2D site -322 m results Kodeda et al. 2015 Real time stress change estimation using strain measurements Kylylahti 2D site -322 m results σ1 (MPa) σ3 (MPa) θ (°) 27.12.2014 0.00 31.12.2014 -11.96 -33.42 20.78 1.1.2015 5.1.2015 -2.21 -10.03 86.74 12.1.2015 136.24 46.99 33.83 22.1.2015 122.27 42.85 33.55 9.3.2015 897.45 320.78 33.03
What went wrong? Plasticity? Rock damage process? EDZ? Kodeda et al. 2015 Real time stress change estimation using strain measurements What went wrong? Plasticity? Rock damage process? EDZ? Unknown rock joints in the area? 3D effects? Not plane strain? Preceding mining activities? Equipment malfunction? Installation errors? Reading errors? …
Kodeda et al. 2015 Real time stress change estimation using strain measurements Conclusions The algorithm performs well for synthetic cases and tolerates noise and missing data. The method has problems coping with instrument crossing joints. In-situ pilot experiment was carried out, but the results are inconclusive as the data is corrupted. The corruption could be caused by plasticity or by presence of undetected joints. After precalculated unit stress responses, the method calculation speed is 20 us (50 Hz) and it is suitable for real time analysis.
Kodeda et al. 2015 Real time stress change estimation using strain measurements Discussion The method shows potential and can be validated, but it needs verification and further in-situ testing. This can be a simple, clean, in-situ experiment or a well documented case history. Plasticity, rock mass damage and fractures will occur near large mine stopes. The method needs further developing to account for these factors. By adding mining sequence the original in-situ stress state can be solved from the time history and a known distribution of strains.
Thank you! Any questions? Kodeda et al. 2015 Real time stress change estimation using strain measurements Thank you! Any questions? Contact us: lauri.uotinen@aalto.fi +358 50 569 1015 Did you enjoy the presentation? Vote for us with the Eurock 2015 app! This research was funded by:
Rock mechanical parameters Kodeda et al. 2015 Real time stress change estimation using strain measurements Rock mechanical parameters Rock unit UCS [MPa] E [GPa] v ρ [kg/m3] KAL 155 60 0.35 2800 MS-SMS 140 85 0.32 3300 OME 95 90 2900 OUM 150 75 0.33 Soapstone 35