Subsidence Monitoring with a Survey Network: Analysis of Network Layout Presented to: National Associate of Abandoned Mine Lands Program Co-authored by: Jeffrey Riedel P.E. (Pioneer Technical Services, Inc) Claire Rasmussen M.S. (Pika Statistical Consulting) 9/26/2016 Introduction • Origin of Work • Part of Red Lodge Subsidence Investigation • Objective – Determine whether historic underground coal mines have any potential for causing (or have caused) ground subsidence. • In response to concerns from several citizens • Abandoned Mines Lands Section, Montana Department of Environmental QualityRemediation Division. • P.M. Bill Snoddy (AML DEQ) • P.M. Tim Ranf (Pioneer Technical Services) Project Background • Red Lodge, Montana Geology and Mining • Tongue River Member of Ft. Union Formation • Red Lodge Coal Field • 1882 thru 1932 • Room and Pillar • 32 square miles • Number of seams and thickness • • • • • • • • • • • No 1 No. 1 ½ No 2 No 3 No 4 No 4 ½ No 5 No 6 No 7 No 8 No 9 11 Feet 5 Feet 8 feet 10 Feet 10 Feet 5 Feet 8 Feet 6 Feet 2 Feet 25 Feet 10 Feet Geology and Mining • Red_Lodge_Coal_Workings_highdef.mp4 Analysis Findings • Seams beneath Red Lodge • Mine depths range from 100 to 600 feet below ground surface • Up to four seams mined in study area • Existing openings range between 6 feet in height and collapsed, average opening is 2 feet. Analysis Findings • Trough Subsidence Estimates 3 – 7 inches • Maximum vertical settlement is estimated between 3-7 inches • Subsidence may accumulate over the next 50 – 100 years • Potential subsidence rates of 0.003 - 0.012 feet per year. • Structural Damage may be slight to appreciable over this time frame (Provide reference) JOB TITLE : Y-Displacement Contours (*10^3) 5.800 FLAC (Version 7.00) LEGEND 5.600 22-Sep-16 11:02 step 117911 -5.556E+02 <x< 5.556E+02 4.714E+03 <y< 5.825E+03 Y-displacement contours -1.20E+00 -1.05E+00 -9.00E-01 -7.50E-01 -6.00E-01 -4.50E-01 -3.00E-01 -1.50E-01 0.00E+00 1.50E-01 5.400 5.200 5.000 Contour interval= 1.50E-01 4.800 Pioneer Technical Services Bozeman, JJR -4.000 -2.000 0.000 (*10^2) 2.000 4.000 Big Picture • Is the ground is subsiding? • Survey 63 points in study area once every 6 months for 2 years • Crux: Can’t just look at points and say definitively if there is subsidence • Measurement error is on the order as the subsidence rate we are trying to detect • Using leveling networks and linear regression will help address this problem Big Picture • Use leveling network to minimize error for each round of the survey • Repeat network measurement on a semi-annual basis • Use linear regression to estimate movement rate at each monitoring point. • Use regression results to make statistical inference about whether each monitoring point is subsiding Survey Design and Accuracy of Estimation • Building more redundancy into the survey design results in more accurate estimates • When using the leveling network method we get: • An elevation estimate for each point in the survey • A Standard Error associated with each elevation estimate Survey Design • The Standard Error is a measure of how variable the elevation estimate is • A small SE means we would likely get a similar estimate if we were to immediately repeat the survey • How do we ensure our survey will produce estimates accurate enough to be useful? • We can use computer simulations to analyze the SEs of our survey points before actually performing the survey • We can compare different surveys to help decide where to include redundancies Survey Design Example Monitoring Point Elevation 1 Elevation 2 Elevation 3 Elevation 4 1 5580.723 5580.454 5580.448 5580.219 2 5581.224 5581.223 5581.219 5581.223 . . . . . . . . . . . . . . . Example - Continued • H0: True Slope = 0 vs. HA: True Slope < 0 • Can make a decision using the p-value from a t-test • Can estimate a confidence interval for the true slope Example - Continued • p-value = 0.025 • This means there is approximately a 2.5% chance of observing a random sample of survey elevations that produce a slope as or more negative than this one, if this point is not actually subsiding • Slope estimate = -0.152 • Associated 95% confidence interval -∞ -0.024 0 • We are 95% confident that the true rate of subsidence of monitoring point 1 is greater than or equal to 0.024 feet per 6-month period Example - Continued • p-value = 0.296 • This means there is approximately a 29.6% chance of observing a random sample of survey elevations that produce a slope as or more negative than this one, if this point is not actually subsiding • Associated 95% confidence interval 0 0.003 -∞ • We are 95% confident that the true rate of subsidence of monitoring point 2 is less than or equal to -0.003 feet per 6month period Summary • Red Lodge Subsidence Investigation • Possible subsidence on the order of 3 - 7 inches over the next 50 - 100 years • Is subsidence occurring? • Crux: • Estimated subsidence rate is small relative to measurement error • Solution: • Use leveling network to minimize survey error • Simulate network to ensure selected survey provides estimates with small enough error to be useful • Obtain confidence interval for subsidence rate at each monitoring point. • Confidence interval • Contains 0, no statistical evidence of subsidence • Negative, statistical evidence of subsidence • LONG STORY LONG: Accurate Survey and Quantified Answer Question or Comments? • Thank You
© Copyright 2026 Paperzz