Lean Mining. A solution for sustainable Development. U. Kumar B. Ghodrati

Transcription

1 Lean Mining A solution for sustainable Development U. Kumar B. Ghodrati

2 What is Lean mining?

3 LEAN MINING PRACTICES brings effectiveness & efficiency in operation by eliminating uncertainty that often leads to wastage of resources: Toyota Production System LEAN ( JIT, TQM, TPM, & KAIZEN)

4 Comparison of Mining and Automotive industry Mining industry Physically challenging environment Inherently variable environment Geographically spread output teams Inherently variable raw materials Remote locations Automotive industry Ambient environment Stable work environment Compact plants Controlled raw materials Large centers Those similarities bring an opportunity to successfully apply lean principle into mining industry. In contrast to innovation approach which emphasize on a quick improvement/change, lean principle is a continuous improvement approach which emphasize on a small but constant improvements.

5 In contrast to innovation approach which emphasize on a quick improvements or STEP CHANGES, lean principle is a continuous improvement approach which emphasize on a small but constant improvements.

6 Lean Mining One approach to apply the lean principle in mining industry is to apply the concept of Overall Production Effectiveness to eliminate waste and increase the operational reliability, production quality and performance through engagement of all personnel. ELIMINATION OF UNCERTAINTY

7 CAMM : LEAN MINING The following four subprojects are considered in our lean mining project: 5 Seniors & 4 PhD Students 1 Improvement of production availability and delivery assurance 2 Rock and machine interface 3 Remaining useful life of mining systems 4 Integration of mine work environment into production systems

8 Production assurance concept

9 1.1 WHAT IT IS? Production assurance models describe to what extent a system is capable of meeting demand for deliveries.

10 1 Introduction to concept Production assurance WHY??? Uncertainty in operation is the main cause for delays, customer dissatisfaction and other waste of resources.

11 HOW? Models are being developed & Tested to make correct decision in order to meet customers requirements in term of volume and quality at short notices.

12 1.2 Production Assurance (PA) The PA concept includes several other concepts such as reliability, maintainability, availability, supportability, capacity.

13 1.3 Production assurance: Emperical Model PA can be described and calculated by combinations of capacity performance and system availability performance for period of (t1, t2) PA = Mean predicted operational availability Planned operational availability t 2 t 1 ( capacity performance) dt t 1 t 2

14 1.4 Production Assurance Case study Consider a simple numerical example of a transportation unit of an underground mine producing iron ore; it consists of three LHD which can be configured as a parallel system: LHD 1 LHD 2 LHD 3

15 1.4 Production assurance Case study The failure of LHDs which consists of different and several components follows Poisson distribution. In this case, the reliability of transportation system (LHD fleet) at time t is: Where, R( t) = exp( λt) λ = Failure rate of LHD N t = Operation time of system ( λt) r= 0 r! N= Total number of required LHD in period t r

16 Each LHD is assumed to have a throughput capacity of either full (c) or 0. Therefore, the possible capacity performance levels for transportation unit are thus 3c, 2c, c, and 0. The planned production rate for transportation unit is: ton/day The Markov diagram for three LHDs in parallel is: 3λ 2λ λ 3c 2c c 0 µ 2 µ 3µ

17 Characteristics of LHDs System Failure rate Repair rate System working capacity System design capacity Capacity performance LHD 3,75 x ,96 x ton/day ton/day 80% Probability distribution of system capacity by Markov model System capacity level (%) Probability 0 3,843 x ,651 x , ,9891

18 Challenges CONTEXT DRIVEN PRODUCTION ASSURANCE MODELS CONSIDERING LOCAL AND GLOBAL BUSINESS RISKS

19 3 Remaining Useful Life (RUL) The remaining useful life (RUL) of the unit indicates its ability and length of surviving in operation in the future. RUL is estimated based on the physics of the failure and statistical analysis.

20 Remaining Useful Life Expected Performance Acceptable Limit Performance x 1 x 2 Degradation starts P 1 P 2 P 3 X 1, X 2 X 3 : Remaining useful life of system based on different degradation (P1, P2, P3) rate x3 Time Luleå University of Technology

21 REMAINING USEFUL LIFE Challenges Remaining useful Life of a Component Remaining useful Life at a System level Remainimng Useful Life at an Asset Level System of System level

22 Remaining Useful Life (RUL) Performance Expected Performance Acceptable Limit P 1 P 2 Time

23 Degradation Behaviour of COMPONENTS IN A SYSTEM T NOM Wear depth (mm) Maintenance Thresh hold limit Safety Limit ASME B MGT/ Age (Years)

24 Data Segregation Components with no degradation T NOM Suspensions Thickness (mm) Maintenance limit Safety Limit Failures ASME B MGT/Age (Years)

25 EQUIPMENT Design & Dev phase Function & Performance Application Environment RAMS, LCC & Risk analysis Safety, Environment, Sustainability, ROI Maintenance & Service program Integrated LEAN Solutions Cost Effective Product Development & Life Cycle Management LIFE CYCLE MANAGEMENT

26 OPERATING ENVIRONMENT BASED MODELS λ( t, z) = λ t z = 0 ( )exp( α) λ0 ( t)exp( α i z i ) R( t) = R0 ( t)exp( α i z i ) n i= 1 n i= 1 β 1 β t n λ( t, z) = exp( α η i= 1 η i z i ) Division of and aintenance MTTF = η Γ(1 + 1 ) β 26-Nov-12 ICQRITTM New Delhi 26

27

28 Probability of failure Time (lifespam) Population that require maintenance Average life Population survive longer than average

29 Most of the items fail around average life with few living longer or needing early repairment Probability of failure Time (lifespam) Population that require maintenance Average life Population survive longer than average

30 Operation phase Condition Monitoring WHAT? Diagnosis WHY? Explaining Describing System state & behavior Prognosis When? Predicting How? Safety, Environment, Sustainability, ROI Controlling Integrated LEAN Solutions Effective Asset & Production Management

31 Challenge FUSION OF QUALITATIVE DATA WITH QUATITATIVE & EXPERIENCE DATA

32 What is Remaining Useful Life (RUL)? t0 t T=0 T1 T2

33 RUL Case study Variable p - Value Final Model Step0 Step1 Step2 Estimate S.E. OPSK Operator skill (OPSK) DUST Maintenance crew skill (MCSK) TEMP Hydraulic oil quality (HOILQ) HOILQ Hydraulic system temperature (TEMP) MCSK Environmental factors (DUST) The effect of three covariates (OPSK, DUST and TEMP) is significant at 10% p-value.

34 Actual Hazard Model λ(t, z) = λ0 (t) exp(-1.201opsk DUST TEMP) λ( t) = = β t ( ) η η β t ( ) η η β 1 β 1 exp( n j = 1 α z j j ) exp(-1.201opsk DUST TEMP ) Covariates Existence Situation (OPSK, DUST, TEMP) State Value State Value S1 (1,1,1) S5 (-1,1,1) S2 (1,1,-1) S6 (-1,1,-1) S3 (1,-1,1) S7 (-1,-1,1) S4 (1,-1,-1) S8 (-1,-1,-1)

35 Actual Reliability Function Age (t) R(t) at different stages of existing covariates and times State 1 State 2 State 3 State 4 State 5 State 6 State 7 State ,00 1,00 1,00 1,00 1,00 1,00 0,99 0, ,00 1,00 0,99 0,97 1,00 0,98 0,93 0, ,00 0,99 0,95 0,79 0,97 0,86 0,57 0, ,99 0,96 0,84 0,46 0,89 0,60 0,15 0, ,98 0,90 0,66 0,16 0,77 0,30 0,01 0, ,92 0,69 0,25 0,00 0,41 0,02 0,00 0, ,82 0,42 0,04 0,00 0,12 0,00 0,00 0, ,69 0,18 0,00 0,00 0,02 0,00 0,00 0, ,52 0,05 0,00 0,00 0,00 0,00 0,00 0, ,44 0,02 0,00 0,00 0,00 0,00 0,00 0, ,28 0,00 0,00 0,00 0,00 0,00 0,00 0,00

36 Remaining Expected Useful Life Age (hrs) Covariates existing states , , , , , , ,94 522, , , , , , ,67 704,22 220, , , ,93 839, , ,50 413, , , , ,75 566, ,23 810,11 236, , , ,12 131, ,32 429,89

37 Wear and remaining useful life prediction of grinding mill liners

38 Wear and remaining useful life prediction using Neural network Goal To optimize: Mill profitability Wear measurement Replacement and maintenance scheduling by means of estimation of the remaining useful life of mill liners

39 Grinding mill liners

40 Data collection from Metso Mineral Condition monitoring data (height and life)

41 2 Cycle condition monitoring data March Sep Dec March-10

42 Prediction of ANN March Sep Dec March-10

43 Real condition monitoring data vs. ANN prediction March Sep Dec March-10

44 Results of the ANN High degree of correlation between the input and output variables The proposed model is able to approximate the input-output function accurately Neural network found to be very effective in defining a function which was capable of establishing good correlation between the input and output variables.

45 CONCLUDING REMARKS LEAN MINING R & I CHALLENGES Generic Production and Deleverary Assurance Model Context Driven Production Assurance Model Undestanding of Rock Mass and Machine interface

46 LEAN MINING R & I CHALLENGES Eatimation of RUL at System and Asset Lebel Context Driven RUL for Operation Decision based on desparate data Planned obsolence vs Sustainability goals (Equipment/System Suppliers

47