MOBILE EQUIPMENT DATA ANALYTICS - CMIC · 6/5/2019 · MOBILE EQUIPMENT DATA ANALYTICS PROJECT...
Transcript of MOBILE EQUIPMENT DATA ANALYTICS - CMIC · 6/5/2019 · MOBILE EQUIPMENT DATA ANALYTICS PROJECT...
MOBILE EQUIPMENT DATA
ANALYTICSCMIC National Mining
Innovation SummitJune 4th 2019
2
SURFACE MINING ROADMAP
IMPROVE PREDICTIVE ANALYTICS OF EQUIPMENT AND
MINING PROCESS PERFORMANCE
Mobile Equipment Analytics
3
MINE MOBILE EQUIPMENT AT SYNCRUDE
Mobile Equipment Analytics
o In the early 90’s Syncrude began to transition its fleet from Dragline, Bucket-Wheel operation…
4
MINE MOBILE EQUIPMENT AT SYNCRUDE
Mobile Equipment Analytics
o …to the current truck and shovel operation
o ~ 130 Ultra Class Haul Trucks (400t)
o 20 large face shovels
o 300+ mobile support equipment
5
HAUL TRUCK MAINTENANCE STRATEGIES
Mobile Equipment Analytics
o Maintenance costs have increased significantly over the past 10 yearso Labour, consumables and components.
o Maintenance plans have been stagnant o Conservative calendar based scheduling.
o 21 day cycle regardless of utilization, equipment duty and component condition.
o Fleet scheduling challenging – size of fleet, production and resource constraints.
o Operating costs have increased.o Labour, fuel, environmental considerations.
o Capital costs have increasedo Trucks cost more.
6
TIME BASED PREVENTATIVE MAINTENANCE TO
OPTIMIZED CONDITION BASED MAINTENANCE
Mobile Equipment Analytics
o Backgroundo Maintenance intervals calendar based including oil changes and oil sample and testing of
component condition.
o Objective – optimize oil change frequencieso Understand lubricant degradation.
o Validate component benchmarks.
o Can we use this data to improve condition monitoring program.
o Scope – work with Queens University Centre for Advance Computingo Provide data – oil sample analysis, component condition on select Ultraclass Haul Truck
drive train components.
o Identify indicators, build model and determine optimal maintenance program.
Big Data -> Advanced Data Analytics -> Information -> Decisions
7
HAUL TRUCK MAINTENANCE STRATEGIES
Mobile Equipment Analytics
o Opportunityo Reduce services by 10%
o Increase availability ½ to 1%
o Avoid major component failures
8
MOBILE EQUIPMENT DATA ANALYTICS PROJECT
Mobile Equipment Analytics
Big Data -> Advanced Data Analytics -> Information -> Decisions
o Purposeo Identify indicators in oil analysis data to assess the current maintenance program.
o Provide used oil analysis from six ultra class haul trucks – upper powertrain (torque, transmission and engine).
o Identify trends on engine oil change frequency.
o Identify failure modes and leading indicators of failure.
o Understand lubricant life degradation.
o Optimize the program to maximize the equipment productivity capacity.
o Change the current “fixed timing” maintenance schedule to automated scheduling and
reduce maintenance costs.
o Methodo Apply data analytics process.
o Predictive models w/ learning algorithm.
o Time series forecasting model.
9
THE DATA ANALYTICS PROCESS
Mobile Equipment Analytics
Modeling
Cross Industry Standard Process for Data Mining (CRISP-DM)
Business
Understanding
Data
Understanding
Data
Preparation
Evaluation
Deployment
10
EXAMPLE: TIME SERIES FORECASTING MODELS
Mobile Equipment Analytics
11
MULTI-VARIANT TEMPORAL DATA – WHEN AND WHAT
Mobile Equipment Analytics
Sampling IntervalMedian: 350hrs (Min: 190hrs – Max: 558hrs)
Oil Hours on Oil ChangeMedian:1,715hrs (Min:101hrs - Max: 3,976hrs)
12
STATUS AND NEXT STEPS
Mobile Equipment Analytics
o Analysis – confidence model taking formo Influence is calculated as the correlation between each attribute.
o Most influential attributes defined.
o Data problems highlighted.
o Early results indicate:o The majority of our powertrain components were being “under serviced” and the
majority of the engines were being “over serviced”.
o Engine Oil is changed every other sample, not enough data to find if oil would have
remained operational for longer.
o Next steps – enhance the predictive model accuracy:o Data cleansing to fix anomalies.
o Increase data set with more units to improve the statistical model predictability.
o Generate new features with more influence on the decision to change oil.
o Build time series forecasting for oil changes.
13
PATH FORWARD
Mobile Equipment Analytics
o Results to be shared with the CMIC Surface Mining Working Group.
o Expand to demonstration piloto More equipment – more data.
o Plan for commercializationo Engage appropriate vendors.
o Syncrude Digitalization Strategyo World Class Maintenance and Reliability
Management.
THANKYOU!