Selection of an Optimum Truck and Shovel Fleet Based on Effective Flat Haul and Capacity Constraint Modelling S Campbell1 and P Hagan2 ABSTRACT A study was undertaken using various methodologies to determine the optimum loading and hauling configuration for different haul routes at Callide Mine, a coal mine located in central Queensland. The study involved development of an effective flat haul model that was used to estimate production rates in truck and shovel circuits that employ two excavators of differing capacities then Queuing Theory to construct a capacity constraint model for scheduling purposes to assist in truck allocation and finally a cost model to determine the most cost effective fleet configuration. The model was calibrated using haul road design and performance data of trucks and excavators currently in use at the mine. The study examined the effects of cycle time for different truck and excavator fleet configurations on production rate and costs. The study found haulage cost was directly related to cycle time while production rate reduced with cycle time. The study indicates the approach taken in combining a cost model with a capacity constraint model offers a simple and effective method of scheduling an optimum number of trucks in a haulage circuit. INTRODUCTION DESIGN OBJECTIVES Optimum design of a truck and shovel fleet is an important aspect in the design of the materials handling process at any mining operation. The main design variables in the process include number and capacity of the dig units and trucks and the layout of the haulage circuit. Truck allocation is important as it can result in higher costs and/or reduced production output. During the planning phase of a mine, scheduling an excavator and fleet of trucks can be difficult when there isn’t detailed knowledge about the performance of trucks in a proposed haulage circuit. Callide Mine, a coal mine operated by Anglo American located in central Queensland, has a homogeneous fleet of trucks that service an excavator in a material handling system removing overburden material. A combination of data over a two-month period was gathered from field measurements and the site-based equipment data recording software were used in the study. The purpose of this study was to investigate the various approaches that could be used to estimate and compare the performance of loading and hauling options in different haulage circuits to derive an optimum fleet design. The method was used to predict the performance of loaded and unloaded haul truck for different haul road designs having differing haulage lengths, grades of haul ramps and position of corners and bends based on an effective flat haul model leading to an estimation of the cycle time for a particular haul route configuration. Once cycle time was estimated then queuing theory was used to estimate production rates and finally operating costs for the different trucking configurations were estimated. The paper outlines an approach based on first principles that can be applied to similar types of mining operations where access to commercially available software might not always be possible or desirable. Filtering of data based on statistical analysis was used to discard unreliable data points from the data set. An effective flat haul model was developed to calculate cycle times for different haul truck configurations. Design graphs were used to predict dig rates and cost of truck fleets. Data was obtained for the following equipment that is currently in use at the mine for prestrip overburden removal: • • • 14 × Hitachi EH4500, 250 t capacity rear dump trucks 2 × Hitachi EX3600, 22 m3 bucket capacity hydraulic excavators 1 × Hitachi EX5500, 27 m3 bucket capacity hydraulic excavator. METHODOLOGY Effective flat haul An effective flat haul (EFH) model was constructed based on the speed that a haul truck will travel at in various haulage scenarios. The scenarios considered included straight runs, bends and corners as well as cases for loaded, unloaded 1. SAusIMM, Student, School of Mining Engineering, University of New South Wales, Sydney NSW 2052. Email: scotty9021@hotmail.com 2. FAusIMM, Lecturer, School of Mining Engineering, University of New South Wales, Sydney NSW 2052. Email: p.hagan@unsw.edu.au EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 19 S CAMPBELL AND P HAGAN and operating on different grades. The results were used to determine an EFH factor for the different scenarios. Data from field time studies was collected for various trucks and truck operators. This included recording truck speeds on different grades, road configurations and times in each of the phases in the haulage cycle as shown in Figure 1. Results for the field measurements were calibrated against values recorded in MinVu, the site equipment performance database, to determine average times for load and dump. FIG 1 - Phases of the shovel truck system (after Ercelebi and Bascetin, 2009). Match factor The productivity of a truck and shovel operation is mutually dependent on the truck and shovel configuration in a fleet. If one of these factors is altered than the overall production can be enhanced or reduced from the optimum rate. Circuits are affected by the occurrence of queuing, idling and bunching. bunching in circuits is a major factor that isn’t accounted for by the match factor. Also it does not account for production targets or cost parameters hence additional methods need to be considered. Capacity constraints The capacity constraint model is a useful theoretical approach based on average load time and changing total cycle time which was noted in some instances by Najor and Hagan (2007) to result in a significant reduction in overall truck productivity and material movement rate. They further commented that by accounting for capacity constraints, mine schedules could better reflect actual fleet capacity. It can be used to determine the productivity of the circuit based on the excavator and number of haul trucks. The total haul cycle time was calculated by adding all four phases in the haulage cycle as per the approach of Ercelebri and Bascetin (2009). To simulate the haulage process required appreciating an excavator had a limiting dig rate. Depending on the haulage circuit, a point could be reached where the addition of another truck will have no effect on increasing the overall production output of the circuit. This behaviour has been explained through the use of queuing theory proposed by Ringwald (1987) whereby as more trucks are added to a circuit, the utilisation of a truck at an excavator approaches 100 per cent. Queuing theory provides a probability factor for queuing to occur and this when incorporated with the match factor can result in more reliable fleet optimisation (Burt and Caccetta, 2007). Equations 2 and 3 from Carmichael (1987) were used to predict the production rate of a circuit based on the number of allocated trucks to an excavator: -1 Pt = 1 - ; (r) i E /in= 0 (n n! - 1) ! (2) Match factor is a method of determining the optimum number of trucks required to service a shovel. The mathematical expression for match factor in relation to homogeneous trucks is given in Equation 1: P = μPtCt Pt = truck availability at excavator, per cent Match factor = (number of haulers)/(number of loaders) × (loader cycle time)/(hauler cycle time) P = production rate, bcm/h n = number of trucks allocated to a circuit r = loading time as a percentage of total cycle time μ = service rate of the excavator based on it not having to wait for a truck (60/excavator loading time) i = integer from zero to n Ct = truck payload, bcm (1) Rai (2000) stated a match factor of one would mean that shovel productivity is equal to truck productivity and that a value of less than one would mean that a circuit is shovel-rich leading to under-utilisation and idling of the shovel. If the value is greater than one then the circuit is truck-rich resulting in overloading the circuit with trucks reducing the efficiency of the circuit through truck bunching of trucks. This makes it important to attain a match factor as close to one as possible. A deficiency of the match factor is that it does not compensate for queuing, heterogeneous trucks and loaders nor explore the consequences when the match factor is not equal to one. Burt and Caccetta (2007) recognised that bunching of faster trucks behind slower trucks is a common problem with many truck fleets which is exacerbated in heterogeneous fleets of differing truck size and performance the end result being an increase in average cycle time. Bunching and queuing are two variables that vary randomly being controlled by such factors as operator experience, equipment quality, equipment age and general maintenance. Using a match factor is a simple method to test the theoretical efficiency of a circuit when comparing the ratio of trucks to shovels. Match factor can be usefully applied to both homogeneous and heterogeneous circuits. Queuing and 20 (3) where: RESULTS Load and dump times Statistics on the dump time of the EH4500 haul trucks are summarised in Table 1. The values shown were obtained from the MinVu database based on a two-month sampling TABLE 1 Average dump and loading times for the haul truck and hydraulic excavators. Dump EH4500 Load EX3600 Load EX5500 Number of loads 8581 793 2023 Sample size 8581 793 2023 Mean load time 93 s (1.55 min) 242 s (4.03 min) 178 s (2.97 min) Standard deviation 24 s (0.40 min) 40 s (0.66 min) 27 s (0.45 min) EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 SELECTION OF AN OPTIMUM TRUCK AND SHOVEL FLEET BASED ON EFFECTIVE FLAT HAUL AND CAPACITY CONSTRAINT MODELLING period involving four dump locations. The mean dump time of 93 seconds is in close alignment with actual times recorded in field measurements having a mean of 82 seconds based on one dump site. The dump time distribution for the trucks is shown in Figure 2. The loading times for the two hydraulic excavators EX3600 and EX5500 are also listed in Table 1 with mean loading times for each of 242 seconds and 178 seconds with a difference of 36 per cent. The loading time distributions for the two excavators are shown in Figures 3 and 4 respectively. From the field measurements it was noted there was a limit on how quick a truck could dump and an excavator could load and this was taken into account when filtering the dataset from MinVu. Due to misreadings, outliers were eliminated such that for example loading times by the EX5500 of less than 150 seconds and truck dump times of less than 50 seconds were deemed too fast. Effective flat haul model When developing an EFH model, one critical factor that needed to be established was the average speed a haul truck could attain on different gradients and road conditions (straight, bend or corner) as well as truck load condition (loaded or unloaded). Different EFH factors were applied for each scenario to derive an equivalent haulage distance. The EFH factors used in the model for the loaded and unloaded trucks are given in Table 2, while Table 3 shows the adjusted truck speeds for different gradients and road conditions. The factors are normalised against a truck travelling straight on the flat at 60 k/h. An actual haulage circuit at the mine was used to verify the EFH values, the circuit shown in Figure 5. Values stated in the circuit are the grades for a loaded truck travelling to the dump point. The calculated values of EFH for the various road conditions in the loaded and unloaded cycle of the circuit are shown in Table 4. TABLE 2 Effective flat haul factors used for loaded and unloaded EH4500 haul truck. Gradient (%) FIG 2 - Dump time distribution curve for the fleet of EH4500 haul trucks. Straight Bend Corner 0-4 1.5/1.0 1.5/1.0 3.0/3.0 4-8 1.7/1.3 1.7/1.3 3.0/3.0 8 - 12 4.6/2.0 4.6/2.0 3.0/3.0 -(0 - 4) 1.5/1.0 1.5/1.0 3.0/3.0 -(4 - 8) 1.7/1.3 1.7/1.3 3.0/3.0 -(8 - 12) 3.0/1.7 3.0/2.4 3.0/3.0 Note: first value is loaded condition and second value is unloaded condition. TABLE 3 Variation in maximum truck speed for different road conditions. Gradient (%) Straight (km/h) Bend (km/h) Corner (km/h) Uphill loaded average speed 0-4 40 40 20 4-8 35 35 20 8 - 12 13 13 20 Downhill loaded maximum speed FIG 3 - Load time distribution curve for the EX3600 excavator. 0-4 40 40 20 4-8 35 35 20 8 - 12 20 20 20 Uphill unloaded maximum speed 0-4 60 60 20 4-8 45 45 20 8 - 12 30 30 20 Downhill unloaded maximum speed 0-4 60 60 20 4-8 45 45 20 8 - 12 35 25 20 Sharp bend FIG 4 - Load time distribution curve for the EX5500 excavator. All EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 30 21 S CAMPBELL AND P HAGAN TABLE 5 Comparison of different phases in truck cycle times based on calculated effective flat haul model and measured times. Phase in truck cycle (min) Load FIG 5 - Haulage route showing varying grades. TABLE 4 Variation in effective flat haul calculated values for haulage route under different road conditions. Straight uphill Loaded haul Straight downhill Sharp bend Grade (%) Distance (m) EFH (m) 0-4 942 1413 4-8 477 818 8 - 12 293 1352 0-4 996 1494 Sharp bend Total Total time (mins) Total Measured time 2.97 6.58 1.37 4.67 15.59 EFH model 2.97 6.38 1.55 4.13 15.03 reviewing the different EFH contributions to haulage cycle time it can be seen that the graded ramps in particular the steep eight to 12 per cent grade ramps despite their short distance and to a lesser extent the corners and curves. The impact that these had on the effective flat haul ranged from a factor of two times through to 4.6 times longer duration. The model showed for example that a loaded truck travelling 293 m up an eight to 12 per cent graded ramp was equivalent to travelling 1.494 km on the flat at 60 km/h, which when compared to the same loaded truck travelling on a four to eight per cent ramp over nearly twice the distance of 477 m had a much shorter EFH of 818 m at 60 m/h. This shows that limiting the amount of steep ramps when designing haul roads can have a much larger impact on cycle time than making ramps steeper to shorten haul distances. Using the EFH model approach can facilitate calculation of the haulage cycle time in any circuit that can in turn be used to determine the optimum number of trucks in a circuit servicing an excavator. This will allow greater flexibility in knowing when to use a certain excavator and how many trucks that are required on that circuit to service the excavator. 4-8 445 763 80 240 All 150 300 3383 6380 Fleet match factor 6.38 A fleet match factor is a simple method of estimating the number of trucks required in a circuit to minimise shovel idle time and maximise truck utilisation. In order to account for the variability observed in load times, calculation of the fleet match factor was based on the summation of values stated in Table 1 for average load time and one standard deviation which is equivalent to 84 per cent of occurrences. 0-4 996 996 4-8 445 593 8 - 12 80 160 0-4 942 942 4-8 477 636 8 - 12 293 502 All 150 300 3383 4130 4.13 When load and dump times given in Table 1 are combined with the travel time in Table 4, the total cycle time can be determined. Table 5 contains the times for each phase in the cycle as a well as the total cycle time, which in the case of the EX5500 excavator was 15.03 minutes, which compares favourably with the field time measurement of 15.59 minutes, representing a difference of only 3.7 per cent. This difference may be partly explained by the model not accounting for truck acceleration times. The results of the EFH model show that actual haul length was not the only major contributor to cycle time. When 22 Return 8 - 12 Total time (mins) Unloaded haul Straight downhill Dump Using the EFH model provides an approach to the calculation of the haulage cycle time in any circuit that can in turn be used to calculate the optimum number of trucks in a circuit to service an excavator. This approach provides some flexibility in knowing when to use a certain excavator and how many trucks are required on that circuit to service the excavator. Total Straight uphill Haul Two sets of results were determined for each excavator. The first being the number of trucks to service an excavator for various cycle times. The second being the fleet match factor. Results for the EX5500 excavator are shown in Figures 6 and 7 with similar trends found for the EX3600. The fleet match factor tended to stabilise around one for longer cycle times. A deficiency in using the fleet match factor approach is that it fails to account for a targeted mining rate as well as impact on mining costs. While the fleet match factor may ultimately reduce bunching and queuing it may not achieve a required production rate. Where this is important then a further approach is required. Capacity constraint model While the deterministic approach of the EFH model may be useful in the first instance to determine optimum fleet configuration, the variability inherent in the various phases of loading and hauling process will often lead to significant EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 SELECTION OF AN OPTIMUM TRUCK AND SHOVEL FLEET BASED ON EFFECTIVE FLAT HAUL AND CAPACITY CONSTRAINT MODELLING FIG 6 - Required number of trucks to service the EX5500 excavator for various cycle times. FIG 7 - Calculated fleet match factor for the EX5500 excavator. FIG 8 - Capacity constraint model based on seven or fewer trucks servicing an EX3600 excavator. FIG 9 - Capacity constraint model based on eight or more trucks servicing an EX3600 excavator. underperformance of the haulage system. The capacity constraint model is a stochastic approach that recognises the random nature of elements in the process. It is based on a probabilistic function that can be used to effectively allocate trucks to a haulage circuit. Mine planners often aim to allocate a sufficient number of trucks in a circuit to achieve a target production rate. Using a combination of Equations 2 and 3, the load times in Table 1 and assuming an average truck capacity of 114 bcm, a capacity constraint model for different total cycle times ranging from five to 40 minutes was generated. The model showing the variation in the production rate based on the EX3600 excavator in a circuit with varying truck numbers and cycle times are given in Figures 8 and 9. For the case of the EX5500 excavator, the capacity constraint model is given in Figures 10 and 11. The results show with shorter cycle times, fewer trucks are required to achieve a target dig rate. A point will eventually be reached when allocating tracks to a circuit that it will become saturated with a ceiling reached in production rate. In the case of the EX3600 and EX5500 the estimated maximum dig rates are 1700 and 2300 bcm/h respectively, each requiring a minimum of seven trucks for this haulage circuit with cycle times of five minutes or less. As one objective of mine planning is to achieve a target dig rate, a combination of the EFH model and the capacity constraint model provides a quick and simple method to estimate the production rate in a circuit that is only in the planning phase. For example, if the required dig rate was 1600 bcm/h in a haulage circuit having a cycle time FIG 10 - Capacity constraint model based on seven or fewer trucks servicing an EX5500 excavator. of 20 minutes then based on Figures 8 and 9, eight trucks would be required to service an EX3600 excavator compared to only six trucks for an EX5500 excavator. Alternatively if only five trucks were available to service and excavator then the maximum production rate that could be achieved for the EX3600 and EX5500 would be approximately 1200 and 1400 bcm/h respectively which is equivalent to 67 per cent and 59 per cent of the respective maximum potential capacity in the circuit. Whereas if the excavators were over-trucked having 14 trucks in a haulage circuit then the maximum potential productivity of 1700 bcm/h for the EX3600 could EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 23 S CAMPBELL AND P HAGAN TABLE 6 A comparison of operating costs for various equipment types relative to haul truck costs. Equipment FIG 11 - Capacity constraint model based on eight or more trucks servicing an EX5500 excavator. be sustained with truck cycle times less than 30 minutes but the 2300 bcm/h productivity for the EX5500 could only be sustained for truck cycle times less that 20 minutes. Costing While achieving an efficient high production circuit is important in terms of equipment utilisation, cost is another important factor that requires consideration. The costing models for the EX3600 and EX5500 excavators shown in Figures 12 and 13 respectively, are based on the relative operating costs provided in Table 6. These models show Cost factors EX5500 excavator 50 EX3600 excavator 30 EH4500 haul truck 1 the variation in unit costs with haulage distance having equivalent cycle times of ten, 20, 30 and 40 minutes based on the number of trucks servicing an excavator. For the shortest haulage route having a cycle time of ten minutes, the optimum number of trucks to achieve the minimum unit cost is approximately 3 - 4 trucks for the EX3600 excavator whereas it is 5 - 6 trucks for the EX5500 excavator. This is explained by more trucks being required with the larger productive excavator to ensure it remains fully trucked up. Beyond this optimum number, unit mining costs increase at a greater rate with the EX3600 as more trucks are needed to be added to the circuit compared to the larger excavator EX5500 which is less sensitive to increases in truck numbers. As haulage distance and hence cycle time increases, the optimum number of trucks increases as does mining cost though the difference narrows between the two excavators as excavator cost becomes a less dominate cost factor. While it is important to achieve a target production rate, the most economical number of trucks allocated to a circuit may not necessarily achieve a required production rate. When combining the capacity constraint model and cost model an overall comparison between the excavators can be made. Figure 14 considers both excavators operating in the same circuit having a 15-minute cycle time indicating the variation in production and costing as additional trucks are added to the circuit. The graph illustrates that the optimum truck fleet size for the EX3600 excavator is four trucks achieving a production rate of 1200 bcm/h. If a higher production rate is required with the same size trucks but using the EX5500, the optimum fleet size is seven trucks achieving a movement rate of 2000 bcm/h, which is 60 per cent more material at the same unit cost. CONCLUSION FIG 12 - The effect of truck fleet size on mining cost for different truck cycle times servicing an EX3600 excavator. It was possible to construct an effective flat haul model that can provide a reasonable estimate of haulage cycle time for FIG 13 - The effect of truck fleet size on mining cost for different truck cycle times servicing an EX5500 excavator. FIG 14 - A comparison of the performance of an EX3600 versus EX5500 excavator based on a 15-minute cycle time. 24 EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 SELECTION OF AN OPTIMUM TRUCK AND SHOVEL FLEET BASED ON EFFECTIVE FLAT HAUL AND CAPACITY CONSTRAINT MODELLING a given haul design scenario that correlated well with actual cycle times. The EFH model was found to be useful in the planning process as it indicated the sensitivity of changes in ramp design (for example gradient versus length) on overall cycle time. A combination of the capacity constraint model and costing model was found to provide a more effective approach in allocating trucks to a circuit as it considered not only cycle time but also production capacity. Keeping cycle times short meant lower cost and higher productive circuits. Fleet match factor was the least useful method in allocating trucks to a circuit as it provided limiting information on the performance of a circuit. Shovel idling and truck availability are best used to gain the best utilisation out of the loader or truck fleet. When comparing the EX3600 and EX5500 excavators in use at the Callide Mine on an identical circuit the smaller EX3600 was more economic for smaller truck fleets size and production rates but the eventual higher productivity of the EX5500 would lower the cost of moving material achieving much higher material movement rates. In the end a choice has to be made between achieving higher production with the EX5500 excavator or a lower cost with the EX3600 excavator in a given haulage circuit. ACKNOWLEDGEMENTS The authors wish to acknowledge Anglo American Metallurgical Coal for allowing access to the data used in this study and to conduct field measurements at the site. They also acknowledge support given by staff at Callide Mine with helping in organising the site visits and giving access to the resources needed to generate this paper. REFERENCES Burt, C and Caccetta, L, 2007. Match factor for heterogeneous truck and loader fleets, International Journal of Mining, 21(4):263-270. Carmichael, D, 1987. Engineering Queues in Construction and Mining (John Wiley and Sons: New York). Ercelebi, S and Bascetin, A, 2009. Optimisation of shovel-truck system for surface mining, The Journal of the Southern African Institute of Mining and Metallurgy, 10(9):433-439. Najor, J and Hagan, P, 2007. Improvements in truck and shovel scheduling based on capacity constraint modelling, in Proceedings Sixth Large Open Pit Mining Conference, pp 87-92 (The Australasian Institute of Mining and Metallurgy: Melbourne). Rai, P, 2000. Optimising shovel-truck combination, Coal International, 5(3):230-231. Ringwald, R, 1987. Bunching theory applied to minimise cost, J Construction Engineering Management, 113(2):321-325. EIGHTH AUSIMM OPEN PIT OPERATORS’ CONFERENCE / PERTH, WA, 18 - 19 SEPTEMBER 2012 25