Open-Pit Mine Truck Dispatching System Based on Dynamic Ore Blending Decisions

Abstract

In the production process of open-pit mines, trucks are applied in the production process of open-pit mines for transporting ores and rocks. Most open-pit mines are equipped with dozens of trucks. It is important to plan the dispatch of trucks in the production process so that the transportation process can be the shortest in distance, the lowest in cost, and the most efficient. At present, many open-pit mining enterprises have realized the use of dispatching systems to schedule trucks to complete production tasks. However, these methods are mostly designed to deploy trucks to reduce production costs without considering the blending problem of the selected ores, and therefore it cannot meet the dual need of ore blending and dispatching. In order to solve the above technical problems and meet the actual needs of the current open-pit mine for ore blending and dispatching, this paper proposes an open-pit mine truck dispatching system based on dynamic ore blending decisions, supported by a 4G/5G wireless network, Beidou positioning, and Internet of Things technology, which can not only realize the optimized truck dispatching of open-pit mine production, but also meet the requirements of downstream concentrators for ore dressing grade. The system has been applied in the Ansteel Group QIDASHAN mine for one year. The proportion of trucks dispatched through the system reached more than 70%. The trucks’ capacity were upgraded from 3.79 to 4 million ton km per set per year, and the efficiency was improved by 5.5%. The limitations of the proposed system and method mainly include the possibility of inaccurate measurement of ore output and the lack of combination with unmanned driving.

1. Introduction

In the process of open-pit mine production, the cost of shovel loading and transportation operation accounts for more than 50% of the cost of the entire open-pit mine production. With the continuous promotion of the construction of smart mines, many enterprises and research institutions have carried out a lot of research work on how to reasonably and efficiently use the shovel loading and transportation equipment to tap the potential of the equipment, and put forward many dispatching methods with positive results.

In the ore blending optimization research, many scholars actively proposed linear programming (LP) to the optimization modelling of production planning and solved many problems in mine blending. Some studies use metaheuristics to improve the production efficiency of the mine and solve some ore blending problems [1,2,3,4,5,6]. Liu et al. [7] proposed an model by LP, which solved the problems of ore blending in iron mines and problems in ore quality control. Boland et al. [8] solved the open-pit mining production scheduling problem based on the method of LP. Lambert et al. [9] who used integer dynamic programming to determine the sequence of extraction for notional three-dimensional production blocks. Moreno et al. [10] considered the impact of ore storage on production and constructed multiple linear programming models to optimize production planning. Some scholars have used the SIP (Stochastic Integer Programming) model in ore blending [11,12,13,14,15,16]. An intelligent optimization algorithm has considerable advantages in solving the optimization model. With the continuous in-depth study of open-pit ore blending optimization, the model has become more and more complex. Cui et al. [17] proposed an artificial bee swarm (ABC) algorithm based on distance fitness and verified it with a standard data set. Bahram and Nader [18] combined ABC with radial basis function (RBF) and backpropagation (BP) neural network to predict phosphate ore grade. Chen et al. [19] proposed a multi-objective ore blending optimization model and non-dominated sorting genetic algorithm III, which can effectively solve the mixing problem of polymetallic ores and obtain a satisfactory ore blending solution. Yao et al. [20] proposed a multi-source and multi-target iron ore blending method in open-pit mine production, so that a single mine can both supply ores to one concentrator and to multiple concentrators.

In the research on truck dispatching system (TDS) in open-pit mines, Kou et al. [21] broadened the evaluation of TDS and provided a theoretical basis for the optimization of TDS. Ahumada et al. [22,23] described a distributed approach based on multi-agent systems to provide a precise dispatching solution. Xia et al. [24] proposed an efficient multi-objective dynamic dispatching method, which will dynamically adjust vehicle–load assignments according to the real-time monitoring of the system. Mohtasham et al. [25] presented a multi-stage optimization framework to solve the real-time truck decision problem in open-pit mines. Smith et al. [26] tested a nonlinear optimization model and mixed integer programming (MIP) model for dispatching trucks in an open-pit mining operation, and the results indicated that the MIP-based dispatching policy was better. The reinforcement learning-based fleet dispatching solution proposed by Huo et al. [27] can reduce direct emissions with existing equipment. In the existing technology, it first solves the problems of the inconsistency between the driving speed and the track of manually driven vehicles, between the time required driving in the same path, and problem of the dispatching system not being able to obtain the driving position in time. Second, it solved of the problem of unbalanced loads in the loading and unloading area caused by failure to predict. Finally, the reasonable planning of traffic flow was realized. Users have a variety of choices, and therefore, they can efficiently choose the appropriate dispatching scheme according to actual needs. In addition, when calculating the distance between the loading point and the unloading point, the real-time congestion degree of the road is taken into consideration, which can effectively save the driving time.

The following sums up the deficiencies of the existing research:

(1)

The existing methods are mostly designed to deploy trucks to reduce production costs, without considering the blending problem of the selected ores.

(2)

In some open-pit mines, an ore blending plan would be made before each shift of production and issued to the truck dispatcher for implementation. However, it often happens that the shovel cannot participate in the production due to faults and other reasons, for example, the ore grades near the shovel vary greatly or the rough crusher fails. In such cases, there would be inconsistencies between the ore blending target and the actual results.

(3)

The existing truck dispatching methods for open-pit mines cannot meet the dual requirements of ore blending and truck dispatching in open-pit mine production.

In order to solve the above technical problems, meet the actual needs of the current open-pit mine for ore blending, and coordinate the existing conflict between the ore blending and truck dispatching in the open-pit mine, this paper proposes an open-pit mine TDS based on dynamic ore blending decisions. It can not only realize the optimized truck dispatching in open-pit mine production, but also meet the requirements of downstream concentrators for ore dressing grade.

2. Open-Pit Mine Truck Dispatching System

The open-pit mine TDS based on dynamic ore blending decision is shown in Figure 1. The whole system consists of an intelligent ore blending module (IOBM), intelligent truck dispatching module (ITDM), intelligent 3D digital geological module (I3DDGM), on-board terminal, and communication network. The IOBM dynamically generates the ore blending plan with the maximum output and the shortest haul distance as the constraint condition. The ITDM executes the received ore blending plan. The I3DDGM is designed to send the grade data of the shovel position participating in ore blending to the IOBM. The on-board terminal receives dispatching instructions and feedback information sent by the ITDM. The communication network ensures the communication between the shovel, truck, and the ITDM.

2.1. Intelligent Ore Blending Module

The IOBM takes the maximum output and the shortest transportation distance as the constraint conditions and simultaneously meets the ore grade requirements of the ore unloading point to dynamically generate the ore blending plan [20]. The ore blending plan lists the ore transport proportion from each shovel to each ore unloading point and the shovel and truck are produced according to this proportion; the ore grade prediction that can be obtained at each ore unloading point. The dynamic refers not only to the generation of an ore blending plan before each shift of production, but also the regeneration of an ore blending plan if the shovel fails to participate in production, or the ore grade deviation at the ore unloading point is too large. The specific steps are as follows:

(1)

Through the interface of the ITDM, the position coordinates of the shovel to be involved in the production of this shift are obtained.

(2)

Send the location coordinate to the I3DDGM through the data interface to obtain the ore grade data of the location coordinate.

(3)

The shovels shall be grouped, and the shovels with grade data less than 3% shall be distributed together, with 2~4 shovels allocated to each group.

(4)

Before the production of each shift, the ore blending plan is generated, with the ore transport proportion from each shovel listed to each uploading point. The plan is then sent to the ITDM.

(5)

If a shovel fails in production, the on-board terminal of the shovel will send the fault information to the ITDM, which would in turn sends the information to the IOBM. The ore blending module will assign the work tasks of the failed shovel to other shovels in the same group according to the previous shovel groups, and send the regenerated ore blending plan to the ITDM for execution.

(6)

The ITDM would record the real-time production data during the implementation process. When the ore grade deviation at the unloading point reaches more than 10%, the ITDM will send a request to the IOBM, and the IOBM will regenerate the ore blending plan and send it to the ITDM for implementation.

2.2. Intelligent 3D Digital Geological Module

The I3DDGM stores the ore body grade data of the entire open-pit mine, and can send the ore grade data of the shovel location to the IOBM through the data interface.

The main functions of the I3DDGM include:

(1)

The I3DDGM can realize 3D solid modeling of mine geology, accurately describe the geological distribution, and meet the ore blending requirements.

(2)

With its data management function, I3DDGM can manage geological basic data such as drilling and survey, and color display the drilling, existing ore body, and surface model.

(3)

With the ore body measurement function, the I3DDGM can quickly calculate and verify the physical volume, ore volume, grade, and other information.

(4)

The I3DDGM has the function of automatically connecting the geological level and automatically updating the geological model of the blasting area.

(5)

The I3DDGM has the function of model modification and management.

2.3. On-Board Terminal

The on-board terminal is installed in the shovel and truck cab to complete the functions of positioning, dispatching command issuing, and information feedback.

2.4. Communication Network

The communication network includes a 4G/5G wireless network, master station, and relay station for data transmission. During the production process, 4G/5G wireless network coverage is available at most locations, which can facilitate communication needs between shovel, truck, and the ITDM. However, with the advancement of mining technologies, there will be a dead corner of network coverage. In such case, the master station of data transmission radio and the relay station of data transmission radio would be used as a supplement for network communication to ensure the real-time performance of the ITDM.

3. Open-Pit Mine Truck Dispatching Method

The open-pit mine truck dispatching method based on dynamic ore blending decision is shown in Figure 2. The specific steps are as follows:

(1)

Obtain the ore blending plan generated by the intelligent ore blending module.

(2)

Set the shovel priority and the unloading point priority.

(3)

The trucks can start working when ready and the truck drivers receive the task order through the on-board terminal.

(4)

Dispatch trucks to the shovel for ore loading according to the priority of the shovel, and record the loading amount.

(5)

According to the priority of the ore unloading point, the trucks should be sent to the ore unloading point for ore unloading, and the amount should be recorded.

(6)

Calculate the proportion of ore discharged at each ore unloading point from each shovel.

(7)

Compare the proportion calculated in step (6) with that of ore blending plan in step (1) to determine the next shovel and the unloading point for the truck.

(8)

Send a truck to the shovel to load ore according to the calculation result in step (7), and record the loading amount.

(9)

According to the calculation results in step (7), send a truck to the ore unloading point for ore unloading and record the ore unloading amount.

(10)

Go back to step (6) until the production of the shift is completed.

In step (2), if the completion ratio is the same, the priority of shovels should be given to the vehicle loading; as for the unloading point, the priority should be given to the vehicle loading.

In step (3), there are two situations indicating that the truck is ready. One is when the shift just starts production, and the driver selects “Prepare to complete the request task” through the on-board terminal; the other is that after one ore unloading task is completed, the driver selects “Ore unloading completion request task” through the truck terminal.

In step (4), the ore unloading volume recorded is calculated according to the average load when the truck of this model is fully loaded.

In order to bring the above method into effect, some technical applications are required. First, 4G/5G wireless network technology is needed to provide a channel for data distribution and upload. A BeiDou Navigation Satellite System realizes the positioning of electric shovels and trucks. The Internet of Things technology uses the message queuing telecommunication transport (MQTT) protocol to realize the management of control instructions and feedback data between the truck and the background server.

4. Experiment and Results

The system has been applied in Ansteel Group QIDASHAN mine for one year. As shown in Figure 3, it is a real-time dispatching screen, through which the user can tell the real-time dispatching information of the entire open-pit mine. On the left side of the screen is the list of equipment, including 24 trucks, four shovels, and the list of standby trucks and standby shovels. In the middle of the screen is real-time dispatching information, including a list of trucks from each shovel to each unloading point and a list of trucks from each unloading point to each shovels. The red truck indicates that the position of the equipment has not changed for a long time during transportation. On the right side of the screen are truck loading and unloading information, including trucks loading ore with an electric shovel and trucks unloading ore at an unloading point. The red label on the truck near the shovel indicates that the truck is loading, and the blue label indicates that the truck is ready to load. The red label on the truck near the unloading point indicates that the truck is in operation of unloading.

Figure 4 shows the real-time statistical information of the system and the status of the trucks and shovels. The statistical information mainly includes current team, current shift, rock and ore transportation time, transportation distance, and production. The status of trucks and shovels mainly includes the current number of running and ready trucks/shovels and troubleshooting.

Figure 5 shows the road networks and unloading point information, as well as real-time shovel and truck location information. The location of the red dot is the unloading point. The “1#” and “2#” marked in red at the front of the figure represent the working shovel, and the “N5” and “H5” represent the working trucks.

Here are two practical examples to illustrate how the system and the method proposed in this paper can be used for truck dispatching.

The QIDASHAN mine is a large open-pit iron ore enterprise, with an annual design production capacity of 51 million tons. At present, the main production equipment of the QIDASHAN mine includes 16 rotary drills, 20 electric shovels, and 54 trucks. In addition to some outsourced mining operations, about five electric shovels and 20 trucks can be put into operation at the same time per shift. Example 1 is the scene of five electric shovels loading ores and two unloading points unloading ores in a shift. It describes in detail how the method proposed in this paper realizes the truck dispatching function in this scenario. Example 2 is how to realize dynamic adjustment in case one shovel fails in the working scenario of example 1.

4.1. Example 1

There are five electric shovels in an open-pit iron mine, among which, No.16, No.17 and No.2 shovels are put into one group because their ore grades are relatively close, whereas No.12 and No.14 shovels are put into another group. It is planned to draw ore to two ore unloading points. The IOBM obtains the ore blending plan of the electric shovels as shown in

Table 1. Shovel ore blending plan.

Shovel No.	Ore Grade (%)	North Crushing Ratio (%)	South Crushing Ratio (%)
16	35.19	13.74	11.10
17	33.12	20	0.00
2	32.38	15.94	53.44
12	25.18	26.72	35.46
14	26.46	23.6	0.00

and the ore blending prediction of the ore unloading points as shown in

Table 2. Ore blending prediction of ore unloading point.

No.	Ore Unloading Point	Grade Prediction (%)
1	North Crushing	29.59
2	South Crushing	30.14

. The ore blending plan of electric shovels lists the proportion of ore transported from each electric shovel to each ore unloading point. The average ore grades near the five shovels are 35.19%, 33.12%, 32.38%, 25.18%, and 26.46%, respectively. The specific data are the proportions of ore that the five electric shovels transported to the north crushing are 13.74%, 20%, 15.94%, 26.72%, and 23.6%, respectively, and those transported to the south crushing are 11.10%, 0%, 53.44%, 35.46%, and 0%, respectively. The ore blending prediction of ore unloading points lists the predicted grade results of each ore unloading point if it is produced according to the ore blending plan of the electric shovel. The predicted results of the north crushing ore unloading point are 29.59% and the south crushing ore unloading point is 30.14%. The shovel ore blending plan is sent to the TDS of the open-pit mine, which then organizes the production.

The TDS of the open-pit mine dispatches trucks according to the following steps:

(1)

Obtain the ore blending plan generated by the intelligent ore blending module, that is, obtain the data in Table 1.

(2)

Set the priority of the shovel and the priority of the unloading point, and the priority sequence of the shovel from high to low is No.16, No.17, No.2, No.12, and No.14. The priority sequence of unloading points from high to low is 1 and 2.

(3)

The current shift starts to work, and truck drivers log in and go online one after another through on-board terminals to request trucks dispatching tasks.

(4)

Arrange the trucks to shovel No.16, No.17, No.2, No.12, and No.14 according to the sequence of tasks requested by truck drivers online, and record the loading capacity.

(5)

The truck load is 120 t. After the truck is full, the truck should be dispatched to unloading point 1 according to the priority of unloading point to unload ore, and the unloading amount should be recorded.

(6)

When five electric shovels complete the loading task of the five trucks, and the ore is unloaded into unloading point 1, the proportion of north crushing completed by each electric shovel is 20%.

(7)

Compare the proportion calculated in step (6) with that of the ore blending plan in step (1). The next truck should be sent to the No.12 electric shovel to load ore at unloading point 2, and the following truck should be sent to the No.14 shovel to load ore at unloading point 1.

(8)

Send a truck to the shovel to load ore according to the calculation results in step (7), and record the loading volume.

(9)

According to the calculation results in step (7), send a truck to the ore unloading point for ore unloading and record the ore unloading amount.

(10)

Return to step (6) to recalculate the proportion of each ore unloading point completed by each shovel until the shift is completed.

4.2. Example 2

According to example 1, when the shovel No.16 fails and cannot continue production, its work tasks need to be allocated to other shovels in the same group, that is, 13.74% of the ratio of shovel No.16 to the north crushing is equally allocated to shovel No.17 and shovel No.2, and 11.10% of the ratio of shovel No.16 to the south crushing is equally allocated to shovel No.17 and shovel No.2. The adjusted ore blending plan of the shovels is obtained as shown in

Table 3. Adjusted shovel ore blending plan.

Shovel No.	Ore Grade (%)	North Crushing Ratio (%)	South Crushing Ratio (%)
16	35.19	0	0
17	33.12	26.87	5.55
2	32.38	22.81	58.99
12	25.18	26.72	35.46
14	26.46	23.6	0.00

below.

If it is implemented according to the adjusted ore blending plan, the ore blending prediction of the ore unloading points is presented in

Table 4. Adjusted ore blending prediction of ore unloading point.

No.	Ore Unloading Point	Grade Prediction (%)
1	North Crushing	29.26
2	South Crushing	29.88

.

5. Discussion

In this study, a dynamic optimization TDS and method was proposed by combining ore blending and truck dispatching in open-pit mine production. Compared with the existing research results, there are mainly two innovations for the proposed method. First, the first version of the ore blending plan is generated before each shift of production and sent to the TDS for implementation. The results are tracked in real time during the production process. When the deviation between the results and the plan is large, the plan can be dynamically adjusted and the deviation can be corrected to achieve the best results, which combines the ore blending results and the shunting results. Second, if the shovel fails in the production process, the ore blending plan can be readjusted to be more assimilated to the actual production site.

According to example 1, when the shovel No.16 fails and it cannot continue production; if the ore blending plan is not adjusted and production continues according to the original plan, the ore blending prediction of ore unloading point are shown in

Table 5. Ore unloading point without adjusting ore blending prediction.

No.	Ore Unloading Point	Grade Prediction (%)
1	North Crushing	28.70
2	South Crushing	29.51

.

The three ore blending predictions of the ore unloading points of the original plan, the adjusted plan, and the non-adjusted plan are shown in the histogram in Figure 6.

Taking the south crushing as an example, the following three points can be seen from Figure 6.

(1)

If the production is carried out according to the original ore blending plan, the grade of ores transported and uploaded to the south crushing is 30.14%.

(2)

The ore blending plan was adjusted due to the failure of the No. 16 electric shovel. If the production is carried out according to the adjusted ore blending plan, the ore grade of the south crushing is 29.88%.

(3)

When the No. 16 electric shovel failed and the ore blending plan was not adjusted, the production was still carried out according to the original plan, and the ore grade of the south crushing was 29.51%.

As is shown in Figure 6, the adjusted ore blending plan has a grade fluctuation of 0.86% compared with the original plan, and the non-adjusted ore blending plan has a grade fluctuation of 2.1% compared with the original plan. The application of this method has contributed to the stability of the ore grade.

Although these studies reveal some important findings, they also have limitations. First of all, the weight of each ore transported by truck is only a rough estimation at present, which has not been accurately measured, so there would be deviation in the calculation of transportation volume. In addition, the current dispatching algorithm does not incorporate more factors that affect the shoveling conditions, such as weather, muddy roads, etc.

We will further conduct follow-up studies on these problems. First of all, we plan to develop a powerful, scalable, and intelligent road network generation method to accurately calculate the truck transportation distance. Second, we will endeavor to combine the system with autonomous trucks to try unmanned operation in open-pit mine production.

6. Conclusions

This paper proposes an open-pit mine truck dispatching system based on dynamic ore blending decisions, which can not only realize the optimized truck dispatching of open-pit mine production, but also meet the requirements of downstream concentrators for ore dressing grade. The system has been applied in the Ansteel Group QIDASHAN mine for one year. The proportion of trucks dispatched through the system reached more than 70%. The trucks’ capacity was upgraded from 3.79 to 4 million ton km per set per year, and the efficiency was improved by 5.5%. The main conclusions are as follows:

(1)

The system and the method proposed in this paper combines open-pit ore blending with truck dispatching, which not only optimizes truck dispatching but also improves ore blending.

(2)

The system and method proposed in this paper have the dynamic optimization function for open-pit ore blending and truck dispatching in emergencies in production (such as sudden equipment failure), which is more in line with the actual situation of the open-pit production site.

(3)

The system and the method proposed in this paper are produced according to the proportion of ore blending plan, which not only ensures the requirements of ore dressing grade at the ore unloading point, but also ensures the requirements of the maximum output and the shortest transportation distance.

Based on the current research results, in the future, the system can be combined with unmanned automatic mining trucks to realize the unmanned operation of ore blending and truck dispatching in open-pit mines.