0% found this document useful (0 votes)
38 views9 pages

Main

Uploaded by

gaby.galindo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
38 views9 pages

Main

Uploaded by

gaby.galindo
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

Available online at www.sciencedirect.

com

ScienceDirect
IFAC PapersOnLine 55-10 (2022) 2276–2281

Mixed Integer Linear Programming Model for Open Pit Mine Scheduling
S. Aallaoui*. A. Azzamouri* N. Tchernev**
* EMINES, School of Industrial Management, Mohammed VI Polytechnic University, 43150 Benguerir, Morocco
(e-mail: soufiane.aallaoui@emines.um6p.ma, ahlam.azzamouri@emines.um6p.ma)
** LIMOS (UMR CNRS 6158), Clermont Auvergne University, 63177 Aubière Cedex,
France (e-mail: Nikolay.tchernev@uca.fr)

Abstract: An open-pit mines consist of the accumulation of ore layers with different chemical
characteristics and aims to extract ores from the surface. Based on a geological discretization of the deposit,
the ore body model is represented as a three-dimensional array of blocks divided into two subsets; ore
blocks, which can be extracted and processed profitably, and waste blocks, which include all the
remaining blocks. The extraction process results from the iteration of a sequence of elementary
operations performed on blocks by specialized or multipurpose machines. Short-term planning of the ore
extraction process in open-pit mines is a complex undertaking that necessitates the identification of the
best spatio-temporal machinery affectation. All of this is constrained by several factors: operations
sequencing, machines compatibility, machines availability, machines motions, and priorities of layers
(urgency of obtaining certain qualities). In this study, a Mixed Integer Linear Programming model
(MILP) is used to generate the short-term extraction schedule. The MILP is based on the well-known flexible
job-shop theoretical problem with generic minimal time-lags between operations. A simple example is
used to clearly explain the Copyright
model and©the 2022
solution obtained.
Copyright © 2022 The Authors. This is an open access article under the CC BY-NC-ND license
(https://creativecommons.org/licenses/by-nc-nd/4.0/)
Keywords: Scheduling, Flexible Job Shop, Time-lags, Open-pit mine, Short-term.
1. INTRODUCTION 𝑂𝑗𝑗 = {𝑂𝑗𝑗,1 , 𝑂𝑗𝑗,2 , . . . , 𝑂𝑗𝑗,𝑘𝑗𝑗 } where 𝑘𝑗𝑗 refers to the number of
The planning of open-pit mine extraction is widely studied in operations of the job j (which means that the jobs could have
the literature since the 1960s (Newman et al., 2010). In this a different number of operations that must be scheduled). The
context, strategic decisions are translated into long and extraction process results from the iteration of a sequence of
medium-term plans, whereas short-term plans are generally
elementary operations 𝑂𝑗𝑗,𝑜 performed on blocks by
referred to as operational decisions. Short-term and
operational plans constitute tactical plans and must comply specialized or multipurpose machine 𝑀𝑗𝑗,𝑜,𝑚 from a set of
with the long-term plans, by fulfilling the operational machines 𝑀𝑗𝑗,𝑜 =
objectives on a shift-to-shift or day-to-day basis. Detailed
{𝑀𝑗𝑗,𝑜,1 , 𝑀𝑗𝑗,𝑜,2 , . . . , 𝑀𝑗𝑗,𝑜,𝑀 }. These resources are limited and
plans at the operational level are expected to provide blocks
extraction sequence and equipment planning over short time urge the planner to find the best spatio-temporal machinery
horizons such as months or weeks (Upadhyay and Askari- affectation. The sedimentary nature of the mine requires that
Nasab, 2016). the extraction of a certain block can be possible only if the
upper level is removed (Azzamouri et al., 2018, 2021). All of
The OCP Group is the largest Moroccan company and a this is constrained by several factors: operations sequencing,
pioneer in the international phosphate market since it holds machines compatibility, machines availability, machines
more than 70% of the currently known phosphate deposits. motions, and priorities of layers (urgency of obtaining certain
Obtaining this commodity requires its extraction from open- qualities). For example, to extract a phosphate block covered
pit mines. Open-pit phosphate mines deposits consist of the by a single block of waste (e.g. a parcel that contains two
accumulation of layers of different geological nature (Figure blocks) means that the job j will be extracted based on 7
1). The deposit contains several panels, each panel is elementary operations, using the available machines as
composed of a group of trenches, and each trench is made up follows: Waste block recovery that consists of {Site
of parcels: rectangles of 4000 m2 (40m * 100m) that contain preparation for drilling: 𝑂𝑗𝑗,1}, {Drilling: 𝑂𝑗𝑗,2},
alternate phosphate and waste blocks. Before extraction and {Blasting: 𝑂𝑗𝑗,3}, {Site preparation for stripping: 𝑂𝑗𝑗,4} and
according to ore grade, some layers are grouped into {Stripping : 𝑂𝑗𝑗,5}; Phosphate block recovery involving
extracted grades while the others correspond to “waste” (i.e. {Stacking: 𝑂𝑗𝑗,6} and {Loading: 𝑂𝑗𝑗,7}.
ore having an insufficient percentage of phosphate). The
phosphate ore extracted is blended to get merchantable ores The decisions to be taken by the planner relate to (Azzamouri
respecting the chemical composition constraints of the et al., 2019): (i) The choice of parcels to be extracted
customer (Hilali et al., 2022). Without loss of generality, knowing that operations are in progress in several parcels; (ii)
each parcel can be considered as a job j and each one of these the allocation of the machines available to perform these
jobs j involves a set of operations operations. In this paper, it is assumed that the blocks
extraction order has been made.
2405-8963 Copyright © 2022 The Authors. This is an open access article under the CC BY-NC-ND license.
Peer review under responsibility of International Federation of Automatic Control.
10.1016/j.ifacol.2022.10.047
S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276– 2

- Specifications of the scheduling problems: The approach


followed, and the methods used in terms of:
 The nature of operations modeled: elementary or
aggregated operations and the perimeter covered
(extraction, processing units, ports, etc.);
 The range of constraints considered: precedence,
capacities, machines, blending, transport, etc.;
 The modeling approach used: simulation, optimization,
etc. (with specifying the objective function that can be
economical or temporal objectives, etc.);
 The solving approaches used.

Blom et al. (2019) state in their review that open-pit mine


scheduling problems have not been as widely considered as
that for the medium- and long-term horizons. They examine
Figure 1. Open-pit block model and the extraction operations. the state-of-art in open-pit mines' short-term scheduling; by
presenting a range of techniques developed for generating
On the whole, the problem is to schedule a set of J jobs where
short-term plans, from mathematical-based methods to
the set of operations 𝑂𝑗𝑗 = {𝑂𝑗𝑗,1, 𝑂𝑗𝑗,2, . . . , 𝑂𝑗𝑗,𝑘𝑗𝑗 } must be
heuristic approaches. Newman et al. (2010) provide a review
processed in a pre-determined order and no pre-emption is of the use of operations research in mine planning, covering
allowed, and each operation is processed by one and only one work on multiple horizons (from day-to-day decision making
machine from the set 𝑀𝑗𝑗,𝑜 = {𝑀𝑗𝑗,𝑜,1, 𝑀𝑗𝑗,𝑜,2, . . . , 𝑀𝑗𝑗,𝑜,𝑀} of the to long-term planning and mine design), and on both
compatible machines. Hence, this paper addressed scheduling underground and open-pit mines. Open-pit long-term plans
and an assignment problem and tried to answer the following use block models that contain millions of blocks and are used
research questions: (i) How can the extraction process of open- for strategic decisions (Newman et al., 2010; Blom et al.,
pit mine be modeled? (ii) Are there classical models from 2019; Fathollahzadeh et al., 2021). At the operational level,
scheduling theory that can translate the short-term planning the planners use short-term plans to make day-to-day or shift-
model of surface mining? to- shift decisions, such as determining the sequence of
Indeed, the problem described above corresponds to the mining which means identifying the blocks to be extracted
Hybrid Flow Shop Scheduling Problem (HFSP) (Pinedo, and their order (L’Heureux et al., 2013; Lamghari and
2016). This means that the resolution of this problem is close Dimitrakopoulos, 2016). Considering elementary operations
to the Flexible Job Shop Scheduling Problem (FJSP) scheduling, we found that Kozan and Liu (2016) propose a
(Brucker and Schlie, 1990), by considering 𝑘𝑗𝑗 similar for all multi-stage scheduling methodology to solve a short-term
jobs open-pit mine scheduling problem by modeling several
production operations such as drilling, sampling, blasting, and
𝑗𝑗𝑗𝑗{1, . . . , 𝐽}. Time-lags can be used as restrictions enforcing
excavating. Moreover, L’Heureux et al. (2013) model the
minimal or maximal delays between two operations, to model
moving shovels activity, drilling, and blasting operations.
general timing relations between jobs like: start-start relations
Generally, the literature review of the open-pit mine short-
between jobs (Lacomme et al., 2011). In this case, minimal
term scheduling is about modeling the mining operations
time-lags impose a minimal delay between the end date of an
using different methods that vary in terms of the level of
operation and the start date of the next one. Hence, the
detail, shovels (for the ore excavation operation), and trucks
problem under study is a Flexible Job Shop Scheduling
(for the ore transportation) allocation. This is due to the large
Problem with generic time-lags, and the objective is to find a
operating cost of the machinery fleet in open-pit mines which
feasible schedule with minimal makespan (minimal global
can reach 60% of the total operating cost (Moradi Afrapoli
duration) by managing machine disjunctions, as the jobs are
and Askari-Nasab, 2019). And by modeling the excavation
processed by mutual machines and assignments.
and haulage operations, all the elementary extraction
The remainder of the paper is organized as follows. First, a operations are aggregated in these two operations. (Alarie and
literature review on short-term scheduling in open-pit mines Gamache, 2002; Moradi Afrapoli and Askari-Nasab, 2019)
is presented in section §0. Next, the problem description and present a review on fleet management in open-pit mines.
modeling are proposed in section §0. Then, we present the Considering quality constraints, Souza et al. (2010) propose a
extraction scheduling model as a MILP. Section §0 gives an dynamic truck allocation to deal with an open-pit mining
example of results obtained using the developed model operational problem to optimize mineral extraction in the
before concluding in section §0. mines, by minimizing the number of trucks used to meet the
production requirements in terms of quantity and quality.
2. RELATED WORKS Additionally, Upadhyay and Askari-Nasab (2016) focus on
the truck-shovel operations modeling and propose a model
To address this problem, an in-depth literature review
using binary variables to allocate shovels to faces (collections of
analysis is realized based on several criteria, such as:
blocks) per period. Integer variables define the number of
- Scheduling or planning type: the mine type, the decision trips made by trucks between faces and destinations (such as
level, the horizon; processing plants), and continuous variables define the
quantities
2 S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276–

extracted from each face in each period. A weighted sum of and blasted.
several objectives (maximizing production, meeting desired
feed characteristics in each plant, meeting blending
requirements, and minimizing shovel movements) is
optimized. However, Lamghari and Dimitrakopoulos (2016)
optimize not only the physical sequence of material
extraction but also the assignment of the mining equipment
fleet based on equipment capacity, availability, and hauling
time.
These short-term schedules must consider several mining
constraints, such as the precedence among the blocks to
comply with the possibility of access to certain blocks, the
safety rules on the degree of slope, and the precedence among
the elementary operations (L’Heureux et al., 2013). We can
also find the constraints related to the customer orders
defined in terms of quantity and quality as considered by
(Souza et al., 2010); a minimum production volume required;
or the uncertainty related to the extracted ore as proposed by
(Lamghari and Dimitrakopoulos, 2016). Lamghari and
Dimitrakopoulos (2016) consider in their short-term
production schedule formulation, the equipment capacity
such as shovels and trucks; and also, the blending constraints
aiming to ensure the good quality and quantity of the ore that
feeds the blending process. In addition to respecting a set of
constraints, generating optimal or near-optimal schedules
depends on the objectives that are optimized when making
decisions. These objectives aim to meet production targets
(Smith, 1998; Souza et al., 2010; Upadhyay and Askari-
Nasab, 2018), minimize costs (Lamghari and
Dimitrakopoulos, 2016), maximize revenue (Jélvez et al.,
2016), or to minimize makespan (time between the start and
the end of the operations) (Kozan and Liu, 2016). All these
problems are modeled using simulation or/and optimization
techniques. In general, simulation is used to evaluate the
result, or the behavior of a given mine plan under different
scenarios due to the large sources of uncertainty as indicated
by (Blom et al., 2019). (Azzamouri et al., 2018; Alaoui et al.,
2019) address through a discrete event simulation approach
the problem of the medium and short-term scheduling
problem in a phosphate mine that defines the blocks to be
extracted, the cumulative feeding curves of extraction ores,
and the assignment Gantt of machinery over time. On the
other hand, Mena et al. (2013) propose a combination of
simulation and optimization approaches to solve the truck
and shovel assignment problem, i.e., a planning model for the
loading and transportation operation. Similarly, Upadhyay
and Askari-Nasab (2018) designed the MOOT tool ("Mine
Operational Optimization Tool") to provide shovel allocation
decisions based on a strategic schedule; and by combining it
with a simulation tool, it manages to give the optimal
assignment of shovels and trucks in the vision of achieving
the production goals by optimally using the machinery fleet.
3. EXTRACTION SCHEDULING MODEL
3.1 Problem description
In this study, we deal with the proposal of an extraction
scheduling model at the operational level of a phosphate
open- pit mine. The mining method used is a selective
method of recovering the phosphate layers: waste blocks are
stripped using Draglines and Bulldozers, after being drilled
S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276– 2
This is followed by phosphate recovery, which consists of
stacking and loading the phosphate for transport by trucks.
These operations are repeated until all levels are mined.
The phosphate extraction process is based on the
elementary operations (Azzamouri, 2018) defined in Table
1 as shown in Figure 1.
Table 1. Elementary operations description

Operation Description
The objective is to remove all the undesirable elements or
Site obstacles existing on the land, to make the surface of the
preparation 𝑂𝑗𝑗,1 land more suitable for the machines, and to prepare the site
for drilling for the next operations. This operation can be carried out by
3 types of machines: Bulldozer, Grader or a Paydozer

Drilling 𝑂𝑗𝑗,2 This operation is processed by a drilling machine to create a


set of holes in the site to house the explosive
The objective of this operation is the fragmentation of the
Blasting 𝑂𝑗𝑗,3 ground by explosive charges (dynamite placed in the holes)
and make it easy to strip
Site The blasted ground surface is prepared for the next
preparation 𝑂𝑗𝑗,4 operation, using 3 types
of machines: Bulldozer, Grader or a
for stripping Paydozer
The ground fragmented by the previous operation is then

Stripping 𝑂𝑗𝑗,5 stripped to access the phosphate layer. This operation can
be processed using a Dragline, a Bulldozer, a Shovel, or a
surface miner
Consists in collecting the phosphate in the middle of the
Stacking 𝑂𝑗𝑗,6 trench to facilitate the loading operation. The machine used
to stack the phosphate is a Paydozer or a Bulldozer
The collected phosphate is loaded, using a Shovel or a
Loading 𝑂𝑗𝑗,7 Loader, into trucks that will then transport it to the desired
destination

The first five operations allow the removal of the waste


block, then perform the last two operations to recover the
phosphate block. The machines used are limited and require
the search for the right allocation plan to extract the blocks
in the shortest time possible. This allocation will be done
by respecting several constraints: i) Precedence is related
to the sedimentary character of layers which means that the
lower block cannot be extracted whether the upper block is
not yet removed. As well as the precedence between
elementary operations dedicated to the waste block that
should be executed before removing on the ore block
operations. Moreover, we follow a staircase extraction logic
in the extraction process allowing us to move from a parcel
to another one; ii) Machine availability, and iii) Machines
motions.
3.2 Problem modeling
The model’s objective is therefore to schedule the
operations 𝑂𝑗𝑗, for all the jobs {1, . . . , 𝐽}, based on the fleet of
machines 𝑀𝑗𝑗,𝑜 available for the exploitation and considering
all the constraints mentioned before. Thus, this problem can
be modeled as a Flexible Job Shop Problem (FJSP) with
generic time-lags. The latter models the precedence
constraints and the movement of the machines between the
parcels to be extracted. In other words, the goal is to find
the optimal sequencing of operations using adequate
machines. Thus, our problem is composed of:
- Assignment problem: the aim is to assign the
available and compatible machines to the elementary
operations, to minimize the Makespan 𝐶𝑚𝑎𝑥.
2 S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276–

- Scheduling problem: this involves defining the start


𝑡𝑙(𝑗𝑗,𝑜),(𝑗𝑗′,𝑜′) : The time-lag between 𝑂𝑗𝑗,𝑜 and 𝑂𝑗𝑗′,𝑜′
date 𝑆𝑗𝑗,𝑜 of the operations 𝑂𝑗𝑗,𝑜 of each job 𝑗𝑗𝑗𝑗{1, . . . , 𝐽} to
:𝑂 The processing time of operation on machine
minimize 𝐶𝑚𝑎𝑥. 𝑗𝑗,𝑜,𝑚 𝑗𝑗,𝑜
𝑀𝑗𝑗,𝑜,𝑚
Theoretically, the FJSP can be described as follows
(Chaudhry
and Khan, 2016): A set of 𝐽 jobs {1, . . . , 𝐽}, must be 𝑑𝑗𝑗,𝑜: The duration of the operation 𝑂𝑗𝑗,𝑜
processed by a set of 𝑚 machines 𝑀𝑗𝑗,𝑜 . Each job 𝑗𝑗𝑗𝑗{1, . . . , 𝐽}
is a sequence of 𝑘𝑗𝑗 operation (𝑂𝑗𝑗). Each operation 𝑂𝑗𝑗,𝑜 can be 𝑆𝑤_𝑇𝑇𝑇𝑚𝑒(𝑗𝑗,𝑜),(𝑗𝑗′,𝑜′),𝑚 : 𝑀𝑗𝑗,𝑜,𝑚 switching time from 𝑂𝑗𝑗,𝑜 to 𝑂𝑗𝑗′,𝑜′
executed by multiple machines in 𝑀𝑗𝑗,𝑜 . Each operation must 𝐻 : large positive number
be processed to complete the job. The execution of each
operation 𝑂𝑗𝑗,𝑜 of a job 𝑗𝑗 requires only one machine among Decision variables
the set of machines 𝑀𝑗𝑗,𝑜 . The processing time of the operation
𝐶𝑚𝑎𝑥: Makespan
𝑂𝑗𝑗,𝑜 by the machine 𝑀𝑗𝑗,𝑜,𝑚 is 𝑑𝑗𝑗,𝑜. FJSP consists of
determining the machine to be mobilized for each operation 𝑆𝑗𝑗,𝑜: Starting time of the operation 𝑂𝑗𝑗,𝑜
and the sequence of operations on the machines (called
1 𝑇𝑇𝑖𝑖 𝑀𝑗𝑗,𝑜,𝑚 𝑇𝑇𝑠 a𝑖𝑖𝑖𝑖𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝑂𝑗𝑗,𝑜
operation
sequencing). In general, the objective of FJSP is to optimize 𝑦𝑗𝑗,𝑜,𝑚 = {
some indicators, for example, Makespan (time between the 0 𝑂𝑡ℎ𝑒𝑟𝑤𝑇𝑇𝑠𝑒
start and the end of the operations), maximum delay, and 𝑥 1 𝑇𝑇𝑖𝑖 𝑂𝑗𝑗,𝑜 𝑇𝑇𝑠 𝑏𝑒𝑖𝑖𝑜𝑟𝑒 𝑂𝑗𝑗′,𝑜′ 𝑜𝑛 𝑀𝑗𝑗,𝑜,𝑚
=
total
duration. Moreover, this problem takes into consideration the (𝑗𝑗,𝑜),(𝑗𝑗′,𝑜′),𝑚 {
0 𝑂𝑡ℎ𝑒𝑟𝑤𝑇𝑇𝑠𝑒
following assumptions and constraints: All machines are
available at time t = 0; all jobs are available at time t = 0; The proposed mathematical model is defined as follows:
each operation is processed by only one machine at a time;
Objectif function
for each job, the sequence of operations is predefined; and it
is impossible to interrupt the processing of an operation 𝑚𝑇𝑇𝑛(𝐶𝑚𝑎𝑥) (1)
already launched.
Subject to
Furthermore, to make the scheduling problem modeling
simple, we consider the following assumptions: (i)The job 𝑗𝑗 𝑆𝑗𝑗,𝑜 + 𝑑𝑗𝑗,𝑜 ≤ 𝐶𝑚𝑎𝑥 , ∀ 𝑂𝑗𝑗,𝑜 ∈ Ω (2)
corresponds to the parcel containing the set of layers (waste
𝑀
and phosphate layers); (ii) the extraction method is a staircase
mining method, to guarantee the stability of the site, the ∑ 𝑦𝑗𝑗,𝑜,𝑚 ∗ 𝑀𝑗𝑗,𝑜,𝑚 = 1, ∀ 𝑂𝑗𝑗,𝑜 ∈ Ω (3)
access of the machines to the site and defines the precedence 𝑚=1
between the different blocks to be mined; (iii) the existence 𝑀
of a machine warehouse: a location where machines are 𝑑𝑗𝑗,𝑜 = ∑ 𝑦𝑗𝑗,𝑜,𝑚 ∗ 𝐷𝑗𝑗,𝑜,𝑚 , ∀ 𝑂𝑗𝑗,𝑜 ∈ Ω (4)
located during their idle time and from which these machines
𝑚=1
move to the operations to be processed; (iv) the minimum
time-lag corresponds to the time needed by a machine to get 𝑥(𝑗𝑗,𝑜),(𝑗𝑗𝘍,𝑜𝘍),𝑚 + 𝑥(𝑗𝑗𝘍,𝑜𝘍),(𝑗𝑗,𝑜),𝑚 − 𝑦𝑗𝑗,𝑜,𝑚 − 𝑦𝑗𝑗𝘍,𝑜𝘍,𝑚 ≥ −1,
back to the warehouse after processing an operation; (v) the ∀ 𝑂𝑗𝑗,𝑜 , 𝑂𝑗𝑗′,𝑜′ ∈ Ω, ∀𝑚 ∈ 𝑀𝑗𝑗,𝑜 ∩ 𝑀𝑗𝑗 𝘍 ,𝑜𝘍 , 𝑂𝑗𝑗,𝑜 ≠ 𝑂𝑗𝑗 𝘍 ,𝑜𝘍 (5)
switching time 𝑆𝑤𝑇𝑇𝑇𝑚𝑒 (𝑗𝑗,𝑜),(𝑗𝑗 𝘍 ,𝑜𝘍 ),𝑚 , corresponds to the
machine 𝑀𝑗𝑗,𝑜,𝑚 (𝑀𝑗𝑗,𝑜,𝑚 = 𝑀𝑗𝑗′,𝑜′,𝑚 ) travel time from the 2𝑥(𝑗𝑗,𝑜),(𝑗𝑗𝘍,𝑜𝘍),𝑚 + 2𝑥(𝑗𝑗𝘍,𝑜𝘍),(𝑗𝑗,𝑜),𝑚 − 𝑦𝑗𝑗,𝑜,𝑚 − 𝑦𝑗𝑗𝘍,𝑜𝘍,𝑚 ≤ 0,
operation 𝑂𝑗𝑗,𝑜 to the operation 𝑂𝑗𝑗′,𝑜′ location, while the ∀ 𝑂𝑗𝑗,𝑜 , 𝑂𝑗𝑗′,𝑜′ ∈ Ω, ∀𝑚 ∈ 𝑀𝑗𝑗,𝑜 ∩ 𝑀𝑗𝑗 𝘍 ,𝑜𝘍 , 𝑂𝑗𝑗,𝑜 ≠ 𝑂𝑗𝑗 𝘍 ,𝑜𝘍 (6)
time-lag between the two operations is infinite.
4. MATHEMATICAL MODEL 𝑆𝑗𝑗𝘍,𝑜𝘍 − 𝑆𝑗𝑗,𝑜 − 𝑑𝑗𝑗,𝑜 + 𝐻 (1 − 𝑥(𝑗𝑗,𝑜),(𝑗𝑗𝘍,𝑜𝘍),𝑚) ≥ 0 (7 )
∀ 𝑂𝑗𝑗,𝑜 , 𝑂𝑗𝑗′,𝑜′ ∈ Ω, ∀𝑚 ∈ 𝑀𝑗𝑗,𝑜 ∩ 𝑀𝑗𝑗 𝘍 ,𝑜𝘍 , 𝑂𝑗𝑗,𝑜 ≠ 𝑂𝑗𝑗 𝘍 ,𝑜𝘍
The following notation is used for the formulation of MILP
Indices and sets 𝑆𝑗𝑗,𝑜 + 𝑑𝑗𝑗,𝑜 + 𝑡𝑙(𝑗𝑗,𝑜),(𝑗𝑗𝘍,𝑜𝘍) ≤ 𝑆𝑗𝑗𝘍,𝑜𝘍 ,
∀ 𝑂𝑗𝑗,𝑜 , 𝑂𝑗𝑗′,𝑜′ ∈ Ω, ∀𝑚 ∈ 𝑀𝑗𝑗,𝑜 ∩ 𝑀𝑗𝑗 𝘍 ,𝑜𝘍 , 𝑡𝑙(𝑗𝑗,𝑜),(𝑗𝑗 𝘍 ,𝑜𝘍 ) ∈ ℝ (8)
𝑗𝑗: Index of jobs
𝑜: Index of operations 𝑆𝑗𝑗,𝑜 + 𝑑𝑗𝑗,𝑜 + 𝑆𝑤_𝑇𝑇𝑇𝑚𝑒(𝑗𝑗,𝑜),(𝑗𝑗′,𝑜′),𝑚 ≤

𝑚: Index of machines 𝑆𝑗𝑗𝘍,𝑜𝘍 + 𝐻 (1 − 𝑥(𝑗𝑗,𝑜),(𝑗𝑗𝘍,𝑜𝘍),𝑚) ,


∀ 𝑂𝑗𝑗,𝑜 , 𝑂𝑗𝑗′,𝑜′ ∈ Ω, ∀𝑚 ∈ 𝑀𝑗𝑗,𝑜 ∩ 𝑀𝑗𝑗 𝘍 ,𝑜𝘍 , 𝑡𝑙(𝑗𝑗,𝑜),(𝑗𝑗 𝘍 ,𝑜𝘍 ) = −∞
Ω = {𝑂𝑗𝑗}𝑗𝑗≤𝐽:The set of operations 𝑂𝑗𝑗,𝑜
(9)

𝑀𝑗𝑗, operation of 𝑂𝑗𝑗,𝑜 can be processed


𝑜 : The set of alternative machines on which
𝐾𝑛 �𝑛𝑔 𝑡ℎ𝑎𝑡: 𝐻 > S.∑
Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276–
𝑚𝑎𝑥𝑚∈𝑀𝑗𝑗,𝑜,𝑚 (𝑗𝑗,𝑜) 2
The objective function (1) minimizes the end date of the last
𝑜𝑤𝑇� (𝐷𝑗𝑗,𝑜,𝑚 )
operation commonly called makespan. Constraint (2) means
Parameters that the end date of the last operation is greater than the end
date of each operation. Constraint (3) enforces that each
2 S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276–

operation is processed by a single compatible machine. Table 3. Start date, duration, and machines assignment
Constraint (4) allows us to calculate the processing time of an
operation by the assigned machine. Constraints (5), (6), and
(7) represent machine disjunctions, i.e., if the processing 𝑂𝑗𝑗,𝑜 𝑆𝑗𝑗,𝑜 dj,o 𝑀𝑗𝑗,𝑜,𝑚
times of two operations overlap, then the two operations must 𝑂1,1 0 108 𝑀1,1,3
be assigned to different machines. Constraints (8) and (9) 𝑂1,2 𝑀1,2,6
123 19
correspond to the time-lag constraints and the time of passage 𝑂1,3 𝑀1,3,10
147 10
of the machines between operations. These constraints 𝑂1,4 𝑀1,4,1
translate those related to precedence between operations: In 160 107
𝑂1,5 282 57 𝑀1,5,5
the case of the existence of precedence between two
operations, the time-lag value is finite. Otherwise, where 𝑂1,6 349 32 𝑀1,6,4
there is no precedence between two operations, the time-lag 𝑂1,7 389 6 𝑀1,7,9
value is infinite. 𝑂2,1 349 108 𝑀2,1,3
𝑂2,2 472 19 𝑀2,2,6
5. RESULTS 𝑂2,3 496 10 𝑀2,3,10
Results have been computed, using CPLEX 12.7 on a Windows 𝑂2,4 509 107 𝑀2,4,1
Server 2012 R2 standard 64 bits with 64 Go of RAM, on a 𝑂2,5 631 57 𝑀2,5,5
generated set of small instances. In order to visualize the 𝑂2,6 698 30 𝑀2,6,1
solution, we propose an instance of 3 jobs, each job contains 7 𝑂2,7 736 5 𝑀2,7,7
operations which means the extraction of 3 parcels composed of 𝑂3,1 698 108 𝑀3,1,3
a single layer of waste and a single layer of phosphate, and a 𝑂3,2 821 19 𝑀3,2,6
fleet of 10 machines are available for the extraction process 𝑂3,3 𝑀3,3,10
845 10
(Figure 2). The result is obtained after 1.72 seconds. 𝑂3,4 𝑀3,4,1
858 107
𝑂3,5 980 57 𝑀3,5,5
𝑂3,6 1047 30 𝑀3,6,1
𝑂3,7 1085 5 𝑀3,7,7

6. CONCLUSIONS
In this study, we address the problem of Flexible Job Shop
Figure 2. Graphical representation of the 3 jobs with 7 operations. Problem with generic time-lags, in the context of the
extraction process in a phosphate open-pit mine. The Mixed
Table 2 shows the distribution of machines by elementary Integer Linear Programming model (MILP) developed
operations. These machines are indexed by the processing considers the precedence constraints among operations and
duration of the operation in question. blocks, the machine disjunction constraint, and their
displacement. The MILP model provides an exact solution
Table 2. Compatible machines and processing time per
operation
for small-scale instances but is quickly overtaken from the
instances of 12 jobs and each job contains 28 operations,
Operation Plan Compatible machines which means the exploitation of 12 parcels that contain 8
alternate waste and phosphate layers. Thus, a search for
𝑂𝑗𝑗,1 𝑀𝑗𝑗,1,1(107), 𝑀𝑗𝑗,1,3(108), 𝑀𝑗𝑗,1,4(109) another approach mainly based on metaheuristics to solve a
𝑂𝑗𝑗,2 𝑀𝑗𝑗,2,6(19) big-sized problem can be considered for future study. The
current results are obtained after 1.72 seconds using CPLEX
𝑂𝑗𝑗,3 𝑀𝑗𝑗,3,10(10) 12.7 on a Windows Server 2012 R2 standard 64 bits with 64
𝑂𝑗𝑗,4 𝑀𝑗𝑗,4,1(107), 𝑀𝑗𝑗,4,3(108), 𝑀𝑗𝑗,4,4(109) Go of RAM.
𝑂𝑗𝑗,5 𝑀𝑗𝑗,5,2(60), 𝑀𝑗𝑗,5,5(57), 𝑀𝑗𝑗,5,8(62), 𝑀𝑗𝑗,5,9(63) Furthermore, it could be interesting to add other mining
𝑂𝑗𝑗,6 𝑀𝑗𝑗,6,1(30), 𝑀𝑗𝑗,6,4(32) constraints imposed by the real scheduling model by taking
into consideration the chemical characteristics of ore and
𝑂𝑗𝑗,7 𝑀𝑗𝑗,7,7(5), 𝑀𝑗𝑗,7,9(6) economic constraints. As long as solving the problem becomes
difficult for large instances, it is important to consider the use
Table 3 represents the solution obtained after running the of robust solving methods to cope with industrial instances.
scheduling model, it’s an operations schedule with a
makespan equal to 1090 periods. This table presents the start REFERENCES
date, the processing time, and the machine assigned per Alaoui, N., Azzamouri, A., Elfirdoussi, S., Fenies, P., and
each operation 𝑂𝑗𝑗,𝑜. Giard, V. (2019, June). Système interactif d'aide à la
décision pour la planification de l'extraction minière.
In 13ème Conférence internationale de CIGI QUALITA
2019.
S. Aallaoui et al. / IFAC PapersOnLine 55-10 (2022) 2276– 2

Alarie, S., and Gamache, M. (2002). Overview of solution Lamghari, A., and Dimitrakopoulos, R. (2016). Progressive
strategies used in truck dispatching systems for open pit hedging applied as a metaheuristic to schedule
mines. International Journal of Surface Mining, production in open-pit mines accounting for reserve
Reclamation and Environment, 16(1), 59-76. uncertainty. European Journal of Operational
Azzamouri, A. (2018). Construction de méthodes et d'outils Research, 253(3), 843-855.
de planification pour l'industrie minière du phosphate L’Heureux, G., Gamache, M., and Soumis, F. (2013). Mixed
en contexte de Lean Management (Doctoral integer programming model for short term planning in
dissertation, Paris 10). open-pit mines. Mining Technology, 122(2), 101–109.
Azzamouri, A., Bamoumen, M., Hilali, H., Hovelaque, V., Mena, R., Zio, E., Kristjanpoller, F., and Arata, A. (2013).
and Giard, V. (2021). Flexibility of dynamic blending Availability-based simulation and optimization
with alternative routings combined with security stocks: modeling framework for open-pit mine truck allocation
a new approach in a mining supply chain. International under dynamic constraints. International Journal of
Journal of Production Research, 59(21), 6419-6436. mining science and Technology, 23(1), 113-119.
Azzamouri, A., Fénies, P., Fontane, F., and Giard, V. (2018). Moradi Afrapoli, A., and Askari-Nasab, H. (2019). Mining
Scheduling of open-pit phosphate mine fleet management systems: a review of models and
extraction. International journal of production algorithms. International Journal of Mining,
research, 56(23), 7122-7141. Reclamation and Environment, 33(1), 42-60.
Azzamouri, A., Fenies, P., Fontane, F., and Giard, V. (2017). Newman, A. M., Rubio, E., Caro, R., Weintraub, A., and
Modelling the tactical decisions for open-pit mines. Eurek, K. (2010). A review of operations research in
Blom, M., Pearce, A. R., and Stuckey, P. J. (2019). Short- mine planning. Interfaces, 40(3), 222-245.
term planning for open pit mines: a review. Pinedo, M. L. (2016). Flow Shops and Flexible Flow Shops
International Journal of Mining, Reclamation (Deterministic). In Scheduling (pp. 151-181). Springer,
and Environment, 33(5), 318-339. Cham.
Brucker, P., and Schlie, R. (1990). Job-shop scheduling with Smith, M. L. (1998). Optimizing short-term production
multi-purpose machines. Computing, 45(4), 369–375. schedules in surface mining: Integrating mine modeling
Chaudhry, I. A., and Khan, A. A. (2016). A research survey: software with AMPL/CPLEX. International Journal of
review of flexible job shop scheduling Surface Mining, Reclamation and Environment, 12(4),
techniques. International Transactions in Operational 149-155.
Research, 23(3), 551-591. Souza, M. J., Coelho, I. M., Ribas, S., Santos, H. G., and
Fathollahzadeh, K., Mardaneh, E., Cigla, M., and Asad, M. Merschmann, L. H. D. C. (2010). A hybrid heuristic
W. algorithm for the open-pit-mining operational planning
A. (2021). A mathematical model for open pit mine problem. European Journal of Operational
production scheduling with Grade Engineering® and Research, 207(2), 1041-1051.
stockpiling. International Journal of Mining Science Upadhyay, S. P., and Askari-Nasab, H. (2016). Short-term
and Technology, 31(4), 717-728. Production Planning in Open Pit Mines using Dynamic
Hilali, H., Hovelaque, V., and Giard, V. (2022). Integrated Shovel Allocations. Mining Optimization
scheduling of a multi-site mining supply chain with Laboratory, 1(780), 1.
blending, alternative routings and co-production. Upadhyay, S. P., and Askari-Nasab, H. (2018). Simulation
International Journal of Production Research. and optimization approach for uncertainty-based short-
https://doi.org/10.1080/00207543.2022.2049909 term planning in open pit mines. International Journal
Jélvez, E., Morales, N., Nancel-Penard, P., Peypouquet, J., of Mining Science and Technology, 28(2), 153-166.
and Reyes, P. (2016). Aggregation heuristic for the
open-pit block scheduling problem. European Journal
of Operational Research, 249(3), 1169-1177.
Kozan, E., and Liu, S. Q. (2016). A new open-pit multi-stage
mine production timetabling model for drilling, blasting
and excavating operations. Mining Technology, 125(1),
47-53.
Lacomme, P., Tchernev, N., and Huguet, M. J. (2011,
September). Dedicated constraint propagation for Job-
Shop problem with generic time-lags. In ETFA2011
(pp. 1-7). IEEE.

You might also like