Chinese Journal of Chemical Engineering: Jin Lang, Jiao Zhao
Chinese Journal of Chemical Engineering: Jin Lang, Jiao Zhao
Chinese Journal of Chemical Engineering: Jin Lang, Jiao Zhao
a r t i c l e i n f o a b s t r a c t
Article history: In this paper, an oil well production scheduling problem for the light load oil well during petroleum field exploi-
Received 16 June 2015 tation was studied. The oil well production scheduling was to determine the turn on/off status and oil flow rates
Received in revised form 28 July 2015 of the wells in a given oil reservoir, subject to a number of constraints such as minimum up/down time limits and
Accepted 29 November 2015
well grouping. The problem was formulated as a mixed integer nonlinear programming model that minimized
Available online 2 May 2016
the total production operating cost and start-up cost. Due to the NP-hardness of the problem, an improved par-
Keywords:
ticle swarm optimization (PSO) algorithm with a new velocity updating formula was developed to solve the
Oil well production problem approximately. Computational experiments on randomly generated instances were carried out to eval-
Scheduling uate the performance of the model and the algorithm's effectiveness. Compared with the commercial solver
Mixed integer nonlinear programming CPLEX, the improved PSO can obtain high-quality schedules within a much shorter running time for all the
(MINLP) instances.
Improved particle swarm optimization © 2016 The Chemical Industry and Engineering Society of China, and Chemical Industry Press. All rights reserved.
1. Introduction Oil fields are characterized by a large number of wells which spread
over a wide geographical area. The wells normally share surface
Oil plays a major role in nowadays society because it covers the high equipment and interconnect with each other through comprehensive
demand for vehicle fuel and is widely used in industrial processing. In pipeline networks as illustrated in Fig. 1. Oil fields contain a lot of reser-
the middle and later periods of the oil field exploitation, oil reserves voirs, each of which contains many wells. The pressure of interconnect-
are finite and most oil fields will run in low-yielding and inefficient ed wells that belongs to the same reservoir decreases globally.
pumping. According to the statistics, there are about 80% light load oil Therefore, the wells which interconnect with each other in the same
wells among all the pumping oil wells. In the oil field exploitation, elec- reservoir can be grouped into one batch. The novel operational ap-
tricity consumption accounts for about 35% of the whole petroleum re- proach proposed accordingly is to schedule the oil wells in batching.
covery cost. When oil well is running in light load state, a great deal of Compared with the conventional well scheduling problem, the prob-
electric power will be wasted and original pressure equilibrium will lem in this paper has the following characteristics that make it difficult
seriously be destroyed. Compared to continuous pumping, the intermit- to be modeled and solved: (1) the problem contains both the discrete
tent production method is more effective to tackle the issue of electric- variables to determine the oil well turn on/off status, and the continuous
ity waste in the oil field. It is possible to enhance oil field's productivity, variables to determine the oil production of oil wells; (2) during the oil
increase operation life of oil well, and reduce production cost by effec- production, not only minimum up and down time constraints, but also
tive pumping dispatch. In consideration of the actual production tech- the mutual influence relations between the pressure of oil well bottom
nique constraints, the oil well production scheduling problem in this and oil production should be considered. These requirements and con-
paper is to determine the turn on/off status in each time period so straints further increase the problem difficulties, resulting in a very
that the total production operation cost of oil wells is minimized. complex NP-hard problem. Since the number of wells scheduled in ac-
tual production is very large, it is impossible or very difficult to obtain
an optimal solution of the problem in a finite time. Thus, proposing ap-
proximation algorithms become a natural choice. Many solution tech-
niques were proposed focusing on the large scale practical production
☆ Supported by National High Technology Research and Development Program of China
scheduling problems in the previous literature [1,2]. In particular, in
(2013AA040704) and the Fund for the National Natural Science Foundation of China
(61374203).
the area of oil production, Ortiz-Gomez et al. [3] studied the three
⁎ Corresponding author. mixed integer multiperiod optimization models in chemical industry.
E-mail address: langjin@ise.neu.edu.cn (J. Lang). The problem took into account a fixed topology which was concerned
http://dx.doi.org/10.1016/j.cjche.2016.04.050
1004-9541/© 2016 The Chemical Industry and Engineering Society of China, and Chemical Industry Press. All rights reserved.
1424 J. Lang, J. Zhao / Chinese Journal of Chemical Engineering 24 (2016) 1423–1430
Reservoir C
Well 1 Well 2
Well j
Manifold
Reservoir A
Reservoir B
with the decisions involving the start-up/shut-down of the operation of the problem. In order to verify the effectiveness of the proposed
the process in each period of time. The models considered the nonlinear PSO, computational experiment is carried out by comparing with La-
behavior of the well flowing pressure. Carvalho et al. [4] studied the oil grange relaxation, binary particle swarm algorithm (BPSO) and the
well platform layout and the oil well distribution in each platform commercial solver CPLEX.
scheduling problem in sea, and proposed a mixed integer linear pro-
gramming model. Iyer et al. [5] researched the scheduling problems 2. Problem Description and Model
for reservoir production offshore fields, and proposed a mixed inte-
ger nonlinear programming model, which considered a lot of eco- 2.1. Problem description
nomic indicators, and then presented a continuous decomposition
technique to solve it. Kosmidis [6] researched the oilfield wells' This work considers the short term planning of the oil production in
daily scheduling problems, taking into account the nonlinear charac- the wells that belongs to the same reservoir over a time horizon T.
teristics of the reservoir, reservoir constraint multiphase flow and Hence, given an oil field, there are n wells within the reservoir and the
surface facilities, and established a mixed integer programming oil wells which interconnected with each other in the same reservoir
model to determine the switching state of the oil wells. Bohannon can be grouped into a batch (block). The scheduling of oil wells in
et al. [7] studied the oilfield plans and production design integrated oilfield production is to determine the optimum operation scheduling
problem, and established a mixed integer programming model. In of oil well during a given planning horizon, which includes the decisions
summary, existing studies only considered the needs of reservoir of determining the turn on/off time in each time period and the yield of
oil production and oilfield reserve constraints, neither of them con- production, so that the total production operation cost of oil wells is
sidered the stability requirements of well production and operation minimized while satisfying the demand of oil recovery and considering
characteristics, nor the relationship belong to an oil reservoir and the impact of oil well's bottom pressure variation on its turn on/off sta-
oil well's frequent turn on/off status on the well's own life and pro- tus, minimum start up and shut down time and so on.
ductivity. Bieker et al. [8] described how operational decisions were
made based on weekly or monthly oil production plans, determining 2.2. Mathematical model
the optimal oil production at each well, and corresponding adjust-
ments of controlling the pressure. Gunnerud and Foss [9] present a The problem can be formulated as a mixed integer optimization
model formulation including manifolds, wells, pipelines and routing problem in which the objective function is to minimize the total operat-
decisions for multiphase petroleum production asset. A piecewise ing costs over the scheduling horizon while satisfying all the constraints.
linear approximation and decomposition approach was used to The objective function includes production operation costs and start-up
solve the problem. Ulstein et al. [10] used a network flow formulation costs. Mathematically, the function is given as follows:
to describe the main production and transportation process of multi-
component flows to satisfy market demand. Güyagüler and Byer [11] XN X
T
min ηi qit zit þ θi zit 1−zi;t−1 : ð1Þ
design an optimization procedure that used an MILP to model a pe- i¼1 t¼1
troleum production allocation problem.
Compared with previous related research, the oil well production
Subject to the following constraints:
scheduling problem in this paper considers the minimum up and
down time constraints. This paper establishes a mixed-integer non- (1) Demand fulfillment on each period of time
linear programming model for the oil well production scheduling
problem considering pressure and minimum up and down con- X
straints. The goal is to minimize the oil well operation cost and the qit ≥Dkt ∀t ∈ T; ∀k ∈ K ð2Þ
start-up costs. Thereafter, an improved PSO is constructed to solve i∈Ωk
J. Lang, J. Zhao / Chinese Journal of Chemical Engineering 24 (2016) 1423–1430 1425
(2) The minimum up and down time constraints where S denotes the particles in the swarm and Sit is any one value be-
tween [0, 10]. The particles are encoding in real number value, which
1 if 1≤Gon it bT i
on
zit ¼ off off ∀i∈Ω; ∀t∈T ð3Þ need to be transformed to turning on/off status of oil wells. The
0 if ‐T i bGit ≤0
decoding process is as follows:
(3) Represent the behavior of the well flowing pressure. Notice that,
1 if sit b ¼ 5
if the well is turn on to flow, then the well flowing pressure will zit ¼ ∀i ∈ W; t ∈ T: ð9Þ
0 if sit N 5
decrease, otherwise the pressure increases.
(4) Generation limit constraints. Do the following treatment if the position of the particle is beyond
[0, 10] after updating:
zit qimin ≤qit ≤zit qimax ∀i∈Ω; ∀t∈T ð5Þ randð0; 5Þ if sit b 0
sit ∈ ∀i ∈ W; t ∈ T: ð13Þ
randð5; 10Þ if sit N10
The parameters in PSO are set as follows: the inertia weight is ob-
(5) Linking constraints from a given time period to the next one.
tained by formula (6). vmax is set as 1, c1 = 2 , c2 = 2. The initial velocity
of each particle is set to 0, while the size of the swarm is 100.
On/off status of oil well obtained by these two methods is ran-
xin out
it ¼ xi; t−1 ∀i∈Ω; ∀t∈T ð6Þ
dom, thereby making most of the particles unable to satisfy the oil
well combination constraints. This paper presents an initial popula-
tion repair strategy to ensure that the particles in the swarm meet
(6) Range of the variables. the minimum up and down time constraints, and improve the solu-
tion quality. Priority list method is used for determine the turn on/
off status of oil well. The characteristic curve of oil well consumption
zit ∈f0; 1g ∀i∈Ω; ∀t∈T ð7Þ is E = γi × x it. Obviously, the larger the oil well coefficient γ i , the
lower the efficiency of the oil well will be. Therefore, the ascending
According to the above analysis, the formulation is a large-scale orders for the oil well are employed to give priority for the oil well
mixed integer programming with both continuous and integer decision starting and stopping order. The fitness function of particle swarm
variables. optimization is calculated directly using the objective function of
oil well scheduling problem. If the fitness value is smaller, then the
3. PSO Algorithm objective function value of the particle is smaller which means that
the particle is more excellent.
This paper proposed an improved particle swarm algorithm to solve
an oil well production scheduling problem for the oil field exploitation. 3.2. An improved PSO
Particle Swarm Optimization (PSO) was first presented by Eberhart and
Kennedy [12,13], and used to optimize the continuous nonlinear func- As the size of the problem in practical scheduling is often
tions. Liao et al. [14] successfully applied PSO algorithm to solve a dis- large, and also due to the complexity of the problem itself, the
crete optimization problem. PSO algorithm falls into local optimal easily. On account of this,
an improved PSO algorithm (IPSO) is presented, in which a new
3.1. Particle encoding mode velocity updating formula is proposed to improve the perfor-
mance of the PSO algorithm. The new velocity updating formula
This paper uses continuous PSO optimization algorithm directly to is listed as follows:
solve the oil well production scheduling problem, which means that
the discrete variables are relaxed to continuous space. In this paper,
vkþ1
it ¼ ωvkit þ c1 ξ ykit −ukit þ c2 η ykgt −ukit þ c3 ψ ukmt −uknt ð14Þ
the discrete variables are first mapped in the interval [0, 10] within
randomly-generated real number of primary particles, and then
decoded to turning-on/off status of discrete oil wells. Particle with N where m and n denote the m-th and the n-th units in the particle,
oil wells and T cycles can be expressed as follows: respectively, k represents the iteration of the algorithm and itself,
ξ , η , ψ ∈ U[0, 1] are random numbers in the interval.
2 3
s11 s12 ⋯ s1T 3.3. Repair turn on/off status
6 s21 s221 ⋯ s2T 7
6
S¼4 7 ð8Þ
⋮ ⋮ ⋮ ⋮ 5 The turn on/off status are repaired to meet the minimum up
sN1 sN2 ⋯ sNT
and down time constraints. The continuous turning on time T on
it
1426 J. Lang, J. Zhao / Chinese Journal of Chemical Engineering 24 (2016) 1423–1430
(3) If the previous cases are not applicable, find an oil well that well as a feasible solution of the primal problem. The demand coupling
does not meet the minimum off time and back forward to constraints are relaxed and embedded as a penalty term multiplied by
time period when it is turned on. Then, turn on the oil well, Lagrangian multipliers in the objective function, leading to a relaxed
as shown in Fig. 6. version of the primal problem, which can be decomposed into several
independent subproblems easier to solve. Here, Lagrangian multiplier
{λkt} is associated with Eq. (2). Thus, the Lagrangian problem is formu-
0 1 1 1 0 0 0 1 1 1 lated as follows:
(LR): L(λ) = min FLR with
0 1
0 1 1 1 1 1 1 1 1 1 X
N X
T X
K X
T X
F LR ¼ ηi qit zit þ θi zit 1−zi;t−1 − λkt @ qit −Dkt A
i¼1 t¼1 k¼1 t¼1 i∈Ωk
Fig. 6. Third case of oil well turns on. N X
X X
K X X X
K X X
N X
T T T T
¼ ηi qit zit − λkt qit þ λkt Dkt þ θi zit 1−zi;t−1
i¼1 t¼1
0 k¼1 t¼1 i∈Ωk k¼1 t¼1
1 i¼1 t¼1
XK X T X X X
K X
T
3.5. IPSO procedure ¼ @ η zit −λkt qit þ η qit zit A þ λkt Dkt
i i
k¼1 t¼1 i∈Ωk i∉Ωk k¼1 t¼1
By the above heuristic, we can obtain the turn on/off status and oil
XN X
T
production of oil wells that meet all the constraints. Thus, the fitness þ θi zit 1−zi;t−1
value of particles can be determined. These are the steps of IPSO. i¼1 t¼1
The steps of IPSO to solve the oil well scheduling problem are as ð18Þ
follows:
subject to Eqs. (3)–(7).
Step 1: Randomly generates a population with population size M. Based on the duality theory, the LR method tries to maximize the
The value of each individual is in [0, 10] real number matrix and dual objective function. The Lagrangian dual problem is
the scale is N× T, k = 1. (LD): Max L(λ) with
0 1
Step 2: Eq. (8) is decoded into a particle turn on/off status to satisfy
XK X
T X X
Eq. (3). A portion of the particles is adjusted according to the priority LðλÞ ≡ min @ ηi zit −λkt qit þ ηi qit zit A ð19Þ
order to get a better on/off status order. k¼1 t¼1 i∈Ωk i∉Ωk
Step 4: Get the fitness value of each particle. If the fitness of the new þ θi zit 1−zi;t−1 þ λkt Dkt
i¼1 t¼1 k¼1 t¼1
particle is better than the local optimal value ykin, then update the ykin
using the fitness value of this new particle. If ykin is better than the subject to constraints (3)–(7).
global optimal value ykgn, then update the value of ykgn using the The subgradient optimization method [15–16] is used to update La-
value of ykin. grangian multipliers λkt at each iteration given by λhkt+ 1 = λhkt + αhgλt,
Step 5: Using Eqs. (14) and (12) to update the particle velocity. where h is the iteration index and gλt is the subgradient of L equal
Step 6: According to Eq. (11) to change the position of the individual, to g λt ¼ Dkt − ∑ qit α h is the step size defined as
i∈Ωk
out of range of the particles is processed using Eq. (13), and re-
encoded and mapped into turn on/off status by Eq. (8).
L −Lh
Step 7: If k b K, k = k + 1, go to Step 2, otherwise go to Step 8. h
α ¼ γ 2 ; 0 b γ b 2: ð20Þ
Step 8: Stop. g h
λt
4. Results and Analysis The results obtained by the relaxed problem are usually not
feasible to the original problem. Thus, a heuristic method is used
4.1. Experimental parameter settings to generate a good feasible solution based on the results of relaxed
problem. Because non-aftereffect property of dynamic program-
The algorithms are coded with C ++ programming language and ming is not satisfied, CPLEX is used to solve subproblem. The dif-
implemented in the development environment of Visual Studio 2005. ference in value between the solution for the primal problem
All the computational experiments are performed on a computer with and the optimum solution for the dual problem yields the duality
Intel Core i5 3.1 GHz CPU and 4 G RAM. gap. The relative duality, gap σ = (Z – L*)/L⁎,is selected as the per-
Supposing the scheduling horizon is 24 h, the experimental data formance of the algorithm. σ = 0.5 is chosen in this paper. The
are randomly generated according to practical production data. The value of the dual function L ⁎ provides a LR lower bound for the
minimum and maximum oil productions of oil well follow uniformed primal problem. A detailed description of the general LR method
distributions which satisfy corresponding at interval between 30 and can be found [17].
100 and the interval between 800 and 1200, respectively. The mini-
mum shut on and off time is uniformed distribution which satisfies 4.3. Comparison of experiment results
interval between 1 and 6. The experiments are tested using 10
sizes with each size of 10 sets of data. The data in the table is the av- In order to verify the effectiveness of the proposed IPSO algorithm,
erage value of 10 set of data for each size. The running time of CPLEX different tests are designed to evaluate the running time and objective
is 100 s. function value of the algorithm. (1) CPLEX. (2) LR algorithm. (3) Binary
particle swarm algorithm, marked as BPSO. The results are shown in
4.2. Lagrangian relaxation method Tables 1 and 2. The numerical value in Table 2 is the ratio that is:
ARi = Zi/ min {Z1, Z2, Z3, ⋯ , Z6}, (i = 1, 2, 3, ⋯ , 6), where Z1/Z2 expresses
In order to evaluate the performance of the IPSO algorithm, the objective function value of CPLEX obtained in solving the single model
Lagrangian relaxation algorithm is applied to obtain a lower bound as and batch model, Z3/Z4 expresses the basic objective function value of
1428 J. Lang, J. Zhao / Chinese Journal of Chemical Engineering 24 (2016) 1423–1430
Table 1 Table 3
Computational result of the proposed model and algorithm Comparison of results for LR and IPSO
Table 2 objective value obtained by IPSO and LR upper bound is 1%, and the av-
Computational time of the proposed model and algorithm erage deviation with LR lower bound is 3.84%. For the batch model, de-
CPLEX BPSO IPSO fine IMP3 = 100 × (UB − Z)/UB and IMP2 =100× (Z − LB)/Z. The results
N show that the average deviation between the objective values obtained
(Single) (Batch) (Single) (Batch) (Single) (Batch)
by IPSO and LR upper bound is 0.95%, and the average deviation with LR
10 57.88 0.02 1.18 0.30 0.86 0.23
lower bound is 3.93%. From the results obtained, it is observed that the
30 90.53 2.41 3.28 3.06 1.28 0.30
50 100.13 36.86 4.93 2.44 2.00 0.47 proposed IPSO algorithm can obtain better solutions compared with the
70 100.05 78.14 6.87 3.51 2.61 0.48 LR algorithm.
90 100.13 90.33 8.80 0.91 3.25 0.56 Figs. 7 and 8 show that the efficiency of proposed model is obviously
100 100.30 80.54 9.76 2.01 3.79 0.59 better than the single-machine model. The proposed IPSO has higher
300 100.14 100.03 29.46 2.58 11.19 1.33
precision, which is obviously better than the other algorithms on com-
500 100.32 100.05 50.34 4.10 18.86 2.01
700 104.59 100.10 71.28 5.64 27.07 2.68 putational time.
900 107.94 100.43 92.22 7.16 35.05 3.37
1100 102.35 100.12 113.2 10.21 51.24 4.87
1300 101.24 100.21 145.21 16.24 77.25 6.99
1500 100.14 101.16 189.36 27.25 97.47 10.27
Ave 97.36 76.18 55.84 6.57 25.53 2.63
BPSO obtained in solving the single model and batch model and Z5/Z6
expresses the objective function value of IPSO obtained in solving the
single model and batch model.
From the results in Tables 1 and 2, the following conclusions can be
obtained:
Table 4
Computational results of the proposed model and algorithm
Table 5
Computational results of the LR and IPSO
LR IPSO IMP
Obj. 249093.27 236982.75 251235.06 239231.32 240322.83 245797.03 1.39 3.5 2.66 2.18
Table 6 Nomenclature
Comparison of computational time
αi0 , ai1 parameter for calculating the pressure increase of well i when
LR IPSO it has been shut in cost coefficients depend on number of wells
Items
Single Batch Single Batch bi0 ,bi1 parameter for calculating the pressure decrease of well i
when it is producing
CPU time 60.81 4.82 2.85 0.45
Dkt the oil demand in time period t for k batch D1t + D2t +
⋯ + Dkt ≥Dt
DPit pressure differential in the well bore when the well i is
producing
5. Conclusions
Goff
it duration of i well is continuously shut down at the t hour
Gonit duration of i well is continuously open at the t hour
This paper focused on the oil well scheduling problem in the prac-
k index for number of batches, k = 1 , 2 , ⋯ , K, where K is the
tical oilfield production. The practical production constraints, such as
total number of batches
the well group's demand for oil production, the restrained pressure
M suitable upper limit for the residual of the equations involving
in the oil well bottom and minimum up and down constraints were
pressure
taken into consideration. A nonlinear mixed-integer programing
qmax
i , qmin
i maximum\minimum allowable oil production of well i
model was established for the problem which aimed to minimize
qit oil production of well i and period t
the cost of oil production and start-up cost. An improved PSO algo-
Rit zero–one variable indicating whether the well bore pressure
rithm was designed to solve the problem. The algorithm was operat-
reaches the maximum allowable value in period t when
ed by mapping discrete variable to the continuous relaxation space,
well i is shut in
changing coded scheme into real coding, and then carrying out the
experimental analysis for the selection of continuous space and ve- Rit ¼ 1 reaches the maximum allowable value
locity range by decoding turn on/off status. Besides, a new velocity Rit ¼ 0 not reaches the maximum allowable value
updating formula was proposed to improve the performance of the
Sit zero–one variable indicating whether the well bore pressure
IPSO algorithm. The performance of the proposed IPSO were com-
go beyond the maximum allowable value in period t when
pared with LR, BPSO and the commercial solver CPLEX by the exper-
well i is shut in
iment results of randomly generated instances, which demonstrated
that the proposed improvement strategies were effective for PSO Sit ¼ 1 not beyond the maximum allowable value
algorithm. Sit ¼ 0 beyond the maximum allowable value
1430 J. Lang, J. Zhao / Chinese Journal of Chemical Engineering 24 (2016) 1423–1430
Tioff minimum down time of well i [4] M.C.A. Carvalho, J.M. Pinto, A bi-level decomposition technique for the optimal plan-
ning of offshore platforms, Braz. J. Chem. Eng. 23 (2006) 1–11.
Tion minimum up time of well i [5] R.R. Iyer, I.E. Grossmann, S. Vasantharajan, A.S. Cullick, Optimal planning and sched-
Ton
i0 initial up time of well i uling of offshore oil field infrastructure investment and operations, Ind. Eng. Chem.
t index for time periods t = 0 , 1 , ⋯ , T Res. 37 (1998) 1380–1397.
[6] V.D. Kosmidis, J.D. Perkins, E.N. Pistikopoulos, A mixed integer optimization formu-
UPit pressure differential in the well bore when the well i is shut in lation for the well scheduling problem on petroleum fields, Comput. Chem. Eng. 29
xin
it well bore pressure of well i at the beginning of period t (7) (2005) 1523–1541.
xmax
i , xmin
i maximum\minimum allowable pressure of well i [7] M.J. Bohannon, A linear programming model for optimum development of multi-
out reservoir pipeline systems, J. Pet. Technol. 22 (11) (1970) 1429–1436.
xit well bore pressure of well i at the end of period t
[8] H. Bieker, O. Slupphaug, T. Johansen, Real-time production optimization of shore oil
ybin i0 initial well bore pressure of well i and gas production systems: a technology survey, SPE Prod. Oper. 22 (2007).
zit zero–one decision variable of well indicating whether well i is [9] V. Gunnerud, B. Foss, Oil production optimization — a piecewise linear model, solved
with two decomposition strategies, Comput. Chem. Eng. 34 (2010) 1803–1812.
up (1) or down (0) in time period t
[10] N. Ulstein, B. Nygreen, J. Sagli, Tactical planning of offshore petroleum production,
ηi cost coefficients Eur. J. Oper. Res. 176 (2005).
θi start-up costs of each unit in time period t [11] B. Güyagüler, T. Byer, A new rate-allocation-optimization framework, SPE Prod. Oper.
Ω set of all oil wells, Ω ={1, 2, ⋯ , N} where N is the total number 23 (2008) 448–457.
[12] R.C. Eberhart, J. Kennedy, A new optimizer using particle swarm theory, Proceeding
of oil wells of 6th international symposium on Micromachine and Human Science, Nagoya,
Ωk set of all oil wells in the kth batch, Ω1 ∪ Ω2 ∪ ⋯ ∪ ΩK = Ω, Japan 1995, pp. 39–43.
Ωi ∩ Ωj =Φ, i = 1 , 2 , ⋯ , K, j =1 , 2 , ⋯ ,K, i ≠ j [13] J. Kennedy, R.C. Eberhart, Particle swarm optimization, Proceeding IEEE internation-
al conference neural networks 1995, pp. 1942–1948.
[14] C.J. Liao, C.T. Tseng, P. Luarn, A discrete version of particle swarm optimization for
References flowshop scheduling problems, Comput. Oper. Res. 34 (2007) 3099–3111.
[15] J.F. Shapiro, Generalized Lagrange multipliers in integer programming, Oper. Res. 19
[1] L.X. Tang, G.S. Wang, Z.L. Chen, Integrated charge batching and casting width selec- (1971) 68–76.
tion at Baosteel, Oper. Res. 62 (4) (2014) 772–787. [16] M.L. Fisher, The Lagrange relaxation method for solving integer programming prob-
[2] L.X. Tang, Y. Zhao, J.Y. Liu, An improved differential evolution algorithm for practical lems, Manag. Sci. 27 (1981) 1–18.
dynamic scheduling in steelmaking-continuous casting production, IEEE Trans. Evol. [17] J.E. Beasley, Lagrangian relaxation, in: C.R. Reeves (Ed.), Modern heuristic tech-
Comput. 18 (2) (2014) 209–225. niques for combinatorial problems, Blackwell Scientific, Oxford 1993, pp. 243–303.
[3] A. Ortiz-Gomez, V. Rico-Ramirez, S. Hernandez-Castro, Mixed-integer multiperiod
model for the planning of oilfield production, Comput. Chem. Eng. 26 (4–5) (2002)
703–714.