SinkrOn : Jurnal dan Penelitian Teknik Informatika

Volume 4, Number 2, April 2020 e-ISSN : 2541-2019

DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

Analysis of Taklinear Performance and Integer

Linear Programming Models in Nurses
Scheduling Problems
Junerdi Nababan Tulus
University Sumatera Utara University Sumatera Utara
Medan, Indonesia Medan, Indonesia
[email protected] [email protected]

Zakarias Situmorang
University Sumatera Utara
Medan, Indonesia
[email protected]

Submitted: Mar 13, 2020

Accepted: Mar 26, 2020
Published: Apr 6, 2020

Abstract— Almost all hospitals schedule nurses' work shifts manually, which is not
effective, mistakes in scheduling nurses in hospitals can make them not work optimally
so that they are prone to making mistakes and will endanger every patient in the hospital.
In the research will be proposed integer linear programming model and branch and
bound method, the purpose of this study is to develop an optimization model of nurse
scheduling problems at the hospital. The optimization model will minimize the total
deviation of nurses' working days from the standard workdays, the optimization model
implemented in hospitals is processed using LINGO software.

Keywords— Nursing Scheduling Issues, Integer Linear Programming, LINGO.

I. INTRODUCTION In this study, the authors tried to solve the

The problem of nurse scheduling is an important problem of scheduling nurses applying the Integer
issue that must be considered in every hospital, good Linear Programming model and the branch and
nurse scheduling must be applied in every hospital so bound method, then processed using the LINGO
that ongoing activities are not interrupted and more application. Branch and bound algorithm is one
optimal. Errors in arranging the scheduling of nurses method that can be used to solve Integer
at the hospital can make them work less optimally so Programming cases. This method divides the problem
they are prone to making mistakes and will endanger into sub-problems that lead to solutions by forming a
every patient in the hospital. Nurse scheduling search tree structure and limiting it to achieve an
problem or known as Nurse Scheduling Problem optimal solution (Sumathi, 2016) (Bakhtiar &
(NSP) (Hakim, Bakhtiar, & Jaharuddin, 2017) (Jafari Jaharuddin, 2017). Branch and Bound algorithm
& Salmasi, 2015), NSP is a nurse scheduling problem procedures are carried out repeatedly to form a search
at the hospital in arranging the shift scheduling for tree and bounding process by determining the upper
each nurse. Nursing scheduling problems at hospitals bound and lower bound in finding the optimal
often occur because many hospitals do manual solution (Suryawan, Tastrawati, & Sari, 2016)
scheduling. Scheduling nurses manually is Literature Review
ineffective and inefficient wherein arranging the
scheduling must pay attention to the limited number (Angeline, Iryanto, & Tarigan, 2014) apply the
of nurses that exist and must pay attention to the branch and bound method in determining the
uniformity of the schedule given to each nurse. optimum amount of production at Cv. Xyz, in his

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 123
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

research, reviewed based on the amount of raw 3. Model solution

material inventory, market demand, profits, and time
A mathematical model for solving real problems
of manufacture of each pant. The results showed that
the optimal amount of production of each type of requires computer performance. To get the
scheduling model solution, LINGO software is used
pants (men's trousers, women's trousers, men's shorts,
women's shorts). (Nur & Abdal, 2016) Using the with the branch and bound method. Then the solution
obtained is the solution that most accommodates all
branch and bound and memory cut methods in
restrictions and minimizes the objective function.
determining integer linear programming solutions,
the results of the Branch and Bound and Gomory Cut The definition of notation that will be used in this
methods can be used to solve linear integer model is
programming problems. Based on the case examples
given, the same results were obtained between the Index:
Branch and Bound and Gomory Cut methods, where i = Nurse (i = 1, 2, …, I).
the company had to produce 47 A flashlights, 62 B
flashlights, and 172 C flashlights in order to produce j = Shift (j = 1, …, J).
an optimum profit of 2,233,000. In the process of 1 = Morning Shift (pukul 07.00-14.00).
completion, the branch and bound method require a
large number of simplex iterations and a log time but 2 = Afternoon Shift (pukul 14.00-21.00).
more convergence is guaranteed. (Suryawan, 3 = Night Shift (pukul 21.00-07.00).
Tastrawati, & Sari, 2016) Applying the branch and
bound algorithm method in optimizing bread k = Day (k = 1, 2, …, K).
production results in a 25.2% increase in profits
T = Total working days in the one month
through calculations by applying the branch and
assignment period.
bound algorithm. (Rafeek & Siswanto, 2015) Make
an ITS course schedule and then develop an integer a = The time constant in the day for one-week
programming (ILP) method to solve NP-Complete assignment.
problems using LINGO software. (Hasan & Arefin,
2017) b = The time constant in hours for one week of
assignment.
II. PROPOSED METHOD c = Constant length of work time for each shift.
1. Datasheet
djk = The number of nurses working for each shift
The data used in this study were taken from Cut j on day k.
Nyak Dien Tapak Tuan Aceh Selatan Hospital, taking
Variabel keputusan:
into account the rules set at the hospital, each nurse
would be scheduled with a fair amount of work and jika perawat i bekerja di shift j pada
shift where all nurses who worked no more than 22
days work, nurses work no more than one shift a day, hari k. selainnya
each nurse works no more than 6 days a week, each 𝑦1𝑖 = Deviation of the shortage of workdays for
nurse works no more than 48 hours a week, if every each nurse i (slack variable)
nurse on duty at the night shift should not be followed
by a morning shift the next day, the nurse's schedule 𝑦2𝑖 = Deviation of excess workdays for each
must meet the minimum number of nurses needed in nurse i (variable surplus).
each shift every day, nurses who work on the night Fungsi tujuan adalah meminimumkan total
shift for 2 consecutive days in one week then the next deviasi dari kendala penyimpangan hari kerja
day is given a day off, and each nurse works no more perawat.
than 2-night shift in one week.
𝑙
2. Modeling Min 𝑧 = ∑ 𝑦1𝑖 + 𝑦2𝑖
After the stage of formulating the problem, the 𝑖=1
next step is to represent the problem in the
mathematical model. Through this model, the Constraints::
problem is described as a system of equations or other 1. Total of the number of workdays per nurse
inequalities and mathematical expressions. Nurse
and deviation of workdays in one month T
work scheduling problems can be modeled as Integer
days,
Linear Programming (ILP) (KOÇ & AKTAN, 2019).

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 124
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

𝐽 𝐾 9. The head nurse gets a vacation every

(∑ ∑ 𝑥𝑖𝑗𝑘 ) + 𝑦1𝑖 − 𝑦2𝑖 = 𝑇 ∀𝑖 = 1, … , 𝐼 Sunday,
𝑗=1 𝑘=1 𝑥1𝑗(7𝑘−𝑑+1) = 0 ∀𝑗 = 1, … , 𝐽, ∀𝑘
2. Each nurse works no more than one shift = 1, … . ,4, 𝑑 = 1, … , 7
every day, 10. Limitation of negative and integer.
𝐽
𝑥𝑖𝑗𝑘 ∈ {0,1} ∀𝑖 = 1, … , 𝐼, ∀𝑘
∑ 𝑥𝑖𝑗𝑘 ≤ 1 ∀𝑖 = 1, … , 𝐼, ∀𝑘 = 1, … , 24 = 1, … . , 𝐽, ∀𝑘 = 1, … , 𝐾
𝑗=1

3. Each nurse works no more than a day a

week, III. RESULT AND DISCUSSION
𝐽 7 1. Nurse Scheduling Results in several Rooms.
∑ ∑ 𝑥𝑖𝑗(𝑘−𝑡+7) ≤ 𝑎 ∀𝑖 = 1, … , 𝐼, ∀𝑘 After the mathematical model is formulated with
𝑗=1 𝑡=1 the form of Integer Linear Programming, then
= 1, … . , 24 processed using LINGO with the branch and bound
method the nurses work schedule is generated for the
4. Each nurse works no more than b hours a emergency room nursing room, class, and ward at the
week with long hours of work c hours per
General Hospital in one month with a minimum
shift t, deviation of working days. Nurses' work schedules in
𝐽 7 each part of the room from the modeling results
∑ ∑ 𝑐. 𝑥𝑖𝑗(𝑘−𝑡+7) ≤ 𝑏 ∀𝑖 = 1, … , 𝐼, ∀𝑘 provide information on comparing working day
deviations between manual scheduling and the new
𝑗=1 𝑡=1
= 1, … . , 24 scheduling in September, as well as the total
workdays of each nurse on morning, evening and
5. Every nurse on duty at the night shift may night shifts. The need for a minimum number of
not be followed by the morning shift the next nurses in each part of the room in the General
day. Hospital from the modeling results is in accordance
with the number of nurses needed. Working days and
𝑥𝑖3𝑘 + 𝑥𝑙1(𝑘+1) ≤ 1 ∀𝑖 = 1, … , 𝐼, ∀𝑘
holidays for each nurse has fulfilled the hospital
= 1, … . , 𝐾 management's requirements.
6. The nurse schedules must meet the need for Implementation of the model that has been
a minimum number of nurses in each shift obtained is done by means of model simulation. The
each day, simulation uses data on the number of nurses and the
𝐼 need for the number of nurses every day in each
∑ 𝑥𝑖𝑗𝑘 ≥ 𝑑𝑗𝑘 ∀𝑗 = 1, … , 𝐽, ∀𝑘 section of the room at the General Hospital.
𝑖=1 The schedule arranged meets the rules set,
= 1, … . , 𝐾 namely:
7. If each nurse is on night shifts for two  All nurses who work do not deviate from 22
consecutive days, then the next day off, working days.
3
 Nurses work no more than one shift in one
𝑥𝑖3𝑘 + 𝑥𝑖3(𝑘+𝑙) + ∑ 𝑥𝑖𝑗(𝑘+2) ≤ 2 ∀𝑖 day.
𝑗=1
= 1, … , 𝐼, ∀𝑘 = 1, … . , 𝐾  Each nurse works no more than 6 days a week.
8. Each nurse works no more than 2-night  Each nurse works no more than 48 hours a
shifts in one week, week.
7
 If every nurse on duty at night shift then it
∑ 𝑥𝑖3(𝑘−𝑙+7) ≤ 2 ∀𝑖 = 1, … , 𝐼, ∀𝑘 should not be followed by the morning shift
𝑡=1 the next day.
= 1, … . , 24
 The nurse schedules must meet the need for a
minimum number of nurses in each shift each
day

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 125
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

 Nurses who work on the night shift for 2 New Model Emergency Room
consecutive days in one week then the next Nurse Mor Aftern Nig Total KB KK
day is given a day off. ning oon ht
 Each nurse works no more than 2-night shifts 1 8 9 5 22 0 0
in one week. 2 11 4 7 22 0 0
 There are nurses who cannot be scheduled on 3 7 8 7 22 0 0
certain days. The head nurse is given a day off 4 9 6 7 22 0 0
on Sundays.
5 10 7 5 22 0 0
The results of mathematical modeling formulated
in the form of Integer Linear Programming and 6 8 7 7 22 0 0
processed using LINGO software using the branch 7 10 7 5 22 0 0
and bound method can be seen in the table below:
8 14 2 6 22 0 0
Table 1. Comparison of Deviations from
Emergency Room Nurse Workdays 9 10 6 6 22 0 0
10 8 7 7 22 0 0
Manual Schedule Emergency Room
Nurse Mor After Nig Total KB KK 11 9 5 8 22 0 0
ning noon ht 12 7 8 7 22 0 0
1 11 11 1 22 1 0
13 8 7 7 22 0 0
2 8 8 7 23 1 0
14 4 12 7 22 0 0
3 7 8 7 22 0 0
Amount 0 0
4 6 9 7 22 0 0
Total Deviations 0
5 6 9 6 21 0 1
6 7 7 9 23 1 0 In table 1. it can be seen that a comparison of the
7 6 8 8 22 0 0 scheduling model between the manual model and the
new scheduling model in September shows that the
8 10 6 7 23 1 0 new scheduling model gives all nurses a more
9 10 4 8 22 0 0 equitable number of working days. In the new
scheduling model, there are no deviations from
10 6 10 7 23 1 0
workdays compared to manual scheduling which
11 5 10 7 22 0 0 there are deviations of 6 workdays with excess 5 days
and a deficiency of 1 workday can be seen in Table 1,
12 8 7 7 22 0 0
The new schedule considers holidays, night shifts and
13 7 6 9 22 0 0 workday needs of each nurse, so that the new
14 9 7 6 22 0 0
schedule has paid attention to the nurses fatigue
factor.
Amount 5 1
Total Deviations 6 Table 2. Comparison of Classroom Nurse
Workday Deviations.

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 126
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

Manual Schedule Class Nursing Room In table 2. it can be seen that a comparison of
Nurse Mor Aftern Nig Total KB KK scheduling models between manual models and the
ning oon ht new scheduling model in September shows that the
1 15 8 0 22 1 0 new scheduling model provides a more equal number
of working days to all nurses. In the new scheduling
2 11 6 6 23 1 0 model there are no deviations from workdays
3 9 6 7 22 0 0 compared to manual scheduling there are deviations
of 8 working days with excess 6 days and lack of 2
4 6 7 9 22 0 0 working days can be seen in Table 2. The new
5 8 7 7 22 0 0 schedule considers the holidays, night shifts and
workday needs of each nurse, so the new schedule has
6 7 7 8 22 0 0 paid attention to the nurses' fatigue factor.
7 8 5 9 22 0 0
Table 3. Comparison of Ward Room Worker Day
8 9 6 7 22 0 0 Deviations.
9 9 8 6 22 0 0 Manual Schedule Ward Room
10 8 9 5 22 0 0 Nurse Mor Aftern Nig Total KB KK
ning oon ht
11 11 4 7 22 0 0 1 20 3 0 23 1 0
12 7 6 8 21 0 1 2 10 6 7 23 1 0
13 4 11 6 21 0 1 3 7 6 8 21 0 1
14 6 9 7 22 0 0 4 5 9 8 22 0 0
Amount 6 2 5 6 8 9 23 1 0
Total Deviations 8 6 8 7 9 24 2 0
7 5 7 9 21 0 1
New Model Class Nursing Room 8 9 5 8 22 0 0
Nurse Mor Aftern Nig Total KB KK
ning oon ht 9 5 7 10 22 0 0
1 8 9 5 22 0 0 10 5 7 9 21 0 1
2 8 7 7 22 0 0
11 7 6 8 21 0 1
3 8 7 7 22 0 0 12 8 5 9 22 1 0
4 10 7 5 22 0 0 13 5 8 8 21 0 1
5 15 2 5 22 0 0 14 6 7 9 22 0 0
6 8 7 7 22 0 0 Amount 9 5
7 8 7 7 22 0 0 Total Deviations 14
8 9 6 7 22 0 0
9 4 10 8 22 0 0
10 7 11 4 22 0 0
11 9 6 7 22 0 0
12 12 4 6 22 0 0
13 9 6 7 22 0 0
14 6 7 9 22 0 0
Amount 0 0
Total Deviations 0

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 127
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

of night shifts and the need for days off

accordingly.
New Model Ward Room
Nurse Mor Aftern Nig Total KB KK  The scheduling model that produces a new
ning oon ht schedule already regulates the need for days
1 10 7 5 22 0 0 off and night shifts so that attention is paid to
the nurse's fatigue.
2 13 4 5 22 0 0
3 14 2 6 22 0 0 2. Suggestion

4 10 6 6 22 0 0  This research can be developed to solve

scheduling problems with a greater number of
5 12 3 7 22 0 0 nurses.
6 9 5 8 22 0 0
 Research can also be developed by completing
7 8 7 7 22 0 0 various cases of nurse scheduling by
considering nurse requests on night shifts and
8 11 5 6 22 0 0
days off so that the level of nurse satisfaction
9 13 3 6 22 0 0 is met.
10 11 5 6 22 0 0
11 8 6 8 22 0 0
V. REFERENCES
12 9 5 8 22 0 0
Angeline, A., Iryanto, I., & Tarigan, G. (2014).
13 11 4 7 22 0 0 Penerapan metode branch and bound dalam
14 10 5 7 22 0 0 menentukan jumlah produksi optimum pada
cv. Xyz. Saintia Matematika , 2 (2), 137-
Amount 0 0
145.
Total Deviations 0 Bakhtiar, H. T., & Jaharuddin, J. (2017). The nurse
scheduling problem: a goal programming
and nonlinear optimization approaches. IOP
In table 3. it can be seen that a comparison of
scheduling models between manual models and the Conference Series: Materials Science and
new scheduling model in September shows that the Engineering .
new scheduling model provides a more equal number Hakim, L., Bakhtiar, T., & Jaharuddin, J. (2017). The
of workdays to all nurses. In the new scheduling nurse scheduling problem: a goal
model there are no deviations from workdays programming and nonlinear optimization
compared to manual scheduling there are deviations approaches. Materials Science and
of 13 workdays with excess 9 days and a deficiency Engineering , 1-7.
of 5 workdays can be seen in Table 3. The new Hasan, M. M., & Arefin, M. R. (2017). Aplication Of
schedule considers the holidays, night shifts and Linear in Scheduling Problem. Dhaka Univ.
workday needs of each nurse so that the new schedule J. Sci , 145-150.
has paid attention to the fatigue factor of nurses. Jafari, H., & Salmasi, N. (2015). Maximizing the
IV. CONCLUSION AND SUGGESTION nurses’ preferences in nurse scheduling
problem: mathematical modeling and a
1. Conclusion meta-heuristic algorithm. J Ind Eng Int .
Based on the results, the conclusions that can be KOÇ, B. A., & AKTAN, M. (2019). The Solution of
drawn are: Nurse Scheduling Problem with Simulated
Annealing Algorithm. Journal of Scientific
 Nurse scheduling modeling can solve nurses and Engineering Research , 6 (4), 153-160.
scheduling problems that minimize the Nur, W., & Abdal, N. M. (2016). Penggunaan Metode
deviation (deviation) of each nurse's workday. Branch and Bound dan Gomory Cut.
 Development of the scheduling model with the JURNAL SAINTIFIK , 2 (1), 9-15.
rules recommended by the General Hospital in Rafeek, F. S., & Siswanto, N. (2015). Solving Course
modeling nurses scheduling more effectively. Timetable Problem by using Integer Linear
This can be seen in the new schedule for the Programming (Case Study IE Department of
number of nurses fulfilled, setting the number

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 128
 SinkrOn : Jurnal dan Penelitian Teknik Informatika
Volume 4, Number 2, April 2020 e-ISSN : 2541-2019
DOI : https://doi.org/10.33395/sinkron.v4i2.10528 p-ISSN : 2541-044X

ITS). International Seminar on Science and Suryawan, G., Tastrawati, N. K., & Sari, K. (2016).
Technology . Penerapan branch and bound algorithm
Sumathi, P. (2016). A new approach to solve linear dalam optimalisasi produksi roti. E-Jurnal
programming problem with intercept values. Matematika , 5 (4), 148-155.
Journal of Information & Optimization
Sciences .

This is an Creative Commons License This work is licensed under a Creative

Commons Attribution-NonCommercial 4.0 International License. 129