ENSAM – Meknès

Module: LEAN SIX SIGMA

Study case : Optimization of the Particle Size Distribution of White Sugar Using Six Sigma
Methodology

Abstract —This study case aims to provide a complete six sigma approach to control several indicators characterizing the white sugar’s
particle size variability at the crystallization station in Morocco. We first defined our perimeter of study, and subsequently, we defined
the critical elements of quality (CTQ) of our project, namely: Average Aperture (OM) with a target of 0.6mm and a variation margin of
± 0.05 and Coefficient of Variation (CV) which varies in the range [38% -45%]. We have adopted the study of repeatability and
reproducibility and the study of the process capabilities to diagnose the initial state of the granulometry. In order to trace the root causes
of the variability in the crystallization process, we carry out an in-depth analysis of our process by adopting the experimental plans. The
desirability study identified the combination of variable parameters that jointly optimize the Average Aperture and Coefficient of
Variation. To verify the validity of the resulting model, we realized monitoring, and we recalculated the new capability indicators and
the level of quality achieved characterized by the Z process. In order to put our process under continuous control, we have carried out
control charts. At the end of this project, we quantified the gains, a 22.51% reduction in non-quality in white sugar.
Keywords- Six Sigma; DMAIC; Experimental Plans; Crystallization; Particle Size Distribution; Desirability Study; Study
Case.

I. INTRODUCTION
The opening up of the Moroccan economy to foreign trade and the exacerbation of competition on the international
scale has forced the Moroccan company to upgrade. The Moroccan sugar industry has experienced increased competition
in the world. The customer's evolving needs, privatization, and openness to the world market impose the perpetual search
for excellence. It is necessary to align with the quality characterized mainly by the Particle Size Distribution of white sugar
to improve competitiveness and meet the internal and external market needs and accompany consumer trends. It is in this
perspective that this project is Optimization of the Particle Size Distribution of White Sugar Using Six Sigma Methodology.
To do this, we adopted the six-sigma approach in order to attack the large variability in the Particle Size Distribution of
white sugar.
Six Sigma or 6 Sigma is a registered trademark of Motorola, Designating a structured management approach to
improve process quality and efficiency [1-6]. The Six Sigma method was first applied to industrial processes [7][8] before
being extended to all types of processes, including logistical [9-11], commercial, services [12-14], healthcare [15-18], and
energy-saving processes [19]. The notion of quality and variability are linked. We can even define non-quality as variability
concerning an expected reference. Now the fight against variability is one of the basic concepts of Six Sigma. The Six
Sigma approach brings its rigorous methodology in improving the quality of production and reduction of waste in the sense
of lean management. It also brings the managerial structure that makes all its strength. Six Sigma is not a method; it is also
a way to organize the business so that we can be able to achieve these significant advances in performance achieved in this
project. Since we aim to improve a current process and an existing product performance which does not meet customer
expectation, we apply the five-phase define, measure, analyze, improve and control (DMAIC) methodology. To define our
project, we used the CTQ diagram, which translates the needs of the client into measurable requirements, namely the
Average Aperture, noted OM, whose target is 0.6mm with a tolerance of ± 0.05mm and the Coefficient of Variation, noted
CV, varying in the Interval [38% -45%]. We studied the performance of our measurement system by studying repeatability
and reproducibility as well as the performance of the crystallization process and precisely the performance of the first jet
by studying the capabilities. To analyze the influencing factors, we used experimental designs, which allowed us to emerge
with two models satisfying the two requirements of the customer—the desirability study made possible the optimization
of both parameters (OM, CV) together. In order to put the process under continuous control, we proposed control cards.

II. DEFINE THE PROBLEM

2.1. Problematic
The crystallization station provides the extraction of sugar crystals in syrup through three successive jets. The draft aims
to extract maximum crystals; it represents 70% of the overall yield and a purity of 99.99%. Particle Size Distribution is an
essential characteristic of sugar. It is not only a quality feature but also an important technological parameter. Indeed, the
grains size distribution is essential insofar as it is a decisive factor for the storage in silos, for its dispatch, and the molding
of the sugar in pieces. Indeed, sugar is better appreciated when its crystals are uniform and regular sizes. This work involves
reducing the variability of white sugar by controlling the Particle Size Distribution parameters characterized by:
Average Aperture (OM) gives the average size of the crystals. It represents the sieve opening, passing half of the grains
and retaining the other half [1].
Coefficient of Variation (CV) is a standard deviation expressed as a percentage, representing the dispersion of the
measurements around the mean opening.

Ibtissam EL HASSANI Filière GI-IADS

Our project aims to achieve an average opening (OM) 0.6 with a tolerance of ± 0.05mm and a coefficient of variation (CV)
in the range [38% -45%]. The current project concerns only crystallizers of the first jet; the other crystallizers come out of
our perimeter of the study illustrated in figure 3.

The non-homogeneity of the grains and the variability in the grain size
What?
distribution of white sugar

Who? All project stakeholders: Group customers; Refining Service; Storage

Where?
Crystallization station

When? Since the start of the crystallization station in 2009 and the start of particle
size analysis of white sugar

Why? The uncontrolled crystallization parameters

How will we solve the problem?

How?
By adopting the six sigma methodology

Figure 1: 5W1H Questions

Improve other indicators of

quality of white sugar

Study the quality of the raw

OUT
material
The second and third jets

Master the particle size of the Reviewing

IN
first jet working methods

Review the method of Optimize the average

analysis of the opening and the
particle size of white coefficient of variation
sugar

Figure 2: In/Out Diagram

The project is performed by studying the sources of variability in these parameters and then proposing improvements at
this level. Our project is SMART (Specific, Measurable, Achievable, Reasonable, is limited in time).

2.2. The raw sugar refining process

2
Refining is the method to have a high purity of refined sugar from raw sugar. The raw sugar refining process is as follows:

Purification:
Raw Sugar

Storage Affinage -Carbonatation Discoloration

-Filtration

Evaporation

Crystallization:
Ingot; Crystallization:
Pieces; Sugar loaves
Granular

Conditioning Conditioning

Storage and
Storage and shipping
shipping

Figure 3: The raw sugar refining process

2.3 Description of the first jet

The first jet is the most important one. Quantitatively by about 70% of all production and qualitative as the outlet sugar has
a purity equal to 99.99%. The first step at the start of crystallization is evacuation. Once the vacuum is established, we
prepare the "massecuite" by covering sufficiently the heat exchanger with the standard liquor, which corresponds to 25%
of the device's volume. The liquor is concentrated until it reaches the point of supersaturating. The beam must remain
submerged all the time. For this reason, the evaporated water is compensated by the continuous introduction of syrup,
which is regulated by a Brix meter. Once the graining point is reached (point of supersaturating), a seeding is performed to
initiate the formation of the other crystals. The seeding is carried out by minimizing the apparatus's heating vapor to avoid
the dissolution of the crystals, which corresponds to a Brix of 78% to 80%. The seeding must be followed by a stabilization
period (waiting for time or start of the magnification). It is a phase of maintaining supersaturating around the value of the
seeding point. The rise lasts about 30 to 40 minutes until the first clamping. This first clamping cooked is the most delicate
phase in so far as the risk of the conglomeration of the crystals is significant. When the maximum level is reached (clamping
phase), the syrup supply is stopped, and evaporation of the massecuite is continued. Cooking ends with the pouring and
washing of the appliance. We return the apparatus to atmospheric pressure by a vacuum breaker, and we stop the agitator,
open the drain valve, and the massecuite flows into a mixer.

2.4 Cartography of the first jet

To visualize the inputs/outputs of the first jet and the various control parameters, we have carried out the following SADT
mapping:

3
 Figure 4: SADT diagram of the first jet

2.5 Critical To Quality

Granulometry is the study of the grain size distribution of white sugar. The fineness of white sugar is characterized by:
opening Medium (OM), the Coefficient of Variation (CV), the percentage of Coarse (% GG), the percentage of fine grains
(% GF). The two indicators Medium Aperture (OM) and Coefficient of Variation (CV) are considered the most critical for
the client. The CTQ diagram below defines the critical characteristics for the client, their targets, and their limits:

Figure 5: CTQ Diagram

The black box allows to visualize the transfer function defined by the relation Y = F (A, B, C ...) which is established
between the response to optimize Y and the process parameters (A, B, C ..). We seek to demonstrate by statistical evidence
the non-random nature of the relationship between the input parameters and Y.

Figure 6: Black box of the process

4
 III. MEASURE: THE PROCESS OF MEASUREMENT AND DATA COLLECTION
3.1. Validation of measurement tools
The measuring means must be attached to the calibration chains and possess accuracy, linearity, and stability
properties. The sugar analysis method corresponds to "Method GS2 / 9 - 37 (2007) for the determination of the particle
size distribution of white sugar and plantation sugars by sieving", taken from the ICUMSA - Methods Book. This method
applies to all white sugars and brown sugars. We use several methods to determine the OM and the CV of the sugar. The
ICUMSA repository cites them: POWERS method, RENS method, RRSB method, BUTLER method. The method used
by the study laboratory is the Butler method.
The refinery's laboratory is certified in ISO 17025 version 2005, which guarantees the accuracy, linearity, and stability of
the CTQ measurement.
3.2. Study of Repeatability and Reproducibility (R & R)
We carried out the R & R study on samples taken over five days, and each sample was analyzed and repeated by
the same operator and by another operator in an independent manner. We have taken the measurements of the two
characteristics: OM and CV. The results of the study are presented in Figure 9.

Figure 7: R & R study for OM

Figure 8: R & R study for CV

5
 Figure 9 : Result of the Study R & R from Minitab Software

The ANOVA table for the repeatability and reproducibility study Minitab software shows the statistical significance of
various sources. We have noted that both the Parts and Operators factors and their interaction in the R & R study (for OM
and CV) have a P-value greater than α = 0.05, which shows that statistically, we do not have an R & R problem. In addition,
we have standardized the sampling depending on the time and frequency of molding.
3.3. Measuring and evaluating process performance
We have set up observation campaigns of the process. The results are presented on the control charts Fig. 10. We tested
the normality of the data by the Anderson-Darling test. The results are shown in Fig. 11. With a significance level of 0.05,
the Anderson-Darling normality test for OM shows that the data follow a normal distribution since P-value = 0.111> 0.05.
It is not the case for CV which does not follow the normal distribution since p <0.005 <0.05. The analysis of the capability
of the OM Fig. 12 gives Cp = 0.43; therefore, the level of capability is very insufficient. As a direct consequence, there is
a performance problem with Pp = 0.41 <1.33. We also have CPL=0.22 < CPU=0.65, CPK=0.22=CPL<CPU; So we have
a problem of decentering. The CV does not follow a normal distribution. For this reason, we used a non-normal capability
Fig. 12 study which gave: Pp = 0.26 <1.33 then we have a problem of deficient performance, and Ppk = 0.11 = PPU < PPL
<PPL then the process is Not centered.

Figure 10: Average of the Opening (OM) and Coefficient of Variation (CV) Control Charts - Minitab
6
 Figure 11: Normality Test for OM and CV - Minitab

Figure 12: OM and CV Process Capability - Minitab

IV. ANALYSE: THE FACTORS INFLUENCING PARTICLE SIZE DISTRIBUTION

During the diagnosis of the crystallization station, we were able to identify several factors as parameters that could
influence the Particle Size Distribution of white sugar, which are:
1. The quantity of seed introduced for graining;
2. Cooking vacuum temperature/pressure;
3. The grinding time of the sugar used for the preparation of the seed / average opening (OM) of the seed sugar;
4. The amount of isopropyl alcohol concerning the amount of sugar used in the preparation of the semen;
5. Vapor pressure during the rising phase;
6. The speed of agitation;
7. The duration of the magnification phase (waiting time);
8. The Brix of graining (the supersaturating indicated by the Brix);
9. Vapor pressure during the magnification phase.

7
To study the effect of these parameters, we proceeded by the experimental design method except three factors that must be
fixed in advance:
• The grinding time of the sugar used for the preparation of the seed / average opening (OM) of the seed sugar =
3h30;
• The quantity of sugar of 5 kg with a total volume of 7.7 l of isopropyl alcohol (i.e., 13 l of seed obtained);
• The vapor pressure during the rising phase = 900mbar.
4.1. Screening Plans
We have chosen to work with the Plackett and Burman plan, a simplified fractional factorial plan that quickly selects the
most influential factors. It is a preliminary study to be used to prepare an optimization plan. The experimental field of the
screening plane is illustrated in Table 1. The experimental matrix obtained for 6 factors screening plane is a Hadamard
matrix with 8 experiments shown in Table 2 with both OM and CV responses.

Table 1: Factors of Study

Factors Unit Level - Level +

A. Magnification Pressure mbar 800 1000
B. Quantity of seed l 15 20
C. Magnification Time s 100 300
D. Stirring speed Tr/min 375 600
E. Brix of graining % 78 80
F. Temperature °C 72 75

Table 2: Factors and their Field of Study

Factors Responses
Test
Number A D OM CV
B (L) C (s) E (%) F (°C)
(mbar) (Tr/min) (mm) (%)

1 1000 20 300 375 80 72 0,48 41,127

2 1000 20 100 500 78 72 0,51 49,308
3 1000 15 300 375 78 75 0,49 41,206
4 800 20 100 375 80 75 0,72 36,915
5 1000 15 100 500 80 75 0,55 40,243
6 800 15 300 500 80 72 0,41 46,266
7 800 20 300 500 78 75 0,61 40,608
8 800 15 100 375 78 72 0,52 46,265

8
 Figure 13: Graphical study of the effects of OM and CV responses

According to the study, the parameters that have a statistically significant influence on the two CTQ (OM and CV) are F:
Temperature; B: Quantity of seed E: The Brix of graining.
These factors will be retained for the optimization plan.

4.2. Optimization plans

In this step, we generate a modeling plan, carry out each test, and measure the answers obtained. These data make it possible
to model OM and CV according to the factors that strongly influence the particle size distribution and their interactions.
The experimental domain of the planes is illustrated in table 3. The experimental matrix obtained for an experimental plan
of 3 factors is 23 repeated tests, that is to say, 16 tests in total. It is represented in table 4.

Factor Unit Level - Level+

B. Quantity of seed L 15 20

F. Temperature °C 72 75

E. Brix of graining % 78 80

9
 Table 3: Factors and their Field of Study

Factors influencing grain size Responses

Test Quantity of seed Temperature Brix of graining
OM (mm) CV (%)
Number (L) (°C) (%)
1 20 75 78 0,617 44,407
2 20 75 80 0,489 53,098
3 20 72 78 0,409 44,572
4 20 72 80 0,476 46,922
5 15 75 78 0,508 41,881
6 15 75 80 0,621 42,342
7 15 72 78 0,447 42,273
8 15 72 80 0,394 39,482
9 20 75 78 0,64 43,919
10 20 75 80 0,534 49,118
11 20 72 78 0,404 44,848
12 20 72 80 0,473 47,301
13 15 75 78 0,574 41,238
14 15 75 80 0,595 39,551
15 15 72 78 0,463 43,1
16 15 72 80 0,395 44,54
Table 4: Experimental Matrix

4.3. Results of the modeling plan

Figures 14, 15, and 15 respectively show the graphs of the effects, the graphs of the interactions, and the Pareto diagrams
to see the factors and the interactions with statistically significant influence.
OM= 0.50244 + 0.06981 B - 0.03906 (A*B*C)
CV= 44.287 + 2.486 A + 1.329 (A*C)
Such as: A: temperature; B: Quantity of seed and C: Brix graining.

Figure 14: Effects graphic for OM & CV - Minitab

10
 Figure 15: Graph of interactions for OM & CV – Minitab

Figure 16: Pareto Diagrams

The previous study made it possible to have the mathematical models for the OM and the CV. As previously stated, we
want to optimize both responses at once; OM should be between 0.55 mm and 0.65 mm and CV between 38% and 45%.
So we do a desirability study.
V. IMPROVE: OPTIMAL POINT
4.4. Desirability Study

11
The composite desirability study makes it possible to find the optimal conditions to bring us closer to the desired answers.
We have chosen the two responses' importance value since both presents the same level of importance for the clients. Table
5 shows the parameters of the desirability function. This study allowed us to identify the combination of the variable
parameters that jointly optimize the two responses fig.17. The desirability value (0.9891) is close to 1, which indicates that
the parameters appear to reach satisfactory results for all the responses. The individual desirability for OM and CV is
0.99798 and 0.98021, respectively. This study gave the optimal values for the three factors presented in Table 6.

Table 5: Parameters of the desirability function

Figure 17: Desirability study for the first jet –Minitab

factors Predicted Responses Desirability

Temperature 74,9697°C≈75°C OM 0,6001 d(OM) 0,99798
Quantity of seed 15,303L≈15,5L CV 41,5693 d(CV) 0,98021
Brix graining 79,99%≈80% D global 0,9891
Table 6: Results of the desirability study for the first jet

To ensure the validity of the values obtained by all of the above studies, we chose to perform tests by the different values
of the parameters obtained. Table 7 shows some of the results. After fixing the parameters in their optimal values, we
observed that OM and CV are within the tolerance intervals. The boxplot of fig. 18 compares the distribution of the two
responses before and after the improvements.

Factors influencing grain size Responses

Test Quantity Vacuum Brix of Rise Quantity of

Grinding OM CV
Numb of seed temperature graining pressure sugar / Qty
time (h) (mm) (%)
er (L) (°C) (%) (mbar) Alcohol (L)
41.5
1 15 74,5 80 900 3,5 13 0,589
02
39.3
2 15,5 74,7 80 900 3,5 13 0,594
31
40.2
3 15 74,9 80 900 3,5 13 0,631
25
41.0
4 16 75 80 900 3,5 13 0,611
03

12
 41.1
5 16,5 74,7 80 900 3,5 13 0,597
23
42.0
6 15,5 75 80 900 3,5 13 0,601
15
41.9
7 16,5 74,5 80 900 3,5 13 0,618
17
41.6
8 16 75 80 900 3,5 13 0,581
22
41.8
9 15 74,8 80 900 3,5 13 0,571
51
Table 7: Parameter Check Plan

Figure 18: Boxplot of OM & CV before and after improvements

4.5. New Process Capability

We made 154 samples over 14 days, 11 samples per day to see the percentage of values that are now statistically included
within the OM and CV tolerance range. We have recalculated the new capability of the process. For the Average Opening
(OM): We have 1.00 <Cp = 1.04 <1.33, so our process is improved compared to the initial state, whose value was 0.43. So
we have Cpk = CPL≈CPU = 1.03, so our process is centered around the mean OM = 0.6mm. For the Coefficient of Variation
(CV): We obtained a capability of Cp = 1.61, which implies that our process has improved.

OM (mm) CV (%)

Figure 19: New Process Capabilities for Both Responses

VI. CONTROL AND STANDARDIZE

The standardization of the process control parameters ensures that the project lasts in time and that the proven solutions
continue to be applied correctly within the company.
5.1. The sustainability Audit
It aims to ensure that CTQ (Critical to Quality), which are OM and CV, are well measured and monitored, that a minimum
of formalization exists and that the procedures are applied correctly. This audit will enable the work team to ensure the
sustainability of the results of this project. We propose to do a monthly audit following an audit grid and see the evolution
of the Particle Size Distribution of the white sugar daily and to act in case there is a problem as soon as possible.
5.2. Control Charts
To follow daily the process and indicators and continuously carry out improvement actions, we realized the control charts
in a spreadsheet shown in Figure 20.

13
 Figure 20: Control Charts

VII. QUANTIFICATION OF EARNINGS

6.1. Z process
To compare the initial state and the current state, we compare the Z-process (quality level achieved) of the two responses
OM and CV. The Z process of OM and CV were respectively 0.54 and -0.06. After the study and implementation of an
optimal model and the standardization of the values of the parameters influencing the white sugar granulometric state, we
reached a level of quality Z of OM and CV of 2.40 and 2.74. Table 8 summarizes the qualitative gains from this project.

Average Aperture (OM) Coefficient of Variation (CV)

Before Improvement After Improvement Before Improvement After Improvement
CP 0.43 1.17 1.05
Pp 0.41 1.00 0.26 1.06
Zprocess 0.54 2.40 -0.06 2.74
Table 8: Summary of qualitative gains

6.2. Money earnings

White sugar that meets quality requirements increased from 69.7% to 92.21%. The refinery reaches a daily output of 2650
tons; Of which 50% of its capacity is destined for export, i.e., 1325 tones. The total cost of producing one ton of granulated
white sugar is $ 700/ton. We then gained a cost of non-quality equal to (92.21%-69.7%)*1325 tones*700$/Tone =
0.208M$/day =76.2 Millions $/ year.

VIII. CONCLUSION
In this case study of the particle size distribution of white sugar optimization within the refinery by the Six Sigma approach,
we aim to improve the quality of white sugar mainly characterized by indicators Opening Average (OM) And Coefficient
of Variation (CV). To achieve our goal, we adopted the Six Sigma approach through the six steps of the DMAICS. We
have first defined the critical indicators of white sugar quality: OM, whose target is 0.6mm with a tolerance of ± 0.05mm
CV varying in the range [38% - 45%]. We then studied the performance of our measurement system by studying
repeatability and reproducibility as well as the performance of the crystallization process of the first jet by studying
capability and performance indicators. We used the experimental design and the desirability study to analyze the influence
factors, optimizing both parameters (OM, CV). To put the process under continuous control, we proposed control charts.
Implementing the standardization sheets in parallel with the proposed audit schedule will ensure the sustainability of the
improvements. The results were considerable:
The capability has increased:
From 0.41 to 1.00 for the OM.
And from 0.26 to 1.06 for the CV.
The Z value has increased:
from 0.34 to 2.40 for the OM.
And from -0.06 to 2.74 for the CV.
It translates directly into gains of 22.51% of production, which costs MAD 76.2M$/ year. In order to ensure the
sustainability of this success and establish a continuous improvement loop, we propose to study the other two jets to
improve the total efficiency of the crystallization station.

REFERENCES

[1] Pande, Peter, Robert Neuman, and Roland Cavanagh. “The Six Sigma way: How to maximize the impact of your
change and improvement efforts”. McGraw Hill Professional, (2014).
[2] George, Mark O. “The lean six sigma guide to doing more with less: cut costs, reduce waste, and lower your overhead”.
John Wiley & Sons, (2010).
[3] Brussee, W. "Statistics for six sigma made easy. Cause-and-effect fishbone diagram. Kindle Book." (2004).
[4] Michael, L. G. "The Lean Six Sigma Pocket Toolbook: A Quick Reference Guide to Nearly 100 Tools for Improving
Process Quality, Speed, and Complexity." (2005).
[5] George, Michael L., and Mike George. “Lean six sigma for service”. New York, NY: McGraw-Hill, (2003).
14
[6] Eckes, George. The Six Sigma revolution: How General Electric and others turned process into profits. John Wiley &
Sons, (2002).
[7] Tjahjono, Benny, et al. "Six Sigma: a literature review." International Journal of Lean Six Sigma 1.3 (2010): 216-233.
[8] Antony, Jiju, and Craig Fergusson. "Six Sigma in the software industry: results from a pilot study." Managerial
Auditing Journal 19.8 (2004): 1025-1032.
[9] Knowles, Graeme, et al. "A conceptual model for the application of Six Sigma methodologies to supply chain
improvement." International Journal of Logistics: Research and Applications 8.1 (2005): 51-65.
[10] Christopher, Martin, and Christine Rutherford. "Creating supply chain resilience through agile six sigma." Critical eye
7.1 (2004): 24-28.
[11] Dasgupta, Tirthankar. "Using the six-sigma metric to measure and improve the performance of a supply chain." Total
Quality Management & Business Excellence 14.3 (2003): 355-366.
[12] Antony, Jiju. "Six sigma for service processes." Business Process Management Journal 12.2 (2006): 234-248.
[13] Laureani, Alessandro, and Jiju Antony. "Reducing employees' turnover in transactional services: a Lean Six Sigma
case study." International Journal of Productivity and Performance Management 59.7 (2010): 688-700.
[14] Chakrabarty, Ayon, and Kay Chuan Tan. "The current state of six sigma application in services." Managing Service
Quality: An International Journal 17.2 (2007): 194-208.
[15] DelliFraine, Jami L., James R. Langabeer, and Ingrid M. Nembhard. "Assessing the evidence of Six Sigma and Lean
in the health care industry." Quality Management in Healthcare 19.3 (2010): 211-225.
[16] Benedetto, Anthony R. "Adapting manufacturing-based Six Sigma methodology to the service environment of a
radiology film library." Journal of Healthcare Management 48.4 (2003): 263.
[17] Powell, A. E., R. K. Rushmer, and H. T. O. Davies. A systematic narrative review of quality improvement models in
health care. NHS Quality Improvement Scotland, 2009.
[18] Proudlove, Nathan, Claire Moxham, and Ruth Boaden. "Lessons for lean in healthcare from using six sigma in the
NHS." Public Money and Management 28.1 (2008): 27-34.
[19] Garza-Reyes, Jose Arturo. "Green lean and the need for Six Sigma." International Journal of Lean Six Sigma 6.3
(2015): 226-248.

15