the application of statistical quality control...
TRANSCRIPT
1
THE APPLICATION OF STATISTICAL
QUALITY CONTROL TOOLS TO
MONITORING ETHANOL PROCESS
PRODUCTION.
Juliana Keiko Sagawa (UFSCAR )
Ricardo Inoue Yamada (UFSCAR )
O presente trabalho tem como objetivo apresentar a aplicação de
ferramentas de Controle Estatístico da Qualidade ao processo de
produção de etanol a partir da cana-de-açúcar. As ferramentas foram
aplicadas às etapas de Fermentação e Tratameento do Fermento em
uma usina localizada na região de Guariba, interior do estado de São
Paulo. Tais etapas apresentam alto grau de complexidade, englobando
tanto reações físicas como bioquímicas, e impactam diretamente na
eficiência da produção de Etanol. As variáveis do processo de
fermentação e de tratamento do fermento foram previamente
relacionadas e a partir de uma análise crítica e estruturada, foi
possível identificar quais etapas e variáveis necessariamente deveriam
ser monitoradas. As análises dos dados amostrais permitiram a
identificação dos índices de capabilidade do processo (Cpk). Como
contribuição, o estudo permitiu a identificação das variáveis com
maior instabilidade, o que, aliado às análises dos resultados
(Produção Total de Etanol), foi determinante para estimar os impactos
do controle para o processo, justificando assim sua aplicabilidade.
Palavras-chaves: SPC - Statistical quality control, Control charts,
Quality management, Sugar cane and Ethanol.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
2
1.1
1. Introduction
Although in most countries the censuses have indicated a decrease in the birth rate, recent
analyses published by the United Nations Population Fund (UNFPA), foresaw a projected
population growth of more than 25% by 2050. Estimates indicate that the world population
will exceed 8.9 billion by that year.
According to the Intergovernmental Panel on Climate Change (IPCC), the equation used to
measure the increasing trend in CO2 emissions and their impacts on climate changes, such as
global warming, is under direct influence of the population growth and the increase in gross
domestic product per capita worldwide.
In February of 2010, the bioethanol produced in Brazil using sugarcane was recognized by the
United States Environmental Protection Agency (EPA) as an advanced biofuel. Tests showed
a reduction in the emission of greenhouse gases by 61% compared to the emissions produced
by gas. This recognition is accorded to those initiatives that reduce the emissions of
greenhouse gases in at least 50%.
According to Costa et al. (2005), major changes in production management have been
observed over the last 60 years, but two points are worth mentioning: the first is the
advancement in technology and technological development applied to information
management, which contributed to a more efficient control of operations; the second, but no
less important, is related to the new concepts and methods of production management. These
methods started to gain prominence in the 80s, more specifically with the spread of the
concepts of quality management in the United States and Japan. Although its development has
emerged in the 20’s, the Statistical Process Control (SPC) came to be applied effectively in
the Western companies in the 80’s, when they were forced to improve their quality of its
products to better serve the demands of their consumers.
According to Martins (2010), many Brazilian companies have not yet identified the
advantages in the use of SPC to control the variations in their processes and consequently
ensure greater uniformity of their products and services.
The following paper presents a case study of the application of Statistical Quality Control
tools in the critical steps of the production process of ethanol from sugar cane. More
specifically, the process of fermentation and treatment of yeast are approached.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
3
2. Research methodology
The present research was developed according to the following steps:
• Theoretical study and literature review on Statistical Quality Control;
• Characterization of the ethanol production process;
• Selection of relevant variables of the ethanol production process;
• Implementation of Control Charts and Process Capability Analysis.
The following words were uses to search the literature: Statistical Process Control (SPC),
Sugar Cane, Ethanol, Control Charts and Statistical Quality Control. As search web sites were
used: Scielo, Virtual Libraries (USP, UNICAMP, UFSCar) and Google Scholar.
In this study, the researcher was an observer and a participant. Thus, the data collection
process was based on direct observation, meetings with the technical team responsible for the
project, formal documents, charts and informal conversations. The analysis and selection of
the critical variables of the production process were carried out by the technical team by
means of brainstorming.
3. Literature review
In the following subsections, a short literature review on statistical process control, control
charts, capability analysis is presented, as well as a characterization of alcoholic fermentation
processes.
3.1. Statistical process control
According to Oliveira (2010), the permanent monitoring of processes is needed, especially for
detecting the presence of special causes that generate disturbances in the process, also serving
as base for making decisions.
The disturbances that affect the processes may be classified into two types. Minor
perturbations caused by natural variations in process, derived from an ordinary or random
cause, represents small deviations that do not compromise or are negligible to the result.
The special causes, on the other hand, are major perturbations that can shift the average of its
target!], as well as increase its dispersion. The perturbances are usually derived from
problems or abnormal operations, are mostly related to physical conditions and structural
projects or deficiencies in standards work. Special causes of variation are caused by know
factors that lead to an unexpected change in the process output. If the process is subjected to
Special Causes of variation, the process output is not stable over time, it is not predictable.
The special causes may lead to a process shift.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
4
According to Montgomery (2004), the SPC has a powerful collection of tools for
troubleshooting that can be applied to any process, its seven main tools are: Ishikawa
(fishbone) Diagram, Check Sheet, Histogram, Pareto Chart, Scatter Diagram, Flowchart and
Control Chart.
3.2. Control chart
According Montgomery (2004), the control charts for variables are used when the monitored
variable can assume numerical values on a continuous scale, and enable the identification of
special causes in a process out of statistical control. However, an important issue that must be
mentioned is that these graphs only indicate the presence or absence of these causes; they do
not exclude the need for an analysis of which are these causes that are acting in a process and
how to eliminate them.
As is known, the chart has three horizontal lines that represent the limits previously measured
or calculated by sampling of a random variable. The Central Limit or Target (T), represents
the average value of the variable and which also corresponds to the control state. The two
other lines, positioned at the ends of the Target (T), are: Upper Specification Limit (USL) and
Lower Specification Limit (LSL), which represents the control limits that the sampling points
should be between while the process is under control.
According Montgomery (2004), in cases where it is possible to establish predefined values as
references for average and standard deviation, these values can be used for Chart of the
average X and Chart of amplitude R without the need to analyze historical database to
establish the target and the upper and lower specification limits. Generally the values of the
population mean (μ) and standard deviation (σ) must be estimated from samples taken from
the controlled process in order to calculate the control limits.
Also according to Montgomery (2004), caution is needed when the values of Mean (μ) and
standard deviation (σ) are already known and referenced, it is possible that these standards are
not really applicable to the process, so may produce many alerts out of control.
3.3. Process-capability analysis
According Montgomery (2004), the magnitude of CpK index is a direct measure of how off-
center the process is operating, in other words, it considers not only the variability of the
process, but is also sensitive to process shift. For analyzing and interpreting the CpK index
results were used reference ranges listed in Table 1.
Table 1 - Classification of processes from the CpK index.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
5
CpK Index Classification Interpretation
CpK < 1,00 Unstable Process The capacity of the process is inappropriate
to the required specification.
1,00 < Cpk < 1,33 Partially able Process The capacity of the process is within the
required specification.
Cpk ≥ 1,33 Stable Process The capacity of the process is adequate to
the required specification.
Source: Montgomery (2004)
3.4. Characterization of the alcoholic fermentation process
The ethanol extracted from sugarcane is obtained by alcoholic fermentation. It consists on a
biological process in which sugars, present in the sugarcane juice are converted into cellular
energy and thereby produce ethanol and carbon dioxide as metabolic waste products.
According Basso et al. (2001), the yeast Saccharomyces cerevisiae, popularly known as
baker's yeast, is the most common specie used for ethanol production. It is a facultative
aerobic fungus and the products obtained from sugar metabolizing vary with the
environmental conditions in which they are taken.
In anaerobic reactions, the metabolized sugar is converted into ATP, i.e. the cellular energy
necessary for survival and cellular growth of the yeast, producing ethanol and carbon dioxide.
For the best performance in the conversion of sugar into ethanol, it is important to evaluate
and control the changes in the conditions of fermentation, such as pressure, temperature, pH,
oxygenation, substrate, species, Lineage, and other contaminations (BASSO et al., 2001)
According to Lopes (2008), the fermentation process can be divided into five stages:
Lag-phase: An adaptation stage where the enzyme reconstruction and the cellular
multiplication occur; an increase in the amount of cells present in the mash is observed.
Acceleration phase: in this stage the speed of cellular multiplication increases and the
sugar in mash begins to be metabolized.
Exponential phase: as the name says, in this stage there is an exponential increase in
the number of cells, characterized by the large amount of metabolic waste products obtained,
such as Ethanol.
Stationary phase: This stage is marked by the exhaustion of nutrients and sugars
present in mash, which ensures the required energy for the emergence of new cells. As
consequence, the number of cells is kept constant.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
6
Decline phase: in this stage is observed a drop in viability of the yeast. In other words,
the number of cells that dies is bigger than the number of new cells. This happens due to a
deficiency in maintaining the necessary conditions of temperature and pH of the fermented
wine. In addition, the ethanol present in fermented wine destroys the cell membrane of yeast,
favoring infections.
According to Lopes (2008), the product obtained after fermentation goes through a centrifuge.
The yeast cream that is separated from the wine is recovered and treated with water, lowering
its concentration from 60% down to approximately 25%. Acid or an antibiotic is also added to
the yeast in order to reduce bacterial contamination. Then, it is pumped back to the yeast
treatment vat and re-added to the next fermentation.
The resulting fermented wine is sent to the distillation process, where the hydrated Ethanol is
separated from the other components with different boiling points. Chemical treatments of
dehydration can be used to reach the specifications of 99.7°GL, resulting in the anhydrous
ethanol used for blending with pure gasoline.
4. Case study – applying statistical quality tools to the ethanol production
The following items present the steps and results of the case study carried out in a chemical
company, which illustrates the application of quality tools to monitor the ethanol production
process.
4.1. Definition of the transformation steps
Initially, the macro phases of the ethanol production were identified. After analyzing the
nature of its operations, the production process was divided in three steps, as proposed in
Figure 1.
Figure 1 - Macro steps of Hydrated Ethanol production.
Source: own author
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
7
First step: the raw material mix is treated to provide favorable conditions for the
following biochemical reactions.
Second step: the mix (called mash) will be metabolized by the reactor (yeast).
Third step: the metabolized products will be taken to additional physical treatments
until reaching the desired specifications.
Although comprehensive, this representation of the production process does not present the
required level of detail to allow the identification of the critical variables of the process. Thus,
the technical group in charge of the project mapped the production process with more detail,
as shown in Figure 2.
Figure 2 - Flow of the stages of production of Ethanol.
Source: own author
As it can be seen in Figure 2, the clarified sugarcane juice, the molasses and the water are
mixed together to form the mash, which is boiled and then cooled down to a specific
temperature. After that, the yeast is added and the fermentation process occurs under
controlled conditions. The resulting mix is filtered and centrifuged, allowing the separation of
the wine from the yeast cream, finally, is distilled to yield ethanol, while the yeast is treated to
be reused in the next fermentation process, as mentioned.
4.2. Identification of the critical steps
After mapping the production process, the project group decided to specifically focus on the
fermentation and the yeast treatment processes. These steps were considered critical since the
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
8
ethanol production efficiency directly depends on the success of the fermentation reactions,
which, on their turn, require the correct control of several variables. These variables will be
discussed as follows.
4.3. Identification of critical variables
By means of a brainstorming, the technical group responsible for the project identified X
variables of influence on the fermentation process. Given the large number of variables
identified, a criterion was established to prioritize them according to their degree of relevance
to the process performance indicators. The following performance indicators were considered:
milling capacity, loss in distillation, loss in final effluent, undetermined loss, accident risks,
and costs of non-quality.
First of all, the group of experts in the area assigned a score to each performance indicator to
reflect its impact on the process results according to the following scale: 1 - Low impact / 2 -
Medium impact / 3 - High impact. Each score was given after the group reached a consensus.
Figure 3 - Calculation Methodology
Source: own author
Similarly, the experts assessed the correlation of the process variables to the performance
indicators, that is, they evaluated the extent with which a given process variable would affect
a given performance indicator. Grades were assigned according to the following scale: 0 -
Nonexistent Correlation / 1 – Weak Correlation / 3 – Median Correlation / 9 – Strong
Correlation. In order to obtain a prioritization index, a weighted sum of grades was calculated
for each variable, as shown in Equation 1 below:
Zi = (Y1. Xi1) + (Y2. Xi2) + (Y3. Xi3) + (Y4. Xi4) + (Y5. Xi5) + (Y6. Xi6) + (Y7. Xi7)
As a result of the analysis, 7 variables were classified as critical: Temperature in the
fermentation vats, Alcoholic concentration in the fermentation vats, Brix of the mash,
Temperature of the mash, Alcoholic concentration in the vats of yeast treatment, Viability in
the fermentation vats and Infection in the fermentation vats. For monitoring the critical
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
9
variables using control charts, the group previously identified the impacts of each critical
variable on the fermentation process and defined the sampling frequencies of the variables.
In addition, a bank of actions to correct deviations was created for each variable, enabling to
solve problems as quickly as possible and decentralizing the power of decision making.
5. Capability of critical variables
For the ethanol production process, previous research has shown that optimum results are
obtained if some critical variables remain between specified intervals. Thus, the specification
limits of variation for these variables were already established from empirical studies. In case
study, rather than conduct a sample analysis of historical data to establish values of Target,
Upper and Lower Specification Limits of the process were used these pre-specified values
referenced in Table 1.
Table 2 - Analysis of the index Cpk for different periods.
Source: own author
Two hundred daily samples of each variable were provided by the industrial laboratory in
order to plot the control charts. The data was divided into 4 periods, each of them with 50
daily samples on chronological order.
The variables analysis was carried out only for two periods, the worse and the better
performance in the Ethanol production, respectively represented by Period 1 and Period 2 in
Table 1. Samples suffered some interruptions due to equipments stoppage during rain periods.
This is a particularity of the ethanol production process.
An important evaluation about the monitoring importance and their effects in results of total
ethanol production is that in Period 2 was observed an increase in total production of ethanol
over 26% compared to Period 1.
The verification of special causes acting in the process can be done using Control Charts and
capability analysis.
The impacts of each critical variable in the Ethanol production process as well as the
results of capability analysis of these variables are provided bellow:
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
10
Temperature in the fermentation vat (ºC)
When below the LSL: Influences the speed
and productivity of fermentation processes.
When above the USL: Possibly reduces
cellular viability of the yeast, due to an
increased probability of infection and
flocculation in mash.
Analysis of Cpk Histogram observations
Period 1:Unstable process (0.77)
Period 2: Stable process (1.59)
Distribution with low variability, however, the
variable is off-center.
Alcoholic concentration in the fermentation vats (ºGl)
When below the LSL: Causes residual losses
in fermentation processes.
When above the USL: Possibly reduces the
cellular viability of yeast, due to excessive
exposure to high alcoholic level.
Analysis of Cpk Histogram observations
Period 1: Unstable process (0.11)
Period 2: Unstable process (0.39)
Distribution with moderate variability and off-center.
Brix of the mash (ºBrix)
When below the LSL: Reduces the
fermentation time and also the process
efficiency.
When the above USL: Increases the
concentration of alcohol on mash, influencing
cellular viability losses.
Analysis of Cpk Histogram observations
Period 1: Unstable process (0.57)
Period 2: Partially able process (1.11)
Distribution with low variability and somewhat
off-center.
Temperature of the mash (ºC)
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
11
When below the LSL: Reduces the speed of
fermentation process and consequently
decreases productivity.
When above the USL: Favors bacterial
contamination by proliferation, reducing the
cellular viability.
Analysis of Cpk Histogram observations
Period 1: Unstable process (0.40)
Period 2: Unstable process (0.14)
Distribution with moderate variability and off-center.
Alcoholic concentration in the vats of yeast treatment (ºGl)
When below the LSL: Favors yeast
growing, depleting nutrients dosed for their
treatment.
When above the USL: Inhibits cellular
growth by inhibiting yeast recovery.
Analysis of Cpk Histogram observations
Period 1: Unstable process (0.95)
Period 2: Unstable process (0.43)
Distribution with moderate variability and
significantly off-center.
Viability in the fermentation vats (%)
When below the LSL: Causes reduction of
metabolic reactions efficiency, resulting in a
worse performance of fermentation
processes.
When above the USL: Unilateral variation,
USL is the best reachable result.
Analysis of Cpk Histogram observations
Period 1: Unstable process (-0.87)
Period 2: Unstable process (-0.52)
Distribution with high variability and significantly
off-center.
Infection in the fermentation vats (x107)
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
12
When below the LSL: Unilateral variation,
LSL is the best reachable result.
When the above USL: Reduces the cellular
viability as consequence of increased
bacterial infections.
Analysis of Cpk Histogram observations
Period 1: Unstable process (-0.15)
Period 2: Unstable process (-0.10)
Distribution with high variability and significantly off-
center.
As it could be observed from the analysis of the histograms, the variables were not centered at
the midpoint of the specifications.
6. Concluding remarks
The objective of the presented study was to illustrate the application of Statistical Process
Control tools in Ethanol production processes. For this purpose, data was collected by means
of direct observation, interviews and document analysis. During the study execution, control
charts were plotted and capability analysis was carried out. Observed deviations were
investigated and some causes of the problem were identified using methodologies for analysis
and troubleshooting, such as statistical and quality tools. Other actions included training the
employees involved in Fermentation and Yeast Treatment processes.
In general, the study demonstrated that principles of Statistical Quality Control could be
widely applied in Ethanol production processes. The quality tools allowed the diagnosis of
which variables should be controlled to improve the alcoholic fermentation efficiency.
Moreover, this diagnosis indicated which processes need improvement actions in terms of
variability reduction and in terms of systematic error corrections.
Based on presented results, some variables presented low or moderate variability, as:
Temperature on the fermentation vat, Brix of the mash, Alcoholic concentration in the
fermentation vats, Temperature of the mash and Alcoholic concentration in the vats of yeast
treatment, although, the average values are significantly off-center in relation to their
specifications. Such variables require the adoption of corrective measures to reverse the
systematic average deviations.
Other variables, besides being off-center, also presented high variability, as: Viability in the
fermentation vats and Infection in the fermentation vats. For such variables, it is necessary not
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
13
only to adopt measures to eliminate systematic average deviations, but also actions to reduce
variability.
For processes improvement, suggests an permanent and constant variables analysis, sharing
and providing the analysis in a structured and organized form. Measure and compare progress
towards the target, turning them into actions to correct problem causes and suiting the
processes to achieve better results. Using tools for analyzing and troubleshooting, attacking
the problems at their root causes, improving work standards, as well as equipment and
installations.
REFERENCES
COSTA, A. F. B; EPPRECHT, E. K. & CARPINETTI, L.C.R. Controle Estatístico de
Qualidade. 2º Ed. São Paulo: Atlas, 2005.
IPCC. Renewable Energy Sources and Climate Change Mitigation Cambridge University Press,
Cambridge, United Kingdom, 2011. 1075 p. Disponível em: <http://www.ipcc.ch/publications_and_data/
publications_and_data_reports.shtml#SRREN > Acesso em: 11 jan. 2013.
LIMA, U. A.; BASSO, L. C.; AMORIM, H. V. Produção de Etanol. In: LIMA, U. A. et al.
(Coord.). Biotecnologia Industrial: Processos Fermentativos e Enzimáticos. São Paulo,
Edgard Blücher, v. 3, 2001.
LOPES, M. M. Estudo comparativo da destilação em batelada operando com refluxo
constante e com composição do destilado constante. Dissertação (Mestrado). Escola
Politécnica da Universidade de São Paulo. São Paulo, 2008.
MARTINS, R. A. Conceitos básicos de controle estatístico da qualidade. EDUFSCar. São
Carlos, 2010.
MONTGOMERY, D. C. Introdução ao Controle Estatístico de Qualidade. 4ª ed. LTC. Rio
de Janeiro, 2004.
XXXIII ENCONTRO NACIONAL DE ENGENHARIA DE PRODUCAO A Gestão dos Processos de Produção e as Parcerias Globais para o Desenvolvimento Sustentável dos Sistemas Produtivos
Salvador, BA, Brasil, 08 a 11 de outubro de 2013.
14
OLIVEIRA, T. S. Aplicação do controle estatístico de processo na mensuração da variabilidade em uma
usina de etanol. ENEGEP. São Carlos, 2010.
ÚNICA. Ethanol: EPA reaffirms sugarcane biofuel is advanced Renewable fuel with 61% less emissions than
gasoline. News, 02 mar. 2010. Disponível em: < http://english.unica.com.br/noticias/> Acesso em: 13 jan. 2013.