Words: 4284
A high-power engine of about 1525 HP is composed mainly of cylinder heads. Power is developed at the bottom surface of the cylinder head and is subjected to high heat and vibration during combustion of the engine. The grey cast-iron casting of the head has inlet passages, intricate water- jacket passages for coolant flow, and exhaust passages to discharge the burnt fuels. The water Jacket is a complex passage for water circulation around the cylinder, the valve ports and seats, the combustion chamber, and any other hot part that requires cooling.
The heat of combustion is conducted through metal walls to the water in the jacket. The cylinder head is cast by assembling the base core, the water-Jacket core, the inlet core, and the exhaust core into one unit. (Antonym, Kumar, & Trial, 2005) Since the process presenting the problem was clearly complex and the root cause of the water-jacket-passage was unknown, the company decided to launch a Six Sigma project to address this issue. Define Based on the DYNAMIC methodology, the company looked for the definition of the scope and goals for the project presented below.
They were trying to develop a process that delivers improvement in terms of the customer requirements, so they implemented a project charter as the first part of this phase to help the team members clearly understand and see the limitations of the project, the objectives, time periods, resources, roles and potential financial gains. As the company was looking for continuous improvements they wanted to be sure about the commitment of the team, so while in this first step of the methodology creating a sense of ownership was one of the priorities.
The project was I * A champion * A Black Belt * A process owner * A green belt * Two graduate engineer trainees. They were focused in answering the questions about what was wrong with the recess and production and what could be the impact of such problem. By doing so they managed to define the goal of the project that was the “reduction of water- jacket-passage Jamming from 0. 194 defects per unit (DIPS) to 0. 029 DIPS, which should result in an immense reduction in the cost of poor quality (COOP). ” (Antonym, et al 2005).
By doing the tests required to get the number of DIPS they also concluded that the root cause letting sand to fuse and penetrate the metal was the porous core. The next diagram shows where the problem was occurring, helping them to conclude that f the 13 locations, location 2, 4, 6, 8, and 12 were more prone to Jamming. This phase also helped them to conclude that the engine overheating has a very severe impact causing warranty failures and customers dissatisfaction. To prevent this, they created a process map (shown below) in which they chose some key processes for deeper observation.
The following ones emerged for enhancing customer satisfaction and reducing COOP in the foundry: * Sand preparation * Core-making process * Wash preparation and coating. Measure Based on the process map made during the previous phase, and the three core recesses chosen (sand preparation, core preparation and wash preparation and coating) a cause and effect analysis was carried out. The main effect considered for this was the porous core (also identified during the previous phase as root cause). The diagram is shown below: This diagram led the team to identify some process variables as a result.
They were sand leakage, blow pressure, the OAFS number of sand, the gap in the core box, bulk density of the sand and vent choking. Moreover, the effect of these variables was considered to contribute 80% to the water-jacket-passage Jamming problem, proving hey were important enough to be considered. In order to better support the results obtained from the Cause and Effect analysis, a cause and effect matrix was created. It included the major customer needs and the importance each process characteristic had satisfying them.
The resulting matrix is shown below: As a result, the most important process characteristics were Depth of porous core and Abeam of wash, whereas the most important customer needs were Proper thickness and Filling of porous core. Having found these vital needs, the following step was to translate them into measurable characteristics. Therefore, specification emits were established for eight process parameters (shown in Table 2 below). This in the end let the following experimentation on each of these variables to determine their performance.
Apart from them, depth of porous core was considered a critical to quality characteristic. In order to ensure the proper execution of any experiment, a gauge R&R study had to be made. This let the identification of variation sources and the determination of the measurement system capability. The criterion used to determine the measurement system acceptation was to have a measurement variability less than 10% of total process variability. As it can be seen on the table 3 shown on the right, the measurement system variability turned out to be of 6. 08%, thus making the system acceptable.
On the other hand the baseline process capability, also known as the CAP of the process, had an initial value based on the existing process conditions and its was estimated at 0. 49. This indicated that the process performance was poor and needed improvement. This in the end, became one of the most relevant result of the project. Analyze In the analyze phase it was necessary to get the necessary information to detect the factors that affect the depth of the porous core. The next table explain all the information that the responsible of the work could get in 36 days of observation in different shifts through all day.
The table shows the main elements that can change the properties of any given piece. Problems like Sand leakage, Blow pressure, Bulk density, Abeam of wash and Fin thickness are some data that is collected in the table. After getting all this information, it was necessary to detect the locations where errors with sand are more usual. The Praetor chart shown above classifies the defects according to their location within the 13 spots initially identified, and as a result the team saw the main robbers showed mostly on stations 8, 2, 12, 6, 4 and 13. To finish this phase it was necessary to find the main factors that led variation.
To do so, they compared themselves with a Brazilian automobile company. After a simple regression analysis, the factors that were chosen are Sand Leakage, Bulk Density and Vent Choking Ratio. The decision to choose them was according to the P value, where the factor was less than 0. 05. The next table shows the results after the regression analysis. Improve In this phase the company decided to perform a designed experiment using three armaments (sand leakage, bulk density, and vent choking ratio). The parameters were identified from the regression analysis in the previous phase. In more detail, each parameter gave the following results.
Sand leakage: It was found that sand was coming out of the core box when the door half and the ram half were closely packed due to a misalignment of the two halves. Both halves of the core box were aligned properly and packing was used to seal any remaining opening. This parameter can be varied from 10 to 30 g/blow. Bulk density: The desirable bulk density of core sand varies from 1. 78 to 1. 95 g/CM. Bulk density was increased by adding iron oxide and mill scale. The iron oxide composition was increased to 3 % in order to achieve the desired bulk density and to improve the high-temperature thermal properties of the core sand.
Mill scale was increased 1. 5 % and helped to increase the density and thermal conductivity of the sand. Vent choking: The resin, which is used for the core sand, blocked the opening of vent by sticking to it during the curing process. This blocking resulted in the production of a porous core because pressurized air from the core shooter could not escape out of the core box. It was found that out of 52 vents on both halves of the core box only 30 were not blocked. To clear the blocked vents thinner was used. The cleaning frequency was increased from once a week to twice a day during shift break.
To control the experiment other parameters of the process were controlled. 1 . OAFS number: Sand grains of only OAFS number 65-70 reached the hopper of the shell core machine. 2. Blow pressure: A pressure gage was installed to keep blow pressure in the range of 4 to 4. 5 keg/CM. 3. Abeam of wash: The wash was regularly checked to maintain its value according to the specification tankard. 4. Fin thickness: Maintenance of the shell core machine brought the fin thickness in the range of 0-0. 85 mm. The experiment was conducted using the three parameters discussed above.
Each process parameter was studied at two levels in order to keep the size of the experiment to a minimum, as well to meet time and cost constraints. To investigate both main effects and interaction effects among the parameters a AAA full factorial design was chosen. Each trial condition was replicated twice in order to have the right degrees of freedom for studying both main effects and interaction effects. Results of the experiment with average depth of the porous core as the variable of interest (or yield) are shown in the next table. The average depth of the porous core before the experiment was 1. 5 mm. The objective of the experiment was to decrease the depth of the porous core so the first objective of the analysis was to determine the effect of the process parameters and to understand the presence of any interactions, if present. Fig. 5 indicates that the optimum levels of the process parameters for minimizing the depth of the porous core are: bulk density in a high level 1. 95 g/scam; a low vent choking ratio, low sand gage level at egg/blow. The interaction plot in Fig. 6 suggested that there was a slight interaction between vent choking ratio and sand leakage.
In order to determine the statistical significance of both main and interaction effects, a normal probability plot of effects was constructed. Fig. 7 show only the main effects that were statistically significant at 10 % significance level. Normal silica or chromites sand was mixed with shell sand of OAFS number 70-75 to achieve better results. Zircon mixed with magnetite was used as a wash substitute to increase the refractoriness of the wash. Implementation of the aforementioned suggestions resulted in further improvements in process capability and process yield.
Confirmation trials were made using the optimal settings and the average depth of the porous core was calculated to be 0. 80 mm. The process variability was reduced. The process capability Cap improved from 0. 49 to 1. 28. The increase in the Cap demonstrates a significant improvement in the process performance. Control In order to keep in control the process and sustain the improvements gained in terms of process yield and average depth of the porous core, the project managers implemented process sheets and control charts.
Thus, operators could easily and constantly monitor the process and detect whenever the process seemed to go out of control, taking the pertinent preventive actions. Besides, they managed to elaborate a complete database with the information collected from the charts. Below are shown the control charts for the depth of the porous core, before (fig. 8) and after (fig. 9) the improvements. It can clearly be seen that after the improvements not only the variability around the mean was reduced but the value of the control limits and the mean itself decreased.
Considering that the optimal results are given at the minimum depth, it can be appreciated that the improvements worked. However, if results may begin to deviate and go out of control at any moment, thanks to the control charts the operator can notice and correct the problem. So it won’t become a cause of customer dissatisfaction. As a result of the whole DYNAMIC implementation, several results were achieved. First of all, the cylinder-head manufacturing process evolved into a eight- steps procedure: 1. Making of the core in a core shooter machine. 2. Finishing of the core. 3.
Coating of the finished core with a wash. 4. Drying the wash coating by passing through a heating section 5. Assembling the core in an automated casting section for metal casting of the cylinder head. 6. Removing the unwanted part from the casting by passing through a fettling section. 7. Cleaning and polishing the cylinder-head casting. 8. Testing the quality of the casting in the inspection/testing unit. With the aim to produce a well-informed savings report, the costs generated through every activity were divided into four categories: labor, raw material, operating expenses and other overhead costs.
The detailed description of the costs involved (with the exception of raw material costs) is presented below: Based on these data the savings derived from this project were calculated. One of the areas where the true power of Six Sigma was demonstrated was on the raw material costs. The savings are shown below. Savings generated from the remaining cost categories are shown on the following table: It is clear that defects were considerably diminished in every steps of the process resulting in $ASS 11,471. 60 in savings.
Financially speaking, this is very evident when the values for Annual Impact before and after Six Sigma are compared. In the end, this resulted in a $ASS 111,524. 93 profit improvement for the company. In terms of the process performance, the key metrics also proved improvements. Defects went down, yield went up, the Cap index went from 0. 49 to 1. 28. Standard deviation decreased and process mean did it as well. Though most of these benefits were mentioned as soon as they were achieved along the DYNAMIC execution, table 10 summarizes all the perspectives that were positively impacted thanks to the Six Sigma implementation.
Critique When considering the performance of this Six Sigma project, we immediately realize hat using DYNAMIC to solve this problem was one of the best options, not only because it was related to the quality improvement and customer satisfaction, but also because it is comprehensive enough to adapt itself to the particular conditions of the problem, such as the defects present in the production of the engine’s elements, this while holding an structured approach to never lose the focus regardless of the complexity every process or every step has.
The real advantages of applying this methodology were more clearly appreciated in the results gotten at the end. The economic impact and of course the quality improvement were so notorious after applying the changes in the Improve phase. Despite of the good decisions taken, we consider there are certain actions within some of the steps that did not really met the requirements demanded for the problem in question.
For example, the process map created during the Define phase was drawn without considering any standard or flow logic, and even though it shows the deep knowledge the team had about the process they were dealing with, it may be assumed with a fairly low risk that this model was made with the expected outcomes in mind. In other words, instead of letting the diagram show new aspects of the process, they simply poured their knowledge into it, which may have produced then a confirmation bias.
To counter this potential misuse of the process map, a SIP could have been created instead. Not only would it show the activities that are already present in the existing process map along with the inputs and outputs each one is tied too (which by the way are not present), it would also let the team go further in the identification of customer dissatisfaction, by addressing not only the characteristics they were most uncomfortable with, but also heir identity (if a subgroup could be identified) in terms of gee-location, socio economic level, etc.
Another tool particularly useful for this project that would have been advisable for the team to use (especially considering they are part of the automobile industry) is the value stream map. This would have summarized from the beginning all the data the team would need in the future, as well as point out the data missing that would be needed to generate during the measure phase, in an easy to read format. Considering the Measure step, it is very interesting to note that spite of the fact that it is a Six Sigma project, the calculation of the current sigma level was never done.
If we take into account the need to produce visible and meaningful results quickly, this relatively simple calculation would have been of enormous utility to the company and the team. It may even had led the team to consider the necessary amount of data to be collected before the question arrive later on. Moreover, they had identified since the define phase the locations where the defects were produced, calculating the sigma level based on the defect per million of opportunities would have only made more sense.
Additionally, during the brainstorming session carried out to make the cause and effect analysis, it would have been g and even necessary to include personnel closely related to the process and not only the Six Sigma team; this may have contributed as well with very enriching comments about possible causes. When moving into the Analyze step, the data collected and used to start the experimentation process seems to be too small to be relevant.
Assuming the enterprise makes several cars a day, as most automotive companies now a days, the data collected from 36 days may be not representative enough of the total population of cars affected by overheating, let alone the total amount of cars using engines manufactured by the company. In addition to that, comparing information gotten from a different company could make more problems in the analysis, even when the company has similar process to the other.
In this point is where comparing information with a Brazilian company could affect the behavior of the part. The environment, the quality of the raw material, humidity and even the weather can be strong factors for the process of part, and of course, they are factors hat both paces differ from each other. Finally, a formal calculation of the sample size would have shed some light on the right sample to use, and this would have reinforced the conclusions gotten in the experiments.
In the Improve phase, we consider that the factors used were adequate and the selection that pointed them to be the few vital is appropriate. However, while the AAA design shows some slight interaction between vent choking ratio and sand leakage, it may be recommendable to use a more comprehensive design that includes more than two levels for each actor, so they could have strong evidence of this interaction in a wider interaction plot.
This may have required to run a fractional experiment instead of a complete one, but since there was no other interaction with an equally promising effect, the results may had been worth the effort. Nevertheless, it would have been highly advisable to perform the regression analysis in order to find the optimum bulk density. Finally, even though we can assume there was a comprehensive process control plan given the fact that the process ended up being standardized into 8 tepees, the sole mention of the control charts employed does not guarantee that it was indeed created.
Moreover, a complementary failure and effect mode analysis could have warned employees when using the control charts shown, by, for example, sensitivity them to recognize special deviation patterns applicable to this particular process, based on the detestability scores assigned to each potential failure. Individual Conclusions Irvin Oswald Ramirez Ramirez: During all this time studying these courses of Lean Six Sigma, I couldn’t understand very well the applications of the method and how to ace them in a real situation.
However, this work has helped me a lot to understand them better. Also I could ‘t notice that there are many more applications for the tools than I ever expected. I enjoyed working in this homework because I could see that we can do many things in this kind of works. More than that, I think that we could realize the way to start our project and how to face the steps. The most difficult of this part was choosing and finding a work with all the data needed, because most of them do not explain properly the origins of the company where the project was done.
However, we could get an organized project with the necessary information for this work, making in this way easier for the team members to understand it. Alexandra Villainies: I think that this research project helped us to view the whole DYNAMIC methodology as the integrated set of activities it really is. Though it may still be seen as a theoretical assignment, thanks to the fact that it documents the actual execution of the Six Sigma philosophy in general and the DYNAMIC methodology in particular, we were able to see how important is to keep the continuity between every step and even between the tools used.
This made me aware of how important is to have an strategic vision to bring this type of projects from definition to closure. This may be very useful in the months to come to better conduct our own certification project. Benefits aside, I consider the most significant challenge this project may present to us compared to what we aim to do in the future, is the manufacturing nature of the project. Though the team portrayed in the original source used tools we know, the final integration and usage in our case may differ because of the transactional approach we want to have.
Yet again, and having these different approaches in mind, the project reminded me as well that it is not necessary to use each and every of the tools we know to succeed, instead we should use only those that are truly relevant for the specific problem we are dealing with, and that will be the true value we would add as Green Belts. Elided Garcia: This activity allowed me to get a nice sneak peak to a full and well implemented six sigma project and its results. Six sigma projects are effective if a critical failure is addressed like the one covered here “Jammed water-jacked passage”.
This failure is the cause of overheating in the motor. As a mechanical engineer I found this research activity very interesting and I think it is a complete tour of the DYNAMIC methodology. Now with a more complete understanding of each step reach and goals, I began to take a grasp of the whole picture. Jorge Ivan Lopez Ulna: For me, this research project has been the most interesting and helpful activity of the course so far. And this is because I could finally see applied in a real, complete, well done and effective Six Sigma project all the knowledge I acquired throughout the semester.
Of course, it does not mean that I find the other activities irrelevant, however with the analysis of this real Six Sigma project I was able to reinforce my understanding of the DYNAMIC methodology and compare it to the DYNAMIC steps we applied to the McDonald’s problem. I realized that we did a great Job and that both as individuals and as a team, that we have a very good understanding of the Six Sigma Methodology and that when the time comes, we will be able to perform perfectly on our Certification Project.
Ana Teresa Espanola Rubles : I think this activity give me a better perspective of how ar or close we were from the practical word and I can see that most of the tools we used so far, were used on the case, so this give me a real motivation to doing things better and secure about them. For this I consider this activity truly helpful and it really prepares me for seeing how methodically the documentation is also for SE that the information has to be resumed and several important things for us in the future as green belts.
Team Conclusions This project meant quite a challenge for us both as individuals and as a team, since it represented the culmination of a long process of introduction to the Six Sigma Methodology and its wide set of tools, expanding our vision and comprehension of every step of the DYNAMIC. Furthermore we were able to compare our integral work on a McDonald’s restaurant to this successful Six Sigma project, clearly understanding the importance every step, and even more important the relation and dependence among them.
However we believe that the most outstanding part of the project was the critique, as it allowed us to put into practice everything we learnt throughout the semester, analyzing, Judging and discussing alternate solutions and tools to every step of the Methodology. Besides we could realize the importance of the synergy between Six Sigma and Lean, insomuch as to our consideration the problems related to the operators would been more effectively solved if Lean tools were implemented alongside the Six Sigma improvements.