Leveraging Six Sigma with industrial engineering tools in crateless retort production

M. M. BUNCE, L. WANG and B. BIDANDA

---

Abstract : This paper integrates Six Sigma objectives/concepts and industrial engineering tools within a quality framework, which are used to improve damaged can claims on a crateless retort production system . An overview of the canning industry and crateless retort processing is provided. Three experiments and a series of analysis tools are discussed. The root-cause analysis showed a correlation between can damage and can fill weight using design of experiments (DOE)-based Six Sigma techniques. Manufacturing system improvement projects and the implementation of a closed-loop control system are discussed as the measures used to continuously improve damaged can claims.

---

Introduction : Six Sigma has been embraced by manufacturing companies not only for its robust tool set but also because of its well-defined application methodology, called DMAIC (define, measure, analyse, improve, and control). Industrial engineering is a discipline that focuses on efficiency, optimiza-tion, and reduction of waste in operations. Leveraging Six Sigma concepts with industrial engineering tools can provide the shop floor engineer with a robust and rich tool set along with well defined objectives.

A process is said to be Six Sigma when six times the square root of the variation is less than the control range. Today, Six Sigma is regarded not only as a process capability equation, but also as a business philosophy for reducing process variation. The term Six Sigma has come to mean a problem-solving methodology that companies must embrace in order to eliminate variation and consequently improve quality in areas that are critical to their customers.

Canneries use a thermal processing manufacturing system, known as retort, in the production of preserved food. Retort processing is defined as heating foods prone to microbial spoilage in hermetically sealed containers to extend their shelf life . Simply put, this means that canned foods are cooked in a sealed can to kill harmful bacteria. Canneries have traditionally used retort processing, but recently manufacturers have developed an alternative, called ‘crateless retort’, that represents a paradigm shift in the process. In traditional systems, cans are conveyed into crates and loaded into a retort vessel. In crateless retort, cans are conveyed directly into a water-filled retort vessel; this water cushion prevents the cans from material handling damage. Crateless retort is desirable because it reduces overall throughput time. However, crateless retort systems need a more complex control system. Its numerous production variables must be closely controlled in order to maintain acceptable quality levels.

---

To know more about the process see the video on the site:
Malo inc

---

The corporate management team identified this improvement project due to a low level of customer satisfaction and unacceptable levels of damaged cans. Based on management experience with different quality improvement approaches, a Six Sigma team was chosen and a project charter was established (see table 1) with a clear goal to reduce damage claims by 50%. The Six Sigma team was inter-departmental, led by trained quality managers who were empowered to make both technical and financial decisions.

2.   Defining the Problem
The first step in the Six Sigma process is to define the effects of the problem and also to commit a project team to focus on customer requirements. A project charter or contract is a useful Six Sigma tool for committing a project team and translating the voice of the customer into project goals.Likewise, supplier input process output customer (SIPOC) charts provide a clear picture of the variables in a process, the boundaries of improvement efforts, and the scope of a project.

A SIPOC diagram typically helps establish the bounds or scope of the problem. It is most common to build a SIPOC chart by working backwards (i.e. by first identifying the customers of the project, then the outputs, etc.). The SIPOC chart for the crateless retort system is shown in table 2. The customer requirement in this application was to eliminate waste from the crateless retort system. The waste comes in the form of damaged can claims that cost the plant money and time to produce. Although it was relatively easy to measure the cost of damaged cans, it is not as easy to measure the clean up efforts that are required to dispose of the damaged cans. Both are considered forms of waste.

In the filling process, cans are depalletized and conveyed into a filling machine in one piece flow. The can is then filled with the edible product. The can filling includes any water or oil that is used to help preservation. The net weight is measured as the weight of the filling in each can. The headspace of the can is measured as the distance between the open can surface and the can filling. FDA specifications guide the net weight and headspace of each can along with the can seam and processing.

The filled metal containers are then sealed with a seaming machine that puts the lid on the can. The seaming machine creates a double seam in which the can lid and body are folded into each other. The quality of the seam is critical because it prevents harmful bacteria from entering the can. Specifications that govern the quality of the seam are frequently audited.

The seamed cans then enter the coding process and are labelled with a manufacturing text-based code, typically printed on the can by an inkjet. This inkjet is important in the auditing process because it contains product information and the exact date, hour and minute that the can has been filled with food. Once the can is coded, it is automatically loaded into the retort vessel by a conveyor. The vessel is filled with a water cushion that prevents the cans from damaging each other. At the proper fill weight and headspace, the cans (which are slightly denser than water) settle to the bottom of the vessel.

Since the retort process uses steam to cook the cans, the water needs to be displaced out of the retort vessel. The water is displaced when the vessel is full or when time and temperature requirements (TTR) between seaming and retort have been reached. Typically, once cans have been seamed, they are required to undergo retort within 1.5 hours. If the cans do not undergo retort within this TTR, then they will be sent to quality control for quarantine. Once the vessel is completely drained of water, the cans are cooked with steam until the retort TTR is reached. It is typical for cans to undergo one and a half hours of thermal processing at 98–106 ⁰ C. The steam used to cook the cans creates an atmospheric pressure inside the retort vessel, and inside the cans. The cooling process begins by displacing the steam with water. TTR also governs the cooling process, which typically lasts 10 minutes before the discharge. Cooling pressure is typically used to create equilibrium of pressure on the inside and outside of the can. The cooling pressure should match the pressure created by the steam in the retort cooking cycle. Without the cooling pressure, the pressure on the outside of the can will drop creating an imbalance because the inside of the can retains the pressure that was created during the retort cooking cycle.

After the retort cooling process, the cans are discharged onto the submerged conveyor system. The water-filled vessel undergoes a vacuum process so that the water from the vessel does not cause the submerged conveyor system to overfill. The water from the vessel is vacuumed into a valve, and the cans are dropped onto the submerged conveyor. The cans continue to cool while they are on the submerged conveyor. The conveyor system brings the cans out of the water where the cans are unscrambled for one piece flow. The cans are then stacked and palletized in preparation for the packaging process. At this point, damaged cans and cans in violation of TTR will be set aside for quality control to review. The quality control team maintains documentation of damaged cans including the product type, processing time, quantity of the claim, and reason for the claim. Cans that meet quality control standards are labelled and packaged for distribution.

3.   Measuring the Problem
The second step in the Six Sigma based DMAIC process is to measure system variability that can create waste. It is convenient to choose reporting metrics that can be easily understood. An engineering economic analysis was performed and to understand the return on investment (ROI) needed to rationalize the benefits of the project. In this case, a count of all damaged can claims was used. The process flow charting discussed previously revealed that a quality control process step measures the number of damaged cans by process date and damage type. This quality control measurement became the basis for several reporting mechanisms. Damaged cans were categorized into the following four types

• Buckled cans.
• Bad seams.
• No inkjet code.
• TTR not to specification.

The financial impact of damaged cans is presented as the product of the percentage of can damage claims, the annual production volume of cans and the unit cost of each filled can. Three different products are run on the crateless retort line and the damage was computed as follows: Results of the analysis are shown in table 3.

A Pareto analysis of the system shows that the damage claims on product 3 are significantly higher than the other since they account for 63% of the overall claims (see figure 4). The type of damage, also measured in a Pareto analysis, indicated that most of the damage on the product is due to buckling, accounting for 87% of the overall claims (see figure 5). Therefore, the focus of further research was on buckling damage on product 3.

Engineering Research Paper Ultrasonic NDT of wind turbine blades using guided waves

R.Raišutis, E.Jasiūnienė, E. Žukauskas

---

Abstract :
In order fully to exploit energy of wind power the construction elements of the wind turbine should be inspected periodically. Ultrasonic air-coupled technique using guided waves has been selected for inspection of wind turbine blades, because only one side access is enough and no contact is needed. Dispersion curves of phase velocities as well as leakage losses versus frequency were calculated using numerical global matrix model. Taking into account the results of the performed simulations the frequency of the ultrasonic transducers was selected to be 290 kHz due to non-dispersive region of phase velocities. The ultrasonic air-coupled technique using guided waves was used for investigation of the artificial internal defects in the wind turbine blade. These defects (diameter 19 mm and 49 mm) were made on the internal side of the main spar. From the ultrasonically obtained images it is possible to recognize the geometry of defects and to estimate approximate dimensions of the defects.

---

Introduction :
Wind power is a fast-growing and very promising source of environmentally safeand renewable energy with a high potential. However, in order to fully exploit energy of wind power the construction elements of wind turbines should be inspected periodically. In order to estimate level of a critical damage at the initial stage before collapsing it is necessary to perform continuous condition monitoring of wind turbine blades and the detailed inspection with elimination of the broken-down components [1].

To keep the wind turbine inoperation, implementation of condition monitoring system becomes very important. There are different techniques, methodologies and algorithms developed to monitor the performance of wind turbines. Inspection methods based on ultrasound, radiography, thermography, acoustics and optics enable to perform quality control and on site inspection [2].

One of the essential components in wind turbines are their blades. Wind turbine blades, while in operation, encounter very complex loading sequences, due to the stochastic nature of wind conditions at wind turbines sites. Blade failure is very costly because it can damage other blades, the wind turbine itself and other wind turbines located in neighbourhood. The efficient NDT procedures should extend wind turbine life and reduce failure possibility [1].

Ultrasonic methods were not applied yet very widely for inspection of wind turbine blades. Ultrasonic C-scan imaging has been used for area mapping of the composite delamination or interface disbond due to fatigue in normal field operation conditions of the turbine blade [3]. Three different ultrasonic measurement techniques were used for such investigation: pulse-echo, through transmission and pitch-catch. However, the influence of overlapped reflections, scattering and attenuation of the reflected ultrasonic waves from the multi-layered structure takes place. The scattering effect also has negative impact on the propagation of ultrasonic waves and requires application of lower frequencies. For example, in results presented by Gieske et althe contact type testing technique with 400 kHz transducers was used. Such set-up was similar to the guided waves generation in a particular layer of the structure and reception in a neighbour layer of the structure. The authors declare,that the delamination region between mentioned layers gave the shadowing effect. Therefore, such feature helped to detect the internal delamination [3].

Ultrasonic NDT using guided waves
Application of the guided waves is promising for the detection and sizing of internal defects between individual defects. In the case of guided wave interaction with a structural discontinuity, scattering of guided waves in all directions as well as mode conversion occurs. There are two approaches commonly used for structure health monitoring using guided waves: pulse–echo and pitch– catch [4]. From various characteristics of the received signal, such as the time of flight, amplitude etc., information about the damage in the inspected structure can be obtained. In order to estimate type of the defect, the signal processing algorithms have to be applied [4 - 9].

Simulations of phase velocity dispersion curves and leakage losses
For effective exploitation of the guided waves it is necessary to select frequency, therefore the global matrix numerical model has been used for calculation of the dispersion group and phase velocities curves [7-10]. Leakage losses versus the frequency were taken into account also. During simulation the scattering losses inside the GFRP (glass fibre-reinforced plastics) layers have been neglected due to short propagation distance of the guided waves inside the segment of the blade (approximately 40 mm) and also low operating frequencies of the air-coupled ultrasonic set-up. Anisotropy was neglected also. The drawings of the structures, for which phase and group velocity dispersion curves were calculated, are presented in Fig. 1. The structure, selected for simulations was similar to the real structure of the inspected wind turbine blade sample. In Fig. 1, a the defect free structure is presented, in Fig. 1, b – defected region (without the third layer, in order to simulate delamination type defect due to bad adhesion of glue/foam) is presented. Parameters of the layers used for simulations are listed in Table 1. The lateral dimensions of the structure have been assumed to be infinite.

The calculated phase velocity dispersive curves as well as leakage losses versus frequency for defected and defect free regions are presented in Fig. 2 -5. From the presented results it can be seen that for ultrasonic NDT of wind turbine blades the 290 kHz transducers may be used due to low leakage losses and less dispersive region. Experimental investigations
The measurements were performed using the air-coupled ultrasonic measurement system, which has been developed at Ultrasound Institute of Kaunas University of Technology. The photo of the experimental set-up is presented in Fig. 6.

The pair of air-coupled transducers has been used for non-contact scanning of the wind turbine blade sample. Positioning of the ultrasonic transducers has been performed by a precise mechanical scanning unit. Only one-side access to the sample surface was used. The structural diagram of the used air- coupled ultrasonic technique for NDT inspection of the wind turbine sample is presented in Fig. 7.

The frequency of the ultrasonic transducers f=290 kHz has been selected taking into account the simulation results obtained using the global matrix calculation technique. The transducers were mounted into pitch-catch configuration for generation and reception of guided ultrasonic waves. The transmitter was driven by the 8 periods and 750 V amplitude radio pulse. The total gain of the measurement system was 77 dB. Averaging of the 4 received signals was performed. The measurements were performed with the scanning step of 2 mm.

The cross-section of the inspected wind turbine blade sample is presented in Fig. 8. The photo of the inspected artificial circular defects with49 mm and 19 mm diameter is presented in Fig. 9. In Fig.10 the A-scans obtained over defected (1) and defect free (2) regions are presented. As can be seen from the waveforms, the signal amplitude over the defected region is considerably smaller. In Fig.11 the B-scan image of the 19 mm defect is presented. Lack of the leaky wave signal corresponds to the defected region. In Fig.12 the C-scan image of the 49 mm and 19 mm defects is presented. Both defects can be easily recognised and detected using a conventional amplitude detection technique. The ultrasonically obtained C-scan image shows a good contrast, which enables to estimate geometry of the defects and their approximate dimensions.

Additionally, the darker line at y=30 mm indicates, that besides the known defects in the sample there are unknown defects or variayions of material properties, which have to be inspected in the future.

Conclusions
For ultrasonic NDT of wind turbine blades ultrasonic technique using the air-coupled generation of guided waves has been selected due to only one side access and non-contact experimental set-up.

Simulations of group and phase velocity dispersion curves as well as leakage losses versus frequency for defected and defect free regions were performed using the numerical global matrix model. From the simulation results it can be seen that for ultrasonic NDT investigations of wind turbine blades fundamental A0 mode should be used due to low leakage losses and less dispersive region at the frequencies higher than 290 kHz.

The first measurements show that the proposed air-coupled ultrasonic technique, using Lamb waves allows finding defects in wind turbine blades. The ultrasonically obtained images (A-scan, B-scan, C-scan) show defects geometry and approximate dimensions.