METAL SURFACE DEFECT DETECTION USING DEEP LEARNING

The purpose of the proposed work is to apply a deep learning technique to quickly and accurately detect surface faults. The Shot MultiBox Detector (SSD) network was chosen and fused with the convolution neural network (CNN) MobileNet to generate the MobileNet-SSD because of the meta structure. A surface defect detecting technique was then planned, mostly employing the MobileNet-SSD. The network topology and settings were changed to form the detection model since the SSD's structure was optimized without sacrificing accuracy. The proposed technique was utilized to detect common defects on a container's protection surface, such as breaches, dents, burrs, and abrasions, within the filling line. The results reveal that our method is more accurate and faster than lightweight network methods and classic machine learning methods at detecting surface faults. The findings add to our understanding of fault detection in real-world industrial settings.

The presence of microbial contamination and biofilms at a beer can filling production line

Contamination of beer arises in 50% of all events at the late stages of production, the filling area. Hereby, biofilms, being consortia of microorganisms embedded in a matrix composed of extracellular polymeric substances, play a critical role. To date, most studies have focused on the presence of (biofilm forming) microorganisms within this filling environment. Our aim was to characterize the microbial status as well as the presence of possible biofilms at a can filling line for beer by determining the presence of microorganisms and their associated matrix components (carbohydrates, proteins and extracellular DNA (eDNA)). Targeted qPCR confirmed the presence of microorganisms at ten sites during operation and three after cleaning (from 23 sites respectively). The evaluation of carbohydrates, eDNA and proteins showed that 16 sites were positive for at least one component during operation and four after cleaning. We identified one potential biofilm hotspot, namely the struts below the filler, harboring high loads of bacteria and yeast, eDNA, carbohydrates and proteins. The protein pattern was different than that of beer. This work deepens our understanding of biofilms and microorganisms found at the filling line of beer beverages at sites critical for production.

Characterisation of biofilm hotspots at a can filling line for beer

<p>Biofilms are thought to play a serious role in the food processing environment. Direct contact with or detachment of microorganisms of biofilms could lead to product contamination resulting in diminished shelf life of the product. Packaging and filling are essential key steps for product safety, as any contamination within this step will directly impact the shelf life and safety of the product. For bottled and canned beverages filling lines are used for filling huge amounts of product in a standardised manner. Most parts of these lines are cleaned and disinfected automatically during cleaning in place (CIP) procedures. The design of these filling lines is of great importance, as accessibility and materials are crucial regarding the success of automatic cleaning and disinfection programs. Some companies implemented additional manual cleaning strategies for the complete removal of potential problem-causing sites. In the brewery setting the term biofilm is manifested since the 90s. However, until now all studies focus only on the presence of microorganisms, neglecting the presence of matrix components, which constitutes an essential component of a biofilm.  </p> <p>Within this study a filling line for cans, capable of filling 60000 cans per hour (volume 0.5 l), was investigated regarding critical sites for biofilm formation. The filling line is used primarily for beer, but mixed beer drinks are also filled. We sampled 23 sites using a scraper-flocked-swab method at two time points. The first sampling was done during operation and the second sampling was conducted one month later after the automated cleaning and disinfection procedure. The samples were characterised regarding their microbial load (qPCR for 16S rRNA and 18S rRNA genes) and the presence of biofilm matrix (phenol-sulfuric assay for carbohydrates, precipitation and SDS-PAGE with subsequent silver staining for proteins and precipitation and spectrophotometric quantitative measurements for extracellular DNA).</p> <p>During operation, we could identify three sites harbouring a biofilm by applying the definition of presence of microorganisms and at least two matrix components. Furthermore, there were seven sites harbouring microorganisms and one matrix component. After cleaning and disinfection no biofilm could be detected. At one site, microorganisms and one matrix component could be detected. The drastic reduction of biofilm positive sites indicates the successful removal of biofilms by the cleaning and disinfection process.</p> <p>The design of future filling plants should emphasise on the principles of hygienic design, as this can help to prevent biofilm formation and targeted removal of biofilms during cleaning and disinfection. The here identified biofilm hotspots indicate potential problem-causing sites and weak-points in the design of the filling line.</p>

TRIZ and Lean Philosophies Applied Together in Management Activities

<p><strong>Purpose: </strong>In the current market, the Portuguese industry faces strong competition from countries with substantially lower operating costs. The agri-food industry is also subject to increasing competition, both nationally and internationally. The utilization of methodologies that, in addition to continuous improvement, provide the development of creative and innovative solutions may be relevant for highlighting and differentiating between organizations. In this work, Teoriya Resheniya Izobretatelskikh Zadach and Lean Philosophies have been implemented together. Methodologies as Matrix of Ideality Matrix of Contradiction, Single Minute Exchange of Die, 5S and the Substance-Field analyses were used as complementary tools to improve the production management activities of agri-food sector. Strongly focused on the issues of time wastage and the organization and management of the filling line, the implementation of these methodologies led to a reduction of setup times, of the operators’ movement and an improvement in the line’s management and organization, and improve the results of 5S audits, depending on the work station. The aims of this study is the implementation of diferent methodologies, philosophies and tools to improve management activities.</p><p><strong>Methodology/Approach:</strong> The methodology presented in this work involves using Lean and TRIZ methodologies together. Namely Matrix of Ideality, the Matrix of Contradictions, SMED, 5S and the Substance-Field Analysis, were used to optimize the management activities, in agro-food industrial sector. The methodology strongly focused on the issues of time wastage, organization and management of an filling line.</p><p><strong>Findings:</strong> The implementation of methodologies led to a reduction of setup times by 60% to 70%, a reduction of the operators’ movement during format changes by 26.2%, and an improvement in the line’s management and organization by 9% to 12%, depending on the work station.</p><p><strong>Research Limitation/implication:</strong> In the management activities, the results are not universal. Therefore, the same methodology applied in another industrial sector will present differents results. Nevertheless, the application of this methodology and the tools choosen only make sense if it facilitates the management of improvements.</p><strong>Originality/Value of paper:</strong> There is a need and demand for innovative solutions that lead to continuous improvement, production processes optimization and resource saving. The combination of different validated methologies, with new concrete studies, is always in scientific and technical interest and continuously sought-after (demanded, welcome) by academic and industrial sectors.

Methodology of using the Adjoint solver optimization tool during flow in the intercooler filling line to minimize pressure drop

The paper deals with numerical modelling of the flow in the intercooler filling line by Adjoint solver to minimize pressure loss. The ANSYS Fluent software was used for the calculations. The basic flow calculation was performed in the first phase. Then the mathematical model with Adjoint solver optimization tool was defined. The numerical calculation was unstable and did not lead to a convergent solution, because of creation of vortexes. The mathematical model was simplified in the second phase. To suppress instabilities and vortices a dynamic viscosity of coolant was adjusted. The pressure gradients between inlet and outlet for unmodified geometry and for modified geometry were evaluated. The final evaluations of pressure drop changes were implemented for modified geometry with original dynamic viscosity of the coolant.

Model for improvement of overall equipment effectiveness of beerfilling lines

In this article, the efficiency loss of a beer filling line is investigated and modeled to improve its system performance in terms of overall equipment effectiveness and productivity. First, research is aimed at problems related to beer filling line resulting from poor production management where utilization and productivity of filling machines are low. This article derives the overall equipment effectiveness of filling machines and overall throughput effectiveness of the filling line, with the overall throughput effectiveness of general system (OTE(S)) obtained based on these measurements. Based on these metrics, the efficiency evaluation system is established. Next, according to the efficiency evaluation indexes, overall equipment effectiveness, and OTE(S), the overall efficiency of filling machines and filling line are calculated, respectively. From the results, the stenciling machine has the lowest efficiency. However, considering overall throughput effectiveness indexes and production capacity, unit 3 is the bottleneck and limits the entire productivity. Finally, efficiency loss is analyzed based on the overall equipment effectiveness and OTE(S) calculation results. Using a fishbone diagram for analysis, the cause of equipment loss is summarized and countermeasures are offered. Theory of constraint is used to analyze the efficiency loss of the filling line and give suggestions for improvement. This research can help beer enterprises achieve continuous improvement of equipment efficiency and production capacity.