applied sciences

Article Absorption Characteristics and Optimal Design of Using Drop Testing

Chien-Chih Wang 1,* , Chin-Hua Chen 2 and Bernard C. Jiang 3

1 Department of Industrial Engineering and Management, Ming Chi University of Technology, New Taipei City 24303, Taiwan 2 Yuen Foong Yu Packaging Inc., Taoyuan Plant, Taoyuan City 33849, Taiwan; [email protected] 3 Department of Industrial Management, National Taiwan University of Science and Technology, Taipei City 24301, Taiwan; [email protected] * Correspondence: [email protected]; Tel.: +886-2-2908-9889

Abstract: The application of corrugated to buffer packaging has increased with the rise of the circular economy. The dynamic buffer curve is the key to designing the buffer packaging structure but requires multiple testing by small- and medium-sized enterprises (SMEs) without resources. In this study, we propose drop testing to perform a fractional factorial experiment and establish a regression model of strength through experimental data. The analysis results show that static stress, falling height, and buffer material thickness are the key variables of impact strength, and an impact strength prediction model (R2 = 94.1%) was obtained. Model verification using the buffer package design of a personal computer showed that the measured values of impact strength fell   within the estimated 95% confidence interval. These results indicate that SMEs can use the proposed analysis procedure to improve the design of corrugated paper using minimal resources. Citation: Wang, C.-C.; Chen, C.-H.; Jiang, B.C. Shock Absorption Characteristics and Optimal Design Keywords: dynamic buffer curve; static stress; drop testing; cushion design; regression model of Corrugated Fiberboard Using Drop Testing. Appl. Sci. 2021, 11, 5815. https://doi.org/10.3390/app11135815 1. Introduction Academic Editors: João Carlos de Buffer packaging is essential for ensuring high-quality logistics services. Although Oliveira Matias, Carina Pimentel, the packaging is part of external quality costs, achieving an effective packaging design is Anabela Carvalho Alves and essential in the industry [1,2]. Foamed are commonly used as buffer packaging Andrea Spagnoli materials. However, with the rise of environmental awareness, some countries have regulated the use of foamed materials [3]. Recently, corrugated board has been Received: 24 May 2021 favored for buffer packaging [4–7]. Biancolini [6] proposed a numerical model to evaluate Accepted: 21 June 2021 the stiffness parameters of corrugated boards; the model was used for the parametric Published: 23 June 2021 investigation of the influence of local parameters on corrugated board panel stiffness. The results showed that the stiffness parameters were weakly controlled, acting on the fluting Publisher’s Note: MDPI stays neutral shape and composition. Further, design methods for corrugated board were with regard to jurisdictional claims in reviewed, and two numerical models were developed using finite elements to evaluate the published maps and institutional affil- iations. buckling and ultimate load predictions [7]. In practice, the six-step method of cushion packaging design and MIL-HDBK-304C form the most important basis for packaging design [8]. Product damage during delivery may occur from shock (accidental drop during handling) and (bumps during transportation). Therefore, shock absorption and vibration transfer characteristics must be Copyright: © 2021 by the authors. considered in packaging design. The dynamic buffer curve can reveal the shock ab-sorption Licensee MDPI, Basel, Switzerland. characteristics of the buffer material. This curve shows the impact strength (maximum This article is an open access article acceleration) of a specific buffer material with a specific thickness and height under different distributed under the terms and conditions of the Creative Commons static stresses. The dynamic buffering curve can be used to determine the thickness and Attribution (CC BY) license (https:// area of the buffering material in the early design stage. However, because the corrugated creativecommons.org/licenses/by/ board is a composite made of multilayer and corrugated core paper, each layer 4.0/). may be changed owing to cost, leading to various permutations and combinations. In

Appl. Sci. 2021, 11, 5815. https://doi.org/10.3390/app11135815 https://www.mdpi.com/journal/applsci Appl. Sci. 2021, 11, x FOR PEER REVIEW 2 of 15

paper, each layer may be changed owing to cost, leading to various permutations and combinations. In addition, the existing buffer material evaluation method to obtain Appl. Sci. 2021, 11, 5815 dynamic buffer curve does not provide the final packaging solution 2and, of 15 thus, causes excessive packaging. The dynamic buffer curve is usually U-shaped (Figure 1). Zone A refers to the weight ofaddition, the heavy the existinghammer buffer being material relatively evaluation light. method When to an obtain impact dynamic is applied buffer curve to the cushioning does not provide the final packaging solution and, thus, causes excessive packaging. material, the impact force fails to deform the cushioning material but generates a high The dynamic buffer curve is usually U-shaped (Figure1). Zone A refers to the weight acceleration.of the heavy hammer Zone B being refers relatively to the light. heavier When weight an impact of is the applied heavy to the hammer. cushioning The cushioning materialmaterial, is the deformed impact force upon fails the to deform application the cushioning of the impact, material prolonging but generates the a high impact time and greatlyacceleration. reducing Zone Bthe refers acceleration. to the heavier Zone weight B ofis thethe heavy most hammer. suitable The area cushioning for cushion design. Zonematerial C refers is deformed to the upon fact the that application the heavy of the hammer impact, prolonging is too heavy the impact to bear time the and impact, and a greatly reducing the acceleration. Zone B is the most suitable area for cushion design. cushionZone C refersbottom to theout fact occurs that the when heavy the hammer hammer is too fall heavys. Currently, to bear the the impact, cushion and material a has no bufferingcushion bottom effect. out A occurs dynamic when cushion the hammer curve falls. can Currently, be used the to cushion determine material the has thickness no and area ofbuffering the cushioning effect. A dynamic material cushion to avoid curve cost can beincr usedeases to determine or product the thicknessdamage and caused area by excessive orof insufficient the cushioning packaging. material to avoid However, cost increases this curve or product is obtained damage caused by the by excessiveASTM D1596 [9] test or insufficient packaging. However, this curve is obtained by the ASTM D1596 [9] test underunder a specificspecific combination combination of buffer of buffer material material type, thickness, type, thickness, and drop height and anddrop can height and can onlyonly be usedused to to describe describe the shockthe shock absorption absorption characteristics characteristics of the buffer of material the buffer under material under thatthat specific specific combination. combination. When When conditions conditions change, change, time-consuming time-consuming experiments experiments must be must be repeatedrepeated toto obtain obtain the the curve. curve.

Figure 1. Dynamic buffer curve under a specific combination of buffer material type, thickness, and Figuredrop height 1. Dynamic by the ASTM buffer D1596 curve test. under a specific combination of buffer material type, thickness, and drop height by the ASTM D1596 test. Thus far, Taiwanese buffer material manufacturers have not provided dynamic buffer curveThus data, far, although Taiwanese they have buffer been widelymaterial used ma innufacturers other countries. have The not cost ofprovided test dynamic equipment is high, and only large institutions such as the Industrial Research Institute have buffersuch equipment. curve data, Generally, although the industrythey have or design been unitswidely do notused have in suchother equipment, countries. and The cost of test equipmentobtaining the is required high, informationand only bylarge self-testing institutions is challenging. such as Since the a seriesIndustrial of tests Research have Institute haveto be performed,such equipment. it is costly andGenerally, time-consuming the industry for small- andor medium-sizeddesign units enterprises do not have such equipment,(SMEs) to obtain and the obtaining required dynamicthe required buffer information curve data. Therefore, by self-testing cushion packagingis challenging. Since a designs often rely on the experience of the designer. seriesIn of practice, tests thehave methods to be for performed, establishing dynamicit is costly buffer and curves time-consuming are ASTM D1596 for for small- and medium-sizedthick plate materials enterprises and ASTM D4168(SMEs) [10 ]to for obtain -in-place the required materials. Theredynamic are relevant buffer curve data. Therefore,method applications cushion topackaging evaluate the designs dynamic often cushioning rely on performances the experience of packaging of the ma- designer. terialsIn [practice,11–13]. However, the methods no dynamic for establishing buffering curve dynamic test method buffer exists curves for laminated are ASTM D1596 for thickcorrugated plate bufferingmaterials materials. and ASTM Research D4168 on the[10] impact for foam-in-place absorption characteristics materials. of There cor- are relevant rugated cardboard has mostly used ASTM D1596, and only a specific combination of methodconditions applications could be studied, to suchevaluate as the corrugatedthe dyna cardboardmic cushioning of a specific performances material, mate- of packaging materialsrial thickness, [11–13]. and drop However, height. Since no dynamic different manufacturers buffering curve adopt test different method standards exists in for laminated corrugatedcorrugated boardbuffering designs, materials. the existing Research dynamic buffer on the curve impact cannot beabsorption directly applied characteristics of corrugatedto a buffer packaging cardboard design. has Therefore, mostly used methods ASTM that canD1596, truly reflectand only the real a specific packaging combination of situation need to be developed so that packaging designers can predict the characteristics conditions could be studied, such as the corrugated cardboard of a specific material, material thickness, and drop height. Since different manufacturers adopt different standards in corrugated board designs, the existing dynamic buffer curve cannot be directly applied to a buffer packaging design. Therefore, methods that can truly reflect the real packaging situation need to be developed so that packaging designers can predict the Appl. Sci. 2021, 11, x FOR PEER REVIEW 3 of 15

Appl. Sci. 2021, 11, 5815 3 of 15

characteristics of cushioning materials. Gorman improved ASTM D4168 and proposed an in-package test method for a foam buffer [14]. The impact testing results obtainedof cushioning using materials.ASTM D1596 Gorman were closer improved to the ASTM actual D4168 packaging and proposedmethod. an in-package test methodIn addition, for a Wang polyethylene and Sun foam [15] buffer investigat [14].ed The the impact bending testing fatigue results behavior obtained of usingflute typesASTM BD1596 and C were corrugated closerto the actual packagingsamples under method. cyclic loading. The results can be appliedIn addition,to the dynamic Wang design and Sun and [15 accelerated] investigated vibration the bending test of stacked fatigue corrugated behavior of . flute Millstypes et B al. and [16] C used corrugated an ABAQUS paperboard implicit samples analysis under to obtain cyclic force–deformation loading. The results curves can for be foamapplied liners to the and dynamic corrugated design and accelerated and conclu vibrationded that testa finite of stacked element corrugated analysis could boxes. predictMills et the al. cushioning [16] used an performance ABAQUS implicit of comple analysisx-shaped to obtainliners better force–deformation than the liner curvescurve designfor foam method. liners andNumerical corrugated methods cartons are and often concluded used to calculate that a finite the elementstrength analysisof boxes. could The finitepredict element the cushioning method has performance been successfully of complex-shaped applied in linersmany betterstudies than on thethe linernumerical curve analysisdesign method. of boxes Numerical made of corrugated methods are cardboard, often used such to calculateas the transverse the strength shear of stiffness boxes. The of corrugatedfinite element cardboard method [17–21]. has been Rami successfully et al. [17] applied applied finite in many element studies methods on the for numerical the finite analysis of boxes made of corrugated cardboard, such as the transverse shear stiffness element analysis of nonlinear materials and structures. Nordstrand [19] proposed a panel of corrugated cardboard [17–21]. Rami et al. [17] applied finite element methods for the compression test rig, which was built to achieve accurately defined load and boundary finite element analysis of nonlinear materials and structures. Nordstrand [19] proposed conditions. The analysis results showed a difference in the order of 15–20% between a panel compression test rig, which was built to achieve accurately defined load and experimentally estimated and analytically predicted critical buckling loads for orthotropic boundary conditions. The analysis results showed a difference in the order of 15–20% plates. Garbowski et al. [20] proposed mixed analytical/numerical methods for estimating between experimentally estimated and analytically predicted critical buckling loads for the strength of boxes with openings of various dimensions and locations. Moreover, orthotropic plates. Garbowski et al. [20] proposed mixed analytical/numerical methods together with practical guidelines, the homogenization method can be used to obtain for estimating the strength of boxes with openings of various dimensions and locations. stiffness properties of corrugated cardboard shell structures and any periodic shell Moreover, together with practical guidelines, the homogenization method can be used to structures [21]. obtain stiffness properties of corrugated cardboard shell structures and any periodic shell structures [21]. 2. Methodology 2.1.2. Methodology Experimental Instrumentation 2.1. ExperimentalIn this study, Instrumentation a set of low-cost test procedures was developed using a falling test machineIn this to study,replace a the set ofdynamic low-cost buffering test procedures curve. We was used developed a drop using testing a falling machine, test accelerometer,machine to replace and thesignal-processing dynamic buffering software curve. in We the used laboratory a drop testing of a machine,cooperative ac- manufacturercelerometer, and to collect signal-processing test data (Figure software 2). inThe the Lansmont laboratory PDT-56ED of a cooperative drop manufacturertest machine wasto collect used testto dataload (Figurethe test2). samples, The Lansmont with PDT-56EDa drop test drop height test machinerange of was 30.0–182.9 used to loadcm, maximumthe test samples, sample with weight a drop of 56 test kg, height and rangedrop ofslope 30.0–182.9 of <2° by cm, the maximum CNS 2999 sample [22], weightASTM ◦ D775of 56 kg,[23], and and drop ISO slope2248 of[24] <2 dropby thetest CNS standards. 2999 [22 The], ASTM ICP Model D775 [ 23356A11], and ISOpiezoelectric 2248 [24] three-axisdrop test standards.accelerometer The was ICP used Model to 356A11 record piezoelectricthe acceleration three-axis changes accelerometer during the wasfall process.used to recordThe measurement the acceleration range changes was ±500 during G, thethe fall sensitivity process. Thewas measurement10 mV/G, and range the frequencywas ±500 range G, the was sensitivity 2–7000 was Hz (X 10 and mV/G, Y-axes) and and the frequency2–10,000 Hz range (Z-axis). was 2–7000 The acceleration Hz (X and sensorY-axes) data and and 2–10,000 impact Hz strength (Z-axis). were The accelerationanalyzed using sensor Lansmont data and TP3 impact signal-processing strength were software.analyzed The using default Lansmont values TP3 for signal-processingthe analysis were software.330 Hz for Thethe defaultfilter frequency values for and the a half-sineanalysis werewave 330for Hzthe forwaveform. the filter frequency and a half-sine wave for the waveform.

(a) (b) (c)

FigureFigure 2. 2. ExperimentalExperimental instrumentation. instrumentation. (a (a) )PDT-56ED PDT-56ED drop drop test test machine, machine, ( (bb)) ICP ICP Model Model 356A11 356A11 piezoelectricpiezoelectric three-axis three-axis accelerometer, accelerometer, and (c) Lansmont TP3 signal-processing software.software.

2.2. Experimental Procedures This study integrated design experiments, a variance analysis, and a regression analy- sis to propose a three-stage procedure to develop a predictive model of the impact strength. The results replaced the traditional dynamic buffering curve and provided an optimal Appl. Sci. 2021, 11, x FOR PEER REVIEW 4 of 15

2.2. Experimental Procedures This study integrated design experiments, a variance analysis, and a regression analysis to propose a three-stage procedure to develop a predictive model of the impact strength. The results replaced the traditional dynamic buffering curve and provided an optimal buffering design of the corrugated board. The first stage involved sample Appl. Sci. 2021, 11, 5815 preparation and setting the experimental environment. In the second stage,4 of 15a screening experiment was conducted to identify the critical factors for the drop test. In the third stage, the in-package test was conducted to establish a prediction model for the impact strength.buffering The design implementation of the corrugated steps board. are Thedescribed first stage as follows: involved sample preparation andStage setting 1: the experimental environment. In the second stage, a screening experiment was conducted to identify the critical factors for the drop test. In the third stage, the 1.in-package Cut the test corrugated was conducted board to establishinto 20 acm prediction × 20 cm, model and forthen, the use impact an strength. The to splice the implementationsample according steps are to described the required as follows: thickness. 2. AccordingStage 1: to the pretreatment regulations of the CNS 13190 [25] packaging goods 1. test,Cut the corrugatedsamples were board left into for 20 at cm least× 20 24 cm, h and before then, the use test an adhesive at 20° and to splice relative the humidity ofsample 50% and according 80%. to the required thickness. 2. According to the pretreatment regulations of the CNS 13190 [25] packaging goods Stage 2: test, the samples were left for at least 24 h before the test at 20◦ and relative humidity 3. Designingof 50% and 80%.the drop experiment: The factors proposed in the literature were integratedStage 2: with practical considerations to select the experimental factors, including 3. theDesigning fixed and the dropdesign experiment: factors. TheThe factors drop proposedtest was inperformed the literature by were considering integrated the actual datawith collection practical considerations situation and to select designing the experimental a blocking, factors, a replicated including the2k factorial fixed and design, to collectdesign the factors. data. The drop test was performed by considering the actual data collection k 4. Thesituation drop and test designing and recording a blocking, the a replicatedimpact strength: 2 factorial We design, placed to collect the thespecimen data. on the 4. The drop test and recording the impact strength: We placed the specimen on the cantilever of the drop test machine and stepped on the drop test pedal to drop the cantilever of the drop test machine and stepped on the drop test pedal to drop the specimenspecimen (Figure(Figure3). 3). The The impact impact time time and accelerationand acceleration curve were curve plotted were using plotted using signal-processingsignal-processing software. software. The The maximum maximum acceleration acceleration gave the gave impact the strength. impact strength.

(a) (b) (c) (d)

FigureFigure 3. 3. DecompositionDecomposition procedure procedure of of the the drop drop test: test: (a) start,(a) start, (b) leaf (b) down, leaf down, (c) leaf ( swing,c) leaf andswing, and (d) specimen(d) specimen drop. drop.

StageStage 3: 5. Designing the in-package experiment: We considered the actual packaging situation 5. Designingand designed the the in-package experiment experiment: with the key factorsWe considered obtained from the Stepactual 2. packaging situation 6. andThe designed in-package the test: experiment The laminated with corrugated the key boardfactors specimen, obtained test from (simulatedStep 2. 6. Theproduct), in-package and filled test: corrugated The laminated board were corrug sequentiallyated board packed specimen, into the corrugatedtest box (simulated product),box to realize and afilled complete corrugated package board specimen. were The sequentially in-package specimenpacked into was the then corrugated boxsubjected to realize to the a drop complete test machine package to complete specimen. the impact The in-package strength measurements specimen was then (Figure4). subjected to the drop test machine to complete the impact strength measurements (Figure 4). Appl.Appl. Sci. Sci. 20212021, 11, 11, x, 5815FOR PEER REVIEW 5 of 15 5 of 15

Figure 4.4. SchematicSchematic diagram diagram of of the the in-package in-package test. test. 2.3. Analysis 2.3. Analysis ANOVA and regression techniques were used to analyze the experimental data [26]. ANOVAANOVA was the and core regression experimental techniques design were method used and to evaluatedanalyze the whether experimental the design data [26]. ANOVAfactors were was significant the core factorsexperimental through design a residual method analysis and and evaluated factor effect whether verification. the design factorsThe residual were analysissignificant was factors performed through using a residualresidual plots analysis to validate and factor the ANOVA. effect verification. The Theresidual residual plots analysis included was the normalperformed probability using re plotssidual of residuals,plots to validate residuals the versus ANOVA. fits, The residualand residuals plots versus included predictors, the normal which probability were used plots to verify of residuals, the assumption residuals of normality, versus fits, and residualsconstant variance, versus predictors, and uncorrelation which with were each used other. to Ifverify the normal the assumption and equal varianceof normality, constanttests were variance, not passed, and a uncorrelation residual analysis with was each performed other. If again the normal through and the Box–Coxequal variance transformation. If the test results still did not pass, the analysis was stopped, and the tests were not passed, a residual analysis was performed again through the Box–Cox experiment was replanned, just as the independent test did not pass. After the residual transformation. If the test results still did not pass, the analysis was stopped, and the analysis, the key factors were obtained from the normal probability of the factor effects or experimentthe ANOVA was table. replanned, A variation just trend as ofthe the indep factorendent effect wastest alsodid not presented pass. After through the the residual analysis,factor effect the graph key factors and interaction were obtained graph. from the normal probability of the factor effects or the ANOVAAfter identifying table. A the variation key factors, trend using of athe regression factor effect analysis was to establishalso presented the prediction through the factormodel, effect we selected graph a and polynomial interaction regression graph. model and established the procedure to delete the insignificantAfter identifying factors throughthe key variable factors, screening. using Thea regression best combination analysis of parametersto establish the predictionwas selected model, based on we a high selected R2, high a adjustedpolynomial R2, low regression S, and Cp values;model then, and the established residual the procedureanalysis was to performed. delete the If passed, insignificant the estimation factor modes through was appropriate. variable Onscreening. the contrary, The best combinationthe Box–Cox transformationof parameters ofwas the selected data was based performed on a high to recreate R2, high the adjusted model until R2, low the S, and residual analysis was passed. In addition, to meet the practical application requirements, Cp values; then, the residual analysis was performed. If passed, the estimation mode was the regression model was required to reach the standard of an R2 greater than 95%. appropriate. On the contrary, the Box–Cox transformation of the data was performed to recreate3. Experimental the model Analysis until the residual analysis was passed. In addition, to meet the practical applicationThe experiments requirements, were conducted the regression using corrugatedmodel was board required and equipment to reach the provided standard by of an Rour2 greater partner than companies. 95% First, the factors affecting the impact absorption characteristics of the laminated corrugated board were summarized from the literature, including the flute 3.height, Experimental thickness ofAnalysis Kraft paper, percentage of recycled pulp, hardness of the corrugated board,The thickness experiments of the cushioning were conducted material, using humidity, corrugated static stress, board and and drop equipment height. After provided discussing with field experts and the actual exploration of the experimental environment, by our partner companies. First, the factors affecting the impact absorption characteristics we set the drop height and static stress (product weight/cushion area) at 76.2 cm and of the laminated corrugated board were summarized from the literature, including the 14 g/cm2, respectively. The experimental factors were the caliper (liner), flute height, flutecushion height, thickness, thickness and basis of Kraft weight paper, (medium). percentage However, of therecycled experiment pulp, could hardness not be of the corrugated board, thickness of the cushioning material, humidity, static stress, and drop height. After discussing with field experts and the actual exploration of the experimental environment, we set the drop height and static stress (product weight/cushion area) at 76.2 cm and 14 g/cm2, respectively. The experimental factors were the caliper (liner), flute height, cushion thickness, and basis weight (medium). However, the experiment could Appl. Sci. 2021, 11, x FOR PEER REVIEW 6 of 15 Appl. Sci. 2021, 11, 5815 6 of 15

not be completed in one day, so the day and relative humidity were used as the blocking factors.completed Table in 1 one gives day, the so information the day and of relative the experimental humidity were andused blocking as the factors. blocking factors. Table1 gives the information of the experimental and blocking factors. Table 1. Level values of the experimental and blocking factors. Table 1. Level values of the experimental and blocking factors. Factors Levels FactorsDay Day1 LevelsDay1 Blocking factor Relative HumidityDay RH 50% Day1 RH Day180% Blocking factor CaliperRelative (Liner) Humidity 0.186 mm RH 50% 0.387 RH mm 80% FluteCaliper height (Liner) 2.7 mm 0.186 mm 4.7 0.387 mm mm Experimentation Factor Flute height 2.7 mm 4.7 mm Experimentation Factor Cushion thickness 4-ply 10-ply Cushion thickness 4-ply 10-ply 2 2 Basis weightBasis weight (Medium) (Medium) 100 g/m100 g/m 2 180180 g/m g/m 2

3.1. Empty Box Drop Test 3.1. Empty Box Drop Test In this stage, the 24 blocking of full-factorial resolution V designs with two block In this stage, the 24 blocking of full-factorial resolution V designs with two block factors were adopted [27]. Ten repetitions were set to reduce the experimental error, and factors were adopted [27]. Ten repetitions were set to reduce the experimental error, and a total of 160 experiments were conducted. During the experiment, two block factors were a total of 160 experiments were conducted. During the experiment, two block factors considered, so the entire experiment was divided into four blocks, and the sequence of were considered, so the entire experiment was divided into four blocks, and the sequence experimentsof experiments in inthe the blocks blocks was was completely completely random. random. The The response response was was the the maximum accelerationacceleration (G(G value,value, impact impact strength) strength) measured measured by the by accelerometer the accelerometer fixed onfixed the producton the productat the time at the of thetime fall of impact. the fall impact. AfterAfter completing completing the the experiments experiments according according to to the the designed designed experimental experimental sequence, sequence, anan ANOVA ANOVA was applied to determine the signi significantficant factors. The significant significant factors from thethe normal normal probability probability plot plot of of the the factor factor ef effectsfects were were B, B, C, BC, BD, CD, ABC, ACD, and BCD.BCD. The The ANOVA ANOVA showed that the effect of th thee blocking factor was not significant significant and thatthat the subsequentsubsequent analysis analysis would would pool pool the the error error term term with with the insignificantthe insignificant factors. factors. Then, Then,the residual the residual analysis analysis of the of pattern the pattern check check was performed. was performed. FigureFigure 55aa showsshows thethe relationshiprelationship betweenbetween thethe predictedpredicted valuesvalues andand residualsresiduals (top-(top- right).right). The The graph graph gradually gradually expands expands from from left left to toright, right, which which is not is notconsistent consistent with with the assumptionthe assumption of equal of equal variance, variance, so sodata data tran transformationsformation is isnecessary. necessary. The The data data were transformedtransformed by by weighted weighted least-squares least-squares regressi regressionon and and reanalyzed. reanalyzed. The The factors factors B, B, C, C, BC, BC, BD,BD, CD, CD, ABC, ABC, ACD, ACD, and and BCD BCD were were significant significant at at a a type type I I error of 0.05. Finally, Finally, a residual analysisanalysis was was performed performed on on the the transformed model. model. Figure Figure 55bb showsshows thatthat thethe datadata werewere consistentconsistent with with normality, normality, equalequal variance, and independence.

(a) (b)

Figure 5. ResidualResidual plots plots for for model model checking: checking: ( (aa)) model model results results of of the the original original data data and and ( (bb)) model model results results of of the the transformed transformed data using the Box-Cox technique. Appl. Sci. 2021, 11, x FOR PEER REVIEW 7 of 15 Appl. Sci. 2021, 11, 5815 7 of 15

The results of both the pre-transformation and post-transformation analyses showed The results of both the pre-transformation and post-transformation analyses showed that the main factors affecting the impact absorption characteristics of the laminated that the main factors affecting the impact absorption characteristics of the laminated corrugated board were the flute height and thickness (number of layers): the higher the corrugated board were the flute height and thickness (number of layers): the higher the flute height, the better the cushioning effect. A-flute is the highest among all flute types flute height, the better the cushioning effect. A-flute is the highest among all flute types and, and,therefore, therefore, has the has best the cushioning best cushioning performance. performance. Due to Due its good to its buffering good buffering property, property, A-flute A-fluteis mostly is usedmostly as aused buffering as a buffering material inmaterial the industry. in the Therefore,industry. Therefore, only A-flute only cardboard A-flute cardboardwas used as was the used target as of the the target subsequent of the subs experiment.equent experiment. The higher theThe thickness higher the (number thickness of (numberlayers), the of greater layers), is the the amountgreater is of the deformation amount of and deformation impact. The and effect impact. of relative The humidityeffect of relativewas not significanthumidity was and differednot significant from that and reported differed in thefrom literature, that reported probably in becausethe literature, of the probablycharacteristics because of the of paperboard.the characteristics Since theof the moisture paperboard. absorption Since of the the moisture corrugated absorption medium ofand the Kraft corrugated paper used medium by different and Kraft manufacturers paper used varies, by different the degree manufacturers of influence varies, of relative the degreehumidity of alsoinfluence varies. of In relative subsequent humidity experiments, also varies. the samples In subsequent were processed experiments, under the samplestemperature were and processed humidity under conditions the temperature specified and by thehumidity relevant conditions test standards specified to reduce by the relevantthe experimental test standards variations. to reduce In the the screening experime experiment,ntal variations. the In thickness the screening of the experiment, Kraft paper theand thickness the base of weight the Kraft of the paper core and paper the base were weight not significant; of the core thus, paper considering were not significant; the costs, thus,the subsequent considering experiment the costs, the was subsequent conducted experiment using Kraft was paper conducted and a core using paper Kraft of lighter paper andbase a weight. core paper of lighter base weight.

3.2. In-Package In-Package Experiment We collected data on the impact strength (G value) of the corrugatedcorrugated cardboard cushioning material from the first first drop. The purpose of the in-package test was to ensure the applicability of the results to thethe actualactual cushionedcushioned packagingpackaging design. The equipment used in this phase was the same as that used used in the the screening screening test, except that the simulated product was replaced by a weight-adjustable test chamber. The test chamber was made according to ASTM D4168, with a length and width of 25 cm each and a height of 21 cm and was made made of of a a stainless-steel stainless-steel plate plate with with a a weight weight of of 3.125 3.125 kg. kg. The The static static stress stress was was 5 2 g/cm5 g/cm2 whenwhen the the sample sample size size ofof the the laminated laminated corrugated corrugated board board was was 25 25 cm cm in in length and width, and the weight of each piece was 1.563 kg; the static stress could be increased by 2 2.5 g/cm 2 forfor each each piece piece to to adjust adjust the the weight weight of of the the box box to to meet the specificspecific static stress requirements of the experiment. The configurationsconfigurations of the experimental and assembledassembled samples are shown in Figure6 6a,b,a,b, respectively.respectively.

(a) (b)

Figure 6. In-packageIn-package test: test: ( (aa)) experimental experimental sample sample and and ( (b)) assembly assembly sample on the test equipment.

This phase initially used seven candidate experimental factors for the investigation: (1). StaticStatic stress: stress: Static Static stress stress is is an an importan importantt parameter parameter in cushion cushion packaging design and isis determined determined by by the the weight weight and and size size of of the the product. product. At At present, present, most of the products usingusing corrugated corrugated cardboard cardboard as as a a cushioning material material are are 3C products, so thin and Appl. Sci. 2021, 11, 5815 8 of 15

light is the mainstream of design, with a weight below 10 kg and static stress below 35 g/cm2. (2). Falling height: The standards of the falling height vary for each packaging test. Before designing cushion packaging, it is necessary to understand the standards and their falling heights quoted by customers during acceptance tests. The highest and second-highest drop heights of the drop test-related standards are listed in Table2.

Table 2. Highest and second-highest drop heights of the drop test-related standards.

Standard Drop Height (cm) Package Weight (kg) 91.4 0~13.6 kg MIL-HDBK-304C [28] 76.2 13.6~34.0 kg 61.0 0~9.1 kg ASTM D4169 [29] 53.3 9.1~18.1 kg 106.7 0~9.1 kg ASTM D3332 [30] 91.4 9.1~22.7 kg 80.0 10 kg or less CNS 10033 [31] 60.0 10 kg~20 kg

(3). Humidity: Although the influence of humidity in the screening test is not significant, the temperature and humidity of buffer test specimens are still regulated in the relevant standards to reduce experimental variations. In ASTM D1596 and D4168, the required temperature is 23 ◦C ± 2 ◦C, and the relative humidity is 50% ± 2%; in CNS 14615 [32], it is stipulated that the specimens should be placed at a temperature of 20 ◦C ± 2 ◦C and relative humidity of 65% ± 5% for more than 16 h, and the hygroscopic specimens should be placed for more than 24 h. Corrugated cardboard is a hygroscopic material, so the pretreatment time should be more than 24 h. In this stage, the samples were treated according to CNS 14615. (4). Kraft paper thickness: The effect of Kraft paper thickness in the screening test was not significant; therefore, considering the costs, Kraft paper with a basis weight of 130 g/m2 and a thickness of 0.186 mm was used in the subsequent experiment. (5). Corrugated height: The A-flute cardboard is more commonly used as a buffering ma- terial, because it has the best buffering characteristics. Hence, only A-flute cardboard was tested in the subsequent tests. (6). Thickness (number of layers): In practice, the thickness (number of layers) of corru- gated cardboard is more than four layers and less than 10 layers. (7). Base weight of the corrugated medium: The effect of the Kraft paper thickness was not significant in the screening test; therefore, the base weight of the corrugated medium of 100 g/m2 was used in the follow-up test. The abovementioned factors were reviewed by experts, and three main factors (drop height, static stress, and cushion thickness) were used, with each factor taking four levels for the 34-factor experimental configuration. The experimental order was completely randomized, and the number of repetitions was set to five to reduce the experimental error; a total of 320 experiments were required. The information of the experimental factors is given in Table3. The experimental fixation factors are described below. The sample pretreatment temperature and humidity were set at 20 ◦C ± 2 ◦C, 65% ± 5% relative humidity, and 24 h, according to CNS 14615. The thickness of the Kraft paper was 0.186 mm, with a basis weight of 130 g/m2. A corrugated flute board with a height of 4.7 mm and a corrugated medium with a basis weight of 100 g/m2 were used.

Experimental Data Analysis The analysis procedure started with a least-squares method to estimate the created cubic multiple models. Then, the initial model was obtained by variable filtering. The residual analysis was performed, and if it passed, the model capability was evaluated, and if it failed, the model and residual analysis were recreated after conversion until the residual Appl. Sci. 2021, 11, x FOR PEER REVIEW 9 of 15 Appl. Sci. 2021, 11, x FOR PEER REVIEW 9 of 15

Table 3. Experimental factors of the in-package test. Table 3. Experimental factors of the in-package test. Appl. Sci. 2021 11 Level , , 5815 Factor Level 9 of 15 Factor 1 2 3 4 1 2 3 4 Drop height (cm) 30 60 90 120 Drop height (cm) 30 60 90 120 Static stress (g/cm2) 5 15 25 35 Staticanalysis stress was (g/cm passed.2) JMP Pro® 16 was 5 used to perform 15 the analysis.25 Figure7a35shows Cushion thickness (ply) 3 6 9 12 Cushionthe results thickness of the effect (ply) summary analysis, 3 where the 6 variables with 9 a p-value less 12 than 0.05 were filtered out, the model was recreated, and the residual analysis was performed. Experimental Data Analysis ExperimentalFigure7b shows Data the Analysis residual by the predicted plot in the residual analysis; since the plot wasThe extrapolated, analysis procedure the assumption started ofwith consistent a least-squares variance method was not to passed.estimate Therefore, the created a The analysis procedure started with a least-squares method to estimate the created cubictransformation multiple models. was required Then, to the recreate initial the model model. was obtained by variable filtering. The cubic multiple models. Then, the initial model was obtained by variable filtering. The residual analysis was performed, and if it passed, the model capability was evaluated, and residual analysis was performed, and if it passed, the model capability was evaluated, and ifTable it failed, 3. Experimental the model factors and ofresidual the in-package analysis test. were recreated after conversion until the if it failed, the model and residual analysis were recreated after conversion until the residual analysis was passed. JMP Pro® 16 was used to perform the analysis. Figure 7a residual analysis was passed. JMP Pro® 16 was used to perform the analysis. Figure 7a shows the results of the effect summary analysis, where theLevel variables with a p-value less shows the resultsFactor of the effect summary analysis, where the variables with a p-value less than 0.05 were filtered out, the model1 was recreated, 2 and the residual 3 analysis 4 was than 0.05 were filtered out, the model was recreated, and the residual analysis was performed.Drop height Figure (cm) 7b shows the residual 30 by the predicted 60 plot in 90 the residual analysis; 120 performed. Figure 7b shows the residual by the predicted plot in the residual analysis; sinceStatic the stress plot (g/cm was2 )extrapolated, the assumpti5on of 15consistent variance 25 was not 35passed. since the plot was extrapolated, the assumption of consistent variance was not passed. Therefore,Cushion thickness a transformation (ply) was required 3 to recreate 6the model. 9 12 Therefore, a transformation was required to recreate the model.

(a) (b) (a) (b) Figure 7. Results of the experimental data analysis: (a) effect summary analysis and (b) residual by the predicted plot. FigureFigure 7. 7. ResultsResults of of the the experimental experimental data data analysis: ( (aa)) effect effect summary summary analysis analysis and and ( (b)) residual residual by the predicted plot plot.. The most-used log transformation was used, and then, the model was recreated. TheThe most-used most-used log log transformation transformation was was us used,ed, and and then, then, the the model model was was recreated. recreated. FigureFigure 88aa–c–c showsshows the the results results of theof normality,the normality, equal equal variance, variance, and independence and independence evalua- Figure 8a–c shows the results of the normality, equal variance, and independence evaluations.tions. The absence The absence of abnormal of abnormal patterns patterns indicates indicates that the establishedthat the established pattern is pattern consistent is evaluations. The absence of abnormal patterns indicates that the established pattern is consistentwith the assumptions. with the assumptions. consistent with the assumptions.

(b) (b)

(a) (c) (a) (c) Figure 8. Residual plot for examining the goodness-of-fit in the regression: (a) normal probability plot of residuals, (b) residuals versus order of data, and (c) residuals versus fits. Appl. Sci. 2021, 11, 5815 10 of 15 Appl. Sci. 2021, 11, x FOR PEER REVIEW 10 of 15

Figure 8. Residual plot for examiningThe optimal the model goodness-of-fit for the established in the regression: the impact (a) normal strength probabilityyˆG is as plot follows: of residuals, (b) residuals versus order of data, and (c) residuals versus fits. p yˆG = e where p = 6.03 − 0.0161x + 0.00569xa − 0.000029x 2 2 + 0.0117x 2 The optimal model for the estabc blished the impact strengthb c 𝑦 is as bfollows: + 0.000008x 2 − 0.000215 x − 0.279x − 0.0001516x 3 ab c 𝑦 𝑒 abc b b where p 6.03 0.0161𝑥+ 0.000443 x 0.00569𝑥ab + 0.000000016 0.000029𝑥xa2b3 +0.00186 0.0117𝑥xbc2 0.000008𝑥− 0.00000027xab2c 0.000215𝑥2 − 0.000432xc3 + 0.279𝑥0.00095 x 0.0001516𝑥ac + 0.000443𝑥0.00000038 xa2 0.000000016𝑥b2 − 0.0000002xa3b+ 0.00186𝑥0.00000362xb3c 0.00000027𝑥 0.000432𝑥 0.00095𝑥 yˆG : Predicted value of impact 0.00000038𝑥 strength 0.0000002𝑥 0.00000362𝑥 a : Drop𝑦 : heightPredicted value of impact strength b : Static𝑎∶Drop stress height 𝑏 : Static stress c : Cushion thickness 𝑐 : Cushion thickness FigureFigure9 shows 9 shows the resultsthe results of the of modelthe model evaluation. evaluation. Figure Figure9a shows9a shows the the results results of of the the actualactual and and predicted predicted models, models, which which are are positively positively correlated. correlated. The Thep-value p-value in inFigure Figure9b 9b is 2 is <0.0001,<0.0001, indicating indicating that that the the model model is is appropriate. appropriate. RR2 = 0.941029 0.941029 in in Figure Figure 9c9c is very very close 2 closeto to the the adjusted adjusted R2 = 0.9393130.939313 andand isis greater greater than than 0.9, 0.9, indicating indicating that that the the model model is is representative.representative.

(b)

(a) (c)

FigureFigure 9. Results 9. Results of the of model the model evaluation: evaluation: (a) actual (a) actual by predicted by predicted plot, (plot,b) lack (b) of lack fit test,of fit and test, (c and) summary (c) summary of the of fit the model. fit model. 3.3. Summary of the Analysis Results 3.3.In this Summary study, of a predictivethe Analysis model Results of the impact strength was developed based on the experimentalIn this design study, to a replace predictive the traditionalmodel of the dynamic impact cushioning strength was curve developed and provide based the on the optimalexperimental cushioning design design to for replace corrugated the traditiona boards.l dynamic cushioning curve and provide the optimalIn the empty cushioning box drop design test, for the corrugated four experimental boards. factors used were caliper (liner), flute height,In the cushion empty thickness, box drop and test, basis the four weight experi (medium)mental factors to collect used the were data caliper using the (liner), 24 blockingflute height, of the cushion full-factorial thickness, design. and From basis theweight results (medium) of the screening to collect experiment,the data using the the 24 mainblocking factors affectingof the full-factorial the impact absorptiondesign. From characteristics the results of of the laminatedscreening experiment, corrugated the boardmain were factors the flute affecting height the and impact cushion absorption thickness. characteristics The higher the of flute the laminated height, the corrugated better was theboard cushioning were the effect. flute height A-flute and was cushion the highest thickness. among The all the higher flute the types, flute so height, only A-flute the better cardboardwas the was cushioning used as the effect. target A-flute of the was subsequent the highest experiment. among all The the higher flute types, the cushion so only A- thicknessflute (numbercardboard of was layers), used the as greater the target was theof amountthe subsequent of deformation experiment. and impact. The higher The the cushion thickness (number of layers), the greater was the amount of deformation and Appl. Sci. 2021, 11, 5815 11 of 15

effects of the Kraft paper thickness, the base weight of the core paper, and relative humidity were insignificant. In the in-package test, three experimental factors—drop height, static stress, and cushion thickness—and five fixed factors—temperature, humidity, the thickness of the Kraft paper, corrugated flute board with a height, and corrugated medium with a basis weight—were used to design the experiment for data collection. By analyzing the results, the statistical model of the impact strength established by the drop height, static stress, and cushion thickness could be used to design the best-corrugated board.

4. Verification In the validation experiment, the cushioned package design was implemented using the actual product to verify the applicability of the developed model to the actual cushioned package design. During the design, the model was used to estimate the impact strength of the drop test, and the drop test was performed on the actual product. The impact strength obtained from the actual test was compared with the estimated value of the regression. The test products were designed using a desktop PC as the test product, and the cushioning package was designed according to three acceptance standards: MIL-HDBK- 304C, CNS 10033, and the customer requirements. The regression estimation formula was used to calculate the static stress or thickness (number of layers) required for the drop test acceptance criteria. The test product specifications and the three acceptance standards are listed in Table4. The standard or the customer determined the drop height; the fragility of the impact strength in the MIL-HDBK-304C computer equipment was 40–59 Gs, and the design standard was set at 50 Gs or less; for the CNS 10033 standard, it was set at 60 Gs or less.

Table 4. Test product specifications and three acceptance criteria.

Item MIL-HDBK-304C CNS 10033 Customer Requirements Weight (g) 12,360 Length(cm) 19.0 Product Width(cm) 43.0 Height(cm) 37.6 Drop height(cm) 91.4 60 76.2 Requirements Acceleration (G’s) Less than 50 Less than 60 Less than 65

Buffer Packaging Design The cushioning package design mainly determines the length, width, and thickness of the cushioning material. Once the length and width of the cushioning material are determined, the static stress is determined. Since the impact strength prediction of the model is a function of the main factors, when the model is used in the cushioning package design, the cushioning effect can be calculated from the design parameters, and the optimal design parameters can be calculated by combining different conditions. To verify the usability of the established regressions, two buffer designs were evaluated for each of the three acceptance criteria. Design 1 was used to calculate the required thickness (number of layers) at a fixed static stress, while design 2 was used to calculate the appropriate static stress with a fixed thickness (number of layers according to design 1) and then calculate the length and width of the buffer from the static stress obtained (Table5). After completing the cushion packaging design, the A-flute corrugated cardboard produced by the partner company was used to produce laminated corrugated cardboard and corrugated boxes for outer packaging according to the design results. Appl.Appl. Sci. Sci.2021 2021, 11, 11, 5815, x FOR PEER REVIEW 1212 of of 15 15

TableTable 5. 5.Comparison Comparison of of the the buffer buffer packaging packaging design design results. results.

MIL-HDBK-304CMIL-HDBK-304C CNS CNS 10033 Customer Customer Requirements ItemItem DesignDesign A1 A1 Design Design A2 A2 Design B1B1 Design Design B2 Design C2 C2 Design Design C1 C1 Drop height (cm) 91.4 60 76.2 Drop height (cm) 91.4 60 76.2 Requirement Acceleration (G’s) (G’s) 50 50 60 65 2 StaticStatic stress(g/cm stress(g/cm2) ) 15.1215.12 19 19 15.12 18.5 18.5 15.12 15.12 18 18 ThicknessThickness (ply) (ply) 99 99 44 4 4 5 5 5 5 Design resultresult LengthLength (cm) (cm) 19.0 19.0 19.0 19.0 19.019.0 19.0 19.0 19.0 19.0 19.0 19.0 Width (cm) 43.0 34.2 43.0 35.2 43.0 36.1 Width (cm) 43.0 34.2 43.0 35.2 43.0 36.1

TheThe test test equipment equipment used used at at this this stage stage was was the the same same as as that that of of the the screening screening test, test, except except thatthat the the sample sample was was changed changed to theto the actual actual product. product. Before Before packaging, packaging, the accelerometer the accelerometer was fixedwas onfixed the on hard the disk hard of disk the PC. of Afterthe PC. packing, After packing, the specimens the specimens were placed were in aplaced humidity in a ◦ ◦ chamberhumidity for chamber pretreatment. for pretreatment. The temperature The temperature and humidity and humidity were set atwere20 setC ± at2 20C °Cand ± 2 65%°C and± 5%, 65% respectively, ± 5%, respectively, according toaccording CNS 14615. to TheCNS samples 14615. testedThe samples after processing tested after are shownprocessing in Figure are shown10. The in drop Figure test 10. was The performed drop test according was performed to CNS according 2999, and ato drop CNS tester 2999, measuredand a drop the tester impact measured strength the of impact the product streng atth the of bottomthe product of the at drop. the bottom of the drop.

FigureFigure 10. 10.Schematic Schematic of of the the drop drop verification verification test. test.

TheThe products products were were packaged packaged with with manufactured manufactured cushioning cushioning materials, materials, corrugated corrugated cartons,cartons, andand drop drop tests tests to to compare compare the the measured measured results results with with the the predicted predicted values. values. The The measuredmeasured impactimpact strengthsstrengths ofof thethe sixsix cushionedcushioned packagingpackaging designsdesigns areare givengiven inin TableTable6 ,6, withwith all all the the values values within within the the 95% 95% confidence confidence interval interval (CI) (CI) of of the the predicted predicted values. values. The validation experiment results showed that the proposed model can be used in Table 6. Comparisonthe actual of the measuredpackaging impact buffer strengths design. of theTheref six cushionedore, in practice, packaging the designs. proposed model can effectively shorten the design time and cost by eliminating the trial-and-error method. In Customer addition, althoughMIL-HDBK-304C the thicknesses (number of CNS layers) 10033 of A2, B2, and C2 were the same as Requirements Item those of A1, B1, and C1, they not only saved the area of the cushioning material under higher static stressDesign but A1 also Designhad a better A2 Designcushioning B1 effect Design than B2 A1, Design B1, and C2 C1. Design C1 Drop height (cm) 91.4 60 76.2 Requirements Acceleration (Gs) 50 60 65 Prediction 46.86 43.34 55.80 52.69 60.15 57.60 Upper bound of 95% CI 49.04 45.19 58.50 55.27 62.91 60.20 Acceleration (G’s) Lower bound of 95% CI 44.78 41.57 53.21 50.24 57.51 55.11 Test 47.65 44.21 54.54 52.48 62.29 56.59

The validation experiment results showed that the proposed model can be used in the actual packaging buffer design. Therefore, in practice, the proposed model can effectively Appl. Sci. 2021, 11, 5815 13 of 15

shorten the design time and cost by eliminating the trial-and-error method. In addition, although the thicknesses (number of layers) of A2, B2, and C2 were the same as those of A1, B1, and C1, they not only saved the area of the cushioning material under higher static stress but also had a better cushioning effect than A1, B1, and C1.

5. Discussion The dynamic cushion curve provides useful information for not only selecting the appropriate cushioning material, thickness, and area but also predicting the cushioning effect and avoiding the use of the trial-and-error method, shortening the time and cost of packaging design. However, owing to the limitations of the test equipment and high costs, it is not widely used in Taiwan, and the industry mostly uses the trial-and-error method for packaging design. The dynamic cushion curve can determine only the application range of the static stress and cannot be accurately estimated. Further, the design using the trial-and-error method can determine whether the design meets customers’ requirements but cannot provide an optimal design. Table7 shows a comparison of this study with other studies [ 10,14], and the different highlights are described and discussed below.

Table 7. Comparison of the differences between this study and other studies.

Item This Study Other Studies Test method In-packages + Drop test ASTM D 1596 Test equipment Drop tester + Test block Cushion tester Specimen Multi-levels Specific material, thickness Testing conditions Multi-levels Specific drop height (impact velocity) Test result Generate regression model Generate dynamic cushion curve Application to the cushion design as same as the Application Predict the performances of varies cushion design testing conditions only

In terms of test methods, this study used the in-package method to perform the drop test, while the other studies used ASTM D1596. In terms of test equipment, the drop test machine and the test chamber were used in this study, while the cushion test machine was used in other studies. However, cushioning testers are limited to cushioning material testing and are very rare in Taiwan, so drop testers offer an inexpensive and easily accessible alternative. In terms of specimens and test conditions, this study conducted multilevel tests on corrugated A-flute board in terms of the drop height, static stress, thickness, and other factors, so the results apply to a wide range of applications. Our study was based on the idea of data decision-making and established the model by experimental design, which can be widely applied to design practice. By contrast, other studies were conducted for specific conditions, and the results can be applied to designs with the same conditions, limiting their applicability.

6. Conclusions In this study, a set of low-cost test procedures using the in-package method based on the experimental design technique was developed using a falling test machine to determine the optimal regression model of critical factors for impact strength to replace the dynamic cushioning curve. Packaging designers can calculate the buffering effect under different design conditions in advance using the established statistical model to deliver the appropriate packaging design in-line with customers’ requirements. The advantage of the proposed method is that retesting for a specific combination of design conditions to establish a dynamic buffer curve is not required. Therefore, SMEs can use the proposed approach to improve the design of corrugated paper and effectively reduce the amount of cushioning materials used and the cost of packaging. Appl. Sci. 2021, 11, 5815 14 of 15

Author Contributions: Conceptualization, C.-C.W. and C.-H.C.; methodology, C.-C.W. and C.-H.C.; validation, C.-C.W., B.C.J. and C.-H.C.; formal analysis, C.-C.W. and C.-H.C.; resources, C.-H.C.; data curation, C.-H.C.; writing—original draft preparation, C.-C.W.; and writing—review and editing, C.-C.W. and B.C.J. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Data Availability Statement: Generated during the study. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Meidut-Kavaliauskien, I.; Aranskis, A.; Litvinenko, M. Consumer satisfaction with the quality of logistics services. Procedia Soc. Behav. Sci. 2014, 110, 330–340. [CrossRef] 2. Chen, M.C.; Chang, K.C.; Hsu, C.L.; Xiao, J.H. Applying a Kansei engineering-based logistics service design approach to developing international express services. Int. J. Phys. Distrib. Logist. Manag. 2015, 45, 618–646. [CrossRef] 3. Rijk, R.; Veraart, R. (Eds.) Global Legislation for Materials; John Wiley & Sons: Hoboken, NJ, USA, 2010. 4. Watkins, T. Corrugated board packaging. In Packaging Technology; Woodhead Publishing: Cambridge, UK, 2012; pp. 240–261. 5. Biancolini, M.E.; Brutti, C. Numerical and experimental investigation of the strength of corrugated board packages. Packag. Technol. Sci. Int. J. 2003, 16, 47–60. [CrossRef] 6. Biancolini, M.E. Evaluation of equivalent stiffness properties of corrugated board. Compos. Struct. 2005, 69, 322–328. [CrossRef] 7. Biancolini, M.E.; Brutti, C.; Porziani, S. Corrugated board containers design methods. Int. J. Comput. Mater. Sci. Surf. Eng. 2010, 3, 143–163. [CrossRef] 8. Lansmont, R.D. Six-Step Method for Cushioned Package Development, Revised ed.; Lansmont Corporation: Monterey, CA, USA, 1992. 9. ASTM D1596. Standard Test Method for Dynamic Shock Cushioning Characteristics of Packaging Material; American Society for Testing and Materials: Philadelphia, PA, USA, 1997. 10. ASTM D4168. Standard Test Method for Transmitted Shock Characteristics of Foam-in-Place Cushion Materials; American Society for Testing and Materials: Philadelphia, PA, USA, 1995. 11. Gorman, S.P. In-package methods improve shock, vibration testing. Packag. Technol. Eng. 1997, 6, 25–30. 12. Schueneman, H.H. A comparison of three different cushion test methods. In Current Trends in Protective Packaging of Computers and Electronic Components; ASTM International: West Conshohocken, PA, USA, 1988. 13. Guo, Y.; Xu, W.; Fu, Y.; Wang, H. Dynamic shock cushioning characteristics and vibration transmissibility of X-PLY corrugated paperboard. Shock Vib. 2011, 18, 525–535. [CrossRef] 14. Wickramasinghe, V.K. Dynamics Control Approaches to Improve Vibratory Environment of the Helicopter Aircrew. Ph.D. Thesis, Carleton University, Ottawa, ON, Canada, 2013. 15. Wang, Z.W.; Sun, Y.C. Experimental investigation on bending fatigue failure of corrugated paperboard. Packag. Technol. Sci. 2018, 31, 601–609. [CrossRef] 16. Mills, N.J.; Masso-Moreu, Y. Finite element analysis (FEA) applied to polyethylene foam cushions in package drop tests. Packag. Technol. Sci. Int. J. 2005, 18, 29–38. [CrossRef] 17. Haj-Ali, R.; Choi, J.; Wei, B.S.; Popil, R.; Schaepe, M. Refined nonlinear finite element models for corrugated fiberboards. Compos. Struct. 2009, 87, 321–333. [CrossRef] 18. Nordstrand, T.; Carlsson, L. Evaluation of transverse shear stiffness of structural core sandwich plates. Comp. Struct. 1997, 37, 145–153. [CrossRef] 19. Nordstrand, T. Analysis and testing of corrugated board panels into the post-buckling regime. Compos. Struct. 2004, 63, 189–199. [CrossRef] 20. Garbowski, T.; Gajewski, T.; Grabski, J.K. Estimation of the compressive strength of corrugated cardboard boxes with various perforations. Energies 2021, 14, 1095. [CrossRef] 21. Garbowski, T.; Gajewski, T. Determination of transverse shear stiffness of sandwich panels with a corrugated core by numerical homogenization. Materials 2021, 14, 1976. [CrossRef][PubMed] 22. CNS 2999. Packaging—Complete, Filled Transport Packages—Vertical Impact Test by Dropping; CNS: Taipei, Taiwan, 1983. 23. ASTM D775. Standard Test Method for Drop Test for Loaded Boxes; American Society for Testing and Materials: Philadelphia, PA, USA, 1986. 24. ISO 2248. Packaging—Complete, Filled Transport Package—Vertical Impact Test ny Dropping; ISO: Geneva, Switzerland, 1985. 25. CNS 13190. Packaging—Complete, Filled Transport Packages and Unit Loads—Conditioning for Testing; CNS: Taipei, Taiwan, 1993. 26. Kenett, R.S.; Zacks, S. Modern Industrial Statistics: With applications in R, MINITAB, and JMP; John Wiley & Sons: Hoboken, NJ, USA, 2021. 27. Montgomery, D.C. Design and Analysis of Experiments, 10th ed.; John Wiley & Sons: Hoboken, NJ, USA, 2019. 28. MIL-HDBK-304C. Design; U.S. Department of Defense: Washington, DC, USA, 1997. 29. ASTM D4169. Standard Practice for Performance Testing of Shipping Containers and Systems; American Society for Testing and Materials: Philadelphia, PA, USA, 1989. Appl. Sci. 2021, 11, 5815 15 of 15

30. ASTM D3332. Standard Test Methods for Mechanical-Shock Fragility of Products, Using Shock Machines; American Society for Testing and Materials: Philadelphia, PA, USA, 1999. 31. CNS 10033. Complete, Filled Transport Packages—General Rules for the Compilation of Performance Test Schedules; CNS: Taipei, Taiwan, 1983. 32. CNS 14615. Testing Method of Performance for Package Cushioning Materials; CNS: Taipei, Taiwan, 2002.