Home

Dynamic analysis of in-plant logistics based on RFID data Simon De

Contents

1. Appendix D Performance report xi Table of Figures Figure 1 Forklift supply vs mizusumashi supply 3 Figure 2 Standard forklift Courtesy of Mitsubishi enne nns 3 Fig re 3 Crane Attachment an saar D S a Sa entente ene nee n e 3 Figure 4 Heavy duty forklift Courtesy Of Toyota aa 4 Figure 5 Narrow aisle forklift Courtesy of AisleMaster a 4 Figure 6 Tugger trainee eee e Eee tUi ea eee enen Nasua u uuu us assqa 4 Figure 7 Tow AGV with cart optically guided courtesy of IntelliCart 5 Fig re 8 FOK AGM enorme trate ai dte maius ubertim eben qu esee u ham 5 Figure 9 Working of an RFID system courtesy to Omicron a a 7 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16 Figure 17 Figure 18 Figure 19 Figure 20 Figure 21 Figure 22 Figure 23 Figure 24 Figure 25 Figure 26 Figure 27 Figure 28 Figure 29 From left to right a passive RFID tag an active RFID tag and an RFID reader 7 The six steps OF MSDP i tete tient tere teeth e a e tete toe teams rev da n 16 Hi
2. 21 4 T MISIDHIES uu au ha vie dees SS ama p o Maeve dte aha ay ee ent 21 41 1 Location of the vehicle neven eer t u a tek dote eee 21 4 1 2 Stat s of the vehicle o de comte eer eie e edens 23 4 1 3 Order list proBress eren Eee e PR ee E seas a Orde EAE Esa 25 4 27 ETfICI n yii ott e rtr ec te ttt i vtt ave tati tata eio dett us 26 4 271 Overall Efficiency tee tetto et ede ed eth 26 4 2 2 Overall tilizatioti cer e ER UR et rU Re RET ehe edat 33 42 3 Percentage loaded esee ere e AUI e cR UEM 33 4 3 Reliability zd de S tenete EP Paene 33 4 3 1 bosttime percentage iiit dtes eie Seg dake SERO ehe a eaa iu end eg 33 44 Productivity tere ehe ec testet site p h beer een RENE 35 4 4 1 Number of orders picked hour sese 35 4 4 2 Number of orders picked driven km a sense ensereenn 35 Metric Presentation u c ctam ies 36 5 1 Definition and key features of a dashboard 36 5 2 Dashboard design ete t nte ttt t e ndi tet e tede ts 37 5 244 Used timeframes 0 eate tet meet ter eR a tete eit 37 5 2 2 Multi screem desig Minne e ERE a e RR Ed e RR oaa aria 38 5 23 Screen 1 S mmiary VIGW isi eee piene RE ka len cie 38 5 2 4 Screen 2 Visibility sins eerte P Rl E eb reset np Pes aret idus 40 52 5 Screen 3 ETfICIenCyas iiie rete
3. 8 2 2 3 Dealing with the unreliability of RFID data 10 2 2 4 Current applicatiOris eer te ER re E EE Tr ee NER ERE 10 2 3 Applied methods for real time analysis of in plant logistics 11 2 3 1 Benefits of the use of RFID technology within in plant logistics 11 2 32 Applications tee ete estet 12 3 Design of a performance measurement system 15 3 1 Measurement System Development Process 15 3 1 1 Step 1 Define the need for measurement 16 3 1 2 Step 2 Define what we d 1 u a n na a Rua a a Ruay s 17 3 1 3 Step 3 Define what we must excelat 17 3 1 4 Step 4 Define how we know if we re successful 18 3 1 5 Step 5 Impl mentthe MS i rne VR Hanon thee Fe dens 19 3 56 Step 6 Utilize the MS emer ente ree E HERES WR Pe seines 19 viii 3 2 Hierarchy of the designed metrics 20 Metric development and programming
4. Figure 47 Result in the route selection tool In the figure above the blue lines represent the actual route that was used to complete the order The red striped lines represent the shortest possible route As can be seen here the taken routes clearly differ from the proposed routes This explains the low efficiency that was seen in the performance report 76 8 Expansion to a real life case In this chapter the expansion of the designed metrics and dashboard to a real life scenario is briefly discussed Though no real life case was covered in this thesis a couple of remarks can be made about the difference between the performed test setup and a real life case and how to use the gathered knowledge in a real situation 8 1 Differences between a real life case and the test setup The purpose of this thesis was to develop new dynamic analysis methods for the internal logistics The proposed metrics and their presentations were tested by monitoring a toy train Though this test was useful to give an idea of how to measure the performance of a logistics vehicle some differences between the test setup and a real life scenario do exist The first big difference is the scale The test setup consisted of an area of approximately 25 m Ina real life situation the used area would be the entire plant layout This area is thus much larger than the one used in the test Some of the consequences of this larger scale are e B
5. measurement Step 6 Utilize the MS Step 5 Implement the MS Step 4 we re successful Figure 11 The six steps of MSDP 3 1 1 Step 1 Define the need for measurement In the first step of the MSDP the purpose of the measurement system is determined and also the types of information that are needed for this Purpose of the measurement system e Monitor the performance of the internal logistics e If multiple logistics vehicles are deployed at the same time check if the workload is equally divided amongst all the logistics workers e Make the internal logistics process more visible e Check for improvements in Work method Equipment Layout Required information Are the used work methods efficient Are there too many breakdowns due to bad equipment Are there obstructions that can hinder the logistics vehicles or are the distances between different areas too big The more information that is available the easier it will be to monitor the performance of the internal logistics However the subject of this thesis is to develop new analysis methods for internal logistics based upon RFID data Therefore the available information will be limited to the RFID data and some other more important information The required and available information here is e RFID data e Order list Location of the vehicle s Contains the details of the orders that need to be fulfilled 16 3 1 2
6. Order list details Progress Order Load Unload Standard Standard time Driveto Load Drive to Unload Complete NBR zone zone time Loading Unloading loading unloading point l sec l sec point l l l 1 Lane2 Lane5 10 10 V V v4 V V 2 Lane7 Lane3 10 10 V V x x 3 Lane4 Lane8 10 10 x x x x x Here the green checkmark represents completed tasks the orange exclamation point represents tasks that are being handled and the red cross represents tasks that are not yet started 25 4 2 Efficiency 4 2 1 Overall Efficiency Overall efficiency compares the actual time taken to complete an order to the expected time to complete an order The actual time can be derived from the status update and the order list progress The expected time is based upon the assumption that the vehicle takes the most efficient route shortest path at the ideal average speed predefined speed and performs the loading and unloading actions at the predefined standard times ll _ Expected total time per order FERONT Actual total time per order The total time it takes to complete an order can be seen as the sum of the times it takes to complete each step of the order In general four steps will have to be completed during an order 1 Transportation to the loading point 2 Loading 3 Transportation to the unloading point 4 Unloading Therefore the overall efficiency can be subdivided in two metrics that focus on the diff
7. 59 Table 9 Adjustable parameters of the dashboard 64 Table 10 Format of the Cleaned RFID data U enne 66 Table 11 Format of the RFID data in the dashboard excel file 66 Table 12 Format of the Order list esent tenente ennt nnne nnne 67 Table 13 Format of the network data n n nn n 68 Table 14 RFID data sheet result after step 1 nnee enen enenneenerserrenensnnnensseversenenannenesenennn 72 Table 15 RFID data sheet result after step 2 72 Table 16 RFID data sheet result after step 3 73 Table 17 RFID data sheet result after step 4 73 Table 18 Update of the Error Event Manager when the status changes to Error 73 Table 19 Update of the Error Event Manager when the error problem is resolved 73 Table 20 Order list with filled in start and end time eene 74 Table 21 Order list with actual and shortest possible distances 74 Table 22 Order list with filled in efficiencies 74 xiv 1 Introduction
8. Nbr Metric KPA Visibility 1 Location of the vehicle 2 Status of the vehicle 3 Order list progress KPA Efficiency 4 Overall efficiency Transportation efficiency 5 6 Average speed 7 Route efficiency 8 Loading efficiency 9 Unloading efficiency 10 Load Unload time variance 11 Overall Utilization 12 Percentage loaded KPA Reliability 13 Lost time percentage 14 Vehicle reliability 15 Mean time to repair 16 Route reliability 17 Picking reliability KPA Productivity 18 orders picked hour 19 orders picked Km Een meer uitgebreide uitleg van het PMDS en de verdere ontwikkeling van de ontworpen metrics kan worden teruggevonden in de thesis die bij dit artikel hoort Dynamic analysis of in plant logistics based on RFID data IV PRESENTATIE VAN DE METRICS Om een doeltreffende performance measurement system te bekomen moeten de ontworpen metrics op een duidelijke manier worden voorgesteld zodat de prestatie van de interne logistiek gemakkelijk kan worden afgelezen Hierbij wordt een dynamisch dashboard ontwikkeld omdat dit in staat is om de prestatie in real time weer te geven In de thesis wordt het ontwerp van een dashboard die uit meerdere schermen bestaat besproken Voor het ontwerp van dit dashboard en de uitleg erbij wordt nogmaals verwezen naar de thesis zelf Naast het voorgestelde dashboard word
9. The route reliability focuses on the portion of lost time that is caused by the vehicle going off track Off track here means that the vehicle is driving in places it should not be and corrective action would be needed to get the vehicle back on track This error could occur due to a mistake of the driver or due to problems within the plant infrastructure e g blocked corridors bad signalisation etc P LostTime due to going Off track Vehicle reliability En C 34 4 3 1 3 Picking reliability Lost time can also be caused by inaccuracies of the loading and unloading actions The wrong items can be picked or the unloading can happen on the wrong place This is however difficult to measure with RFID data alone Manual input will be needed here to indicate and explain the error However other errors can also occur during the loading or unloading The items that have to be transported can be hard to reach during the loading or hard the place on the correct location during unloading These errors can be detected if the loading or unloading action takes too long As mentioned under 4 2 1 2 Loading and unloading efficiency an upper bound for the loading and unloading action is given If this upper bound is exceeded a picking error can be detected The picking reliability can be seen as the portion of lost time caused by errors that occur during the loading or unloading of a logistics vehicle m na Lost Ti
10. The user must enter the name of the zone and the coordinates of the four corners Then for each bordering line the user has to determine where the zone is located against the position of the line To facilitate this the zone is displayed in a screenshot and the current line is highlighted in blue on this screenshot The instructions for the use of this tool are given in the lower left corner of the window In this example the zone is located to the right of the current line so the greater than checkbox is chosen This is repeated for the other three lines Once all the steps in the tool have been completed the new zone is added to the Zones sheet and can be used in the dashboard 7 3 3 4 Available routes and intersections This last input is also connected to the layout of the manufacturing plant or in this test the layout of the train track To be able to calculate the efficiency of the train Dijkstra s Shortest Path algorithm needs to be calculated For this algorithm a network or graph with the enclosed nodes and edges needs to be defined As seen under 4 2 1 1 2 Route efficiency a plant layout can be translated to a network by translating every point of interest loading points unloading points intersections etc to a node and defining the edges roads that connect each node The network data needs to be entered into the ShortestPathCalculations sheet in the following format Table 13 Format o
11. To maintain their competitive advantage companies need to make sure they excel in their business processes In the last few years more and more companies have started implementing lean principles on their production process By implementing performance measurement systems the performance of the production can be continuously monitored and improved However the in plant logistics also requires attention and this can be a very time and money consuming task Nowadays the performance of the in plant logistics can already be measured though this is mostly done afterwards instead of during the performed logistics tasks To keep up with the increasing complexity and fast paced behaviour of in plant logistics new dynamic analysis methods should be developed The goal of this thesis is to develop new analysis methods so the performance of the in plant logistics can be monitored in real time The subject of the analysis will not be the entire in plant logistics but only the performance of the logistics vehicles RFID technology will be used to acquire the real time locations of the logistics vehicles Based upon this information and the task description order list of a logistics vehicle some metrics will be designed to measure its performance These metrics will then be used to develop a dashboard which will visually displays the performance of the logistics vehicles First an overview of the used literature is given Next the development of a
12. Off track Table 15 RFID data sheet result after step 2 Timestamp Coordinates Distance Speed Area Purpose Status hh mm ss x y m m s of area 14 12 07 4 401 5 757 04143 0 143 Lane 5 7 3 4 3 Step 3 Calculation of the purpose of the area The purpose of the area is needed to calculate the status of the vehicle Under 4 1 1 Location of the vehicle the purpose of the area was described as the functional name of a zone as opposed to the descriptive name of a zone that is used to identify the area In this step the current area Lane 5 in the example is compared with e The loading and unloading zone of the current order e The predefined parking area s e Off track locations 72 If the current area is none of the above the purpose of the area will be Driving lane by default In this case Lane 5 is noted in the order list as the unloading point for the current order Table 16 RFID data sheet result after step 3 hh mm ss x y m m s of area 7 344 Step 4 Calculation of the status of the vehicle Timestamp Coordinates Distance Speed Area Purpose Status 14 12 07 4 401 5 757 0 143 0 143 Lane5 Unloading station The calculation of the current status of the vehicle is a rather complex process It can best be explained by a flowchart This extensive flowchart is added in the appendices at the end of the thesis see appendix B
13. au se Jjpu SseiSo1d ayy azijensip swia qoud sayyo pue sip J p ous aj21Yyan ayy jo fouednooo au azijensi Jouapyja anou pue uoneoo jo main 33 nb e moll Japso au u sse1Bord pue sapso quo uno ayy Aejdsia 40139 40 lp Bupeoq un Buntem Juag q ue smeas ou euoz uonezo jo easy s 1euipioo2 papeoj a8equacsad inn IIP AO eoueueA euin peolun peo1 Jouopiyje Surpeojun Aouarouyja Sujpeo Aouapijje einoy paads aSe1oAV 4ouapyje uoneuodsue1j ouai lesa 0 asy o ueuuojiad Ady ss 1301d 3S1 JopJO ap y a ou Jo smexs 11490 au jo uOReIOT Ssa20Jd sonsio euj lu au Jo AqIqISIA ay3 A01duui easy 9IUEWIOHIAd o JaUMO aman Jo asodind aman ejnuuo4 o pue uonjujag jeuoneuado ue d uonaajjog eeq uBisaq jeAeiuod XIE 1u uido A d SIJ N uoneaypeds ual Flowchart of the Status update Appendix B snes JOS pspuejs iiyspueys ilo J q oua BunieM noejeq oua Jo uogeinp eumuo 10 BunieM noejeq zoug Jo uoneunp anua 104 Buntem Snes snes SES Duipeo1 snels snes Duipeo1 Snes SMEIS SMIS Y A r3 4 piousaJu jjedd puy piouseiujjedd lt pjouseJuJous gt pioyseuyLJeddn lt plousaJu Lamo ueewjag piouseJu LI A01 gt M 2 Jo uoneing gt lt s ua t eue unuq uonejs Buipeojun ployse ulJeddn lt piousaju LiamoT uveomjeg
14. pioyseuyLJeddn puy plous u LI A01 gt I aan lt gt Jo uoneJnq 4 IPI DELL HO JOUR SES Xem snis sn ls Bunya MOBIL HO JONA SMIS SNES 4 4 LS ES joe z Supyed PALIO 39811 HO lt gt YORIL Jo uoneis Bulpeo7 mon jo ho ee UA x uole007 x29uo eae ay jo asodin eui sl EUM uone07 X90049 A S3A bi ejolyen eui S Z EN smeis Y au ANIJE ih 9 51043 Em uone2013u240n e sajeuipJ00 graue 4T ue graue ST ue1 tT ue ET aue7 ZT aue TT ue OT aue 6 aue g ue1 ue1 9 ue1 e 5 aue gt taue ue Tauern Taue Test dashboard output UL TE vT 8T 0 vT LU O bT 0 0 bT cVO0c vt 0 02 T 9T 6L vT VLI 6L vL TELE VI 80 ST tT Liv T VI Scy vL OZ TT PT A x uone3o1 uogeing unmusgs SayNuUIW OT 3se ur 28s ZO ew YSO EOL ipaq 2013388 742243430 40413 0 iN Met u ON 0 0 saznulw QT 3se u aui 350 jo s sne u lqoid euin 3s0 ui eSueu uonezipno ur eSueu uoneaoy 10413 SeyNUIW OT sey 14euonieuSeds Hoday zud 2 Appendix C 11 pieoquseq Aadw3 SOT SLvI SC tlest vT L 1 1 Buji 8 Buipeo1m REESE Supeogung E SugieM E spia 10438 so3nuiu QT Jad Aauednaag sn3e3s spiri Buipeojun i
15. 2 7 Angeles R 2005 RFID Technologies Supply chain applications and implementation issues Information Systems Management Journal winter 2005 p 51 65 Arkan l 2010 Data analysis of a prototype RFID system and results Industrial Management Faculteit Ingenieurswetenschappen UGent Chow H Choy K L Lee W B and Lau K C 2006 Design of a RFID case based resource management system for warehouse operations Expert Systems with Applications 30 2006 p 561 576 Coimbra E A 2009 Chapters 9 and 10 Internal logistics flow Total Flow Management Achieving Excellence with Kaizen and Lean Supply Chains p 95 128 Zug Switzerland Kaizen Institute Consulting Group Ltd Few S 2006 Dashboard design for rich and rapid monitoring http www pmone de fileadmin documents studien Dashboard_Design_for_Rich_and_R apid Monitoring pdf Gonzalez H Han J Li X and Klabjan D 2006 Warehousing and analyzing massive RFID data sets www cs uiuc edu hanj pdf icdeO6 whrfid pdf Jeffery S R Garofalakis M and Franklin M J 2006 Adaptive cleaning for RFID data streams Proceedings of the 32 international conference on Very Large Data Bases p 163 174 Kootbally Z Schlenoff C Madhavan R 2009 Performance assessment of PRIDE in manufacturing environments http info ornl gov sites publications files Pub21558 pdf Ludwig T and Goomas D 2009 Real time performance monitoring
16. 45 Used equipment RFID reader left and active RFID tag right courtesy of Ubisense 46 xii Figure 30 Figure 31 Figure 32 Figure 33 Figure 34 Figure 35 Figure 36 Figure 37 Figure 38 Figure 39 Figure 40 Figure 41 Figure 42 Figure 43 Figure 44 Figure 45 Figure 46 Figure 47 Location of an RFID tag in the Ubisense software 47 Trails of multiple tags in the Ubisense software 47 Layout of the train track with the four RFID readers 48 Tag range parameters Selection of the QoS 49 Layout of the train track with the defined zones 52 Cropped test dashboard output 55 Total lost time as displayed on the performance report 56 Causes of the low efficiency as indicated on the performance report 57 Errors that occurred during the test 58 Relationship between the number of updated rows and the duration of the update 59 Test dashboard i e etit etr tette t acit title 62 Too
17. 85 3 Lane 4 Lane 8 14 13 14 14 13 22 14 13 41 14 13 52 125 91 91 88 4 Lane 7 Lane 3 14 14 05 14 14 09 14 14 49 14 15 07 250 56 65 47 5 Lane 2 Lane5 14 15 11 14 15 20 14 15 46 14 15 55 111 111 53 67 6 Lane 7 Lane3 14 17 17 14 17 23 14 18 35 14 18 49 167 71 mae 24 y Lane 5 Lane 7 14 18 56 14 19 13 14 19 55 14 20 02 14396 9596 8 Lane 2 Lane5 14 24 05 14 24 15 14 24 26 14 24 38 100 83 15 9 Lane 7 Lane 3 14 24 48 14 24 59 14 25 14 14 25 25 91 91 97 89 10 Lane2 Lane5 14 26 27 14 26 35 14 26 46 14 26 57 125 91 Bijma 35 11 Lane 2 Lane5 14 27 17 14 27 28 14 27 40 14 27 50 91 100 67 77 12 Lane 2 LaneS 14 28 10 14 28 20 14 28 31 14 28 42 100 91 68 77 13 Lane2 Lane5 14 29 32 14 29 41 14 29 52 14 30 03 111 91 49 14 Lane 18 Lane 9 14 35 08 14 35 25 14 36 13 14 36 31 59 56 50 15 15 Lane 18 Lane9 14 37 51 14 38 08 14 39 20 14 39 38 59 56 65 29 16 Lane9 Lane13 14 39 39 14 39 56 14 53 38 14 53 48 59 100 4 17 Lane 9 Lane 13 14 54 04 14 54 14 14 54 30 14 54 39 100 111 92 98 Figure 37 Causes of the low efficiency as indicated on the performance report Finally some conclusions can be drawn from the errors in the Error Event Manager In total 22 errors occurred during the test The planned errors as described under 6 4 Used order list were correctly analyzed by the dashboard and are indicated in blue on figure 38 In this figure the Off track error is also indicated on the spaghetti chart
18. Loading gt Driving gt Unloading apart from unexpected errors or waiting periods in between Every time the status changes to Loading or Unloading the start time is recorded in the OrderList sheet When the loading or unloading action is finished the end time is also recorded This way the progress of the order list can be followed in the OrderList sheet Table 20 Order list with filled in start and end time Order list details Times hh mm ss Order Load Unload Standard Standard time Load Load Unload Unload NBR zone zone time Loading Unloading start time end time start time end time sec sec 1 Lane2 Lane5 10 10 14 11 48 14 11 57 14 12 11 14 12 23 Apart from the recorded times the actual travelled distance the shortest possible distance to a loading point and the shortest possible distance between the loading point and the unloading point are added in the table The shortest possible distance is calculated by using Dijkstra s algorithm as described under 4 2 1 1 2 Route efficiency Table 21 Order list with actual and shortest possible distances Order list details Distances m Order Ideal distance to Driven distance Ideal distance to Driven distance NBR U loading point toloading point unloading point to unloading point 1 m 2 47 16 04 l 4 64 5 37 Once the times and distances have been entered the efficiencies can be calculated as seen
19. Speed Threshold is defined When the speed is lower than this threshold the vehicle is considered to have stopped moving In the table above this threshold is set at 0 11 m s This value is used in the test setup as it provides a good result The two Handling Time Thresholds are used to determine the minimum and maximum duration of a loading or unloading action These thresholds are expressed as percentages of the standard time of the specific action This standard time is included in the used order list To explain this better a short example is given Suppose an order is being executed and the vehicle has stopped to unload items The standard time for this unloading action is 10 seconds according to the given order list The thresholds of the previous table are used Here the Lower Handling Time Threshold is set to 50 which means that the vehicle has to stand still for at least 5 seconds 50 of the given standard time before the action will be considered as a valid unloading action The Upper Handling Time Threshold is set to 180 meaning that the loading action can take at the most 18 seconds Once this threshold is reached the status will change from Unloading to Error The Error Threshold has a similar meaning it determines how long a vehicle has to stand still before it is considered to be defected The last parameter is the Ideal average speed This parameter presents the ideal speed of the vehicl
20. The coiled antenna of the reader creates a magnetic field which detects the presence of the tag The tag subsequently draws energy from this magnetic field and sends waves back through its own antenna These waves are then interpreted by the reader and transformed into digital information This is how the EPC or Electronic Product Code of a tag can be read and the object with the tag attached to it can be identified Electromagnetic ield RFID Transponder Reader Reader Coil I l Transponder Coil Figure 9 Working of an RFID system courtesy to Omicron In general RFID systems can be divided into three classes active passive and semi passive The difference between these three classes is the power supply of the tags Active RFID tags require an own power supply This is needed to power the internal circuits and also to broadcast radio waves to the RFID readers The power supply can be provided by connecting the tags to a powered infrastructure or by simply using an integrated battery In this last case the lifetime of the tag is limited by the battery s capacity However active tags do have mechanisms that allow them to expand their lifetime some active tags have the possibility to go into sleep mode when the tag is not moved for a certain duration so it can preserve its power e g Ubisense tags Because active tags provide their own power to broadcast signals the signal is stronger and can be sent over longer d
21. Total lost time 18 1 min Bb Defect 17 85 min B Off Track 0 25 min Cause of errors Figure 36 Total lost time as displayed on the performance report Apart from the low utilization the performance report also shows that the efficiency of the vehicle is insufficient 52 When the order list in the report is reviewed it can be seen that this low efficiency is mainly caused by the low route efficiency of most orders the routes between the various loading and unloading points were poorly chosen On the following figure the efficiencies of all the orders are displayed The route efficiencies of lower than 50 have been indicated in red as they are one of the most import factors of the low overall efficiency In some orders it also occurred that the overall efficiency is low while the load unload efficiency and the route efficiency are high er indicated in green This was caused by defects during those orders which caused the vehicle to stand still for a long time and thus prolonged the total duration of the order 56 Efficiency and Utilization Average Efficiency of the vehicle Total Utilization of the vehicle 48 Order hann Unload Load Load Unload Unload NBR zone starttime endtime starttime endtime Efficiency Efficiency Efficiency Efficiency Lane 2 Lane 5 14 11 48 14 11 57 14 12 11 14 12 23 111 83 33 gt 42 2 Lane 7 Lane3 14 12 33 14 12 41 14 12 57 14 13 10 125 77 89
22. coordinates zone of the logistics vehicle gt Location coordinates zone of the loading and unloading point gt Layout of the plant available aisles intersections gt Shortest path algorithm The actual driven distance can easily be derived from the cleaned RFID data and the order list progress From the order list progress it can be seen when the vehicle begins a new order and how it advances in this order In the two transportation steps of the order each time a start and an endpoint will be defined These points are indicated by their start and stop times as tstart and tstop 28 The distance between the consecutive data points is calculated and the sum of all these distances between tstart and tstop is taken tstop Actual Driven Distance gt x Xr yt Vi t tstartt1 The shortest distance in meters and the accompanying shortest path can be calculated by using an existing shortest path algorithm Because in this shortest path there can be no negative edges the distance between two points can never be negative Dijkstra s Algorithm can be used Before the algorithm can be run first a graph or network with all the needed nodes vertices and arcs edges will have to be constructed This graph will be a representation of the plant where the nodes are all the possible points of interest loading points unloading points intersections etc and the arcs will be the aisles or roads that connec
23. 403 3 Kootbally Z Schlenoff C Madhavan R 2009 Performance assessment of PRIDE in manufacturing environments http info ornl gov sites publications files Pub2 1 558 pdf last Update lases 161557 TRAIN 1 Figuur 1 Ontwerp van het test dashboard Empty Dashboard Update stop Current Status Current Location Current Order 18 Progress e to loadin Status Occupancy per 10 minutes Spaghettichart last 10 minutes Error location Error Event Manager Start time End time 144457 145456 Total lost time e last 10 minutes Starttime Duration Location zone k omm xm desos 00025 Dynamic analysis of in plant logistics based on RFID data Simon De Buyser Supervisor s H Van Landeghem V Lim re Abstract In this thesis new dynamic analysis methods are developed to monitor the real time performance of in plant logistics vehicles In first instance a set of metrics is developed by using the Performance Measurement Development System A dashboard is then designed to dynamically present the defined metrics The accuracy and real time aspect of the metrics and their presentation in the dashboard are then tested in a test setup which consists of a toy train with an RFID tag attached to it The results of the test prove that the dashboard is capable of providing correct
24. 54 3 285 3 387 0 582 14 54 53 2 718 3 522 0 212 14 54 52 2 515 3 584 0 230 14 54 51 1 833 3 350 0 249 EEEN Dess TS As mentioned before the cleaned RFID data is automatically taken from the excel file that is referenced in the Parameters sheet This RFID data is then placed in the RFID data sheet and completed by adding the calculated distance between the points This is displayed below Table 11 Format of the RFID data in the dashboard excel file Timestamp Coordinates Distance Speed hh mm ss x y m m s 14 54 56 4 038 3 364 0 590 0 590 14 54 55 3 448 3 359 0 166 0 166 14 54 54 3 285 3 387 0 582 0 582 14 54 53 2 718 3 522 0 212 0 212 14 54 52 2 515 3 584 0 722 0 230 14 54 51 1 833 3 350 0 249 0 249 14 54 50 1 655 3 176 0 665 0 665 66 7 3 3 2 Used order list To analyse the performance of the internal logistics first the tasks of the logistics vehicles need to be known These tasks and the order in which they need to be done are given in the form of an order list Here the assumption is made that every tasks consists of loading one item transporting it to another place and unloading it there The format of the order list is displayed below Table 12 Format of the Order list Order Load Unload Standard time Standard time Description NBR zone zone Loading sec Unloading sec 1 Lane 2 Lane 5 10 10 Info 2 Lane 7 Lane 3 10 10 3 Lane 4 Lane 8 10 10 4 Lane 7 Lane 3 10 10 5 La
25. 847 4050 lane2 5 Defect 14 17 31 0 00 21 2 995 4 547 Lane1 6 Defect 14 19 14 0 00 01 3 856 5 30 LaneS 7 Defect 14 19 16 0 00 28 3 612 5 652 lanes 8 Off track 14 20 03 0 00 01 0 977 4702 lane7 9 Off track 14 20 12 0 00 01 1 315 3 772 lane7 10 Off track 14 30 03 0 00 05 3 462 5 62 lanes n Off track 14 30 17 0 00 01 3 282 5 492 lane5 12 Defect 14 30 18 0 00 25 3 279 5 525 lanes 13 Defect 14 31 14 0 03 04 2 500 5 642 lane 5 m vii Figure 1 Design of the test dashboard Table of Contents Acknowledgements u uu u Breen hen ereleden adeleae hal vo rends i OVEWIEW H iii Extended abstract Nederlands ccccccccccssssececssssececsssaecesessececsesaececseasececessaaeeeesesseeeceeaaeeeeseaeeeeneaas iv Extended Abstract English nie rtr eee te ee Ere T S Ne a Eee TEN Te BORNE renee vi FablecOf EIB L esr rare tan icd i ee tie dts ee heron EI xii ME te A en xiv Ts SIATKOGUETION eren eneen tan tet teta teret oett dede 1 2 Literature study een au eee aaa ated Ge cc ea ti m eT eos 2 2 1 a plant 0E OR 2 2 1 1 Logistics vehicles u a LAI UI San a Nu AH SS S Nu BNIE Sus 3 2 2 RFID technology sl oi u exe Pe OR HIS IGnnI 6 2 2 1 Introduction to RFID technology 6 2 2 2 Dealing with massive RFID data sets
26. Eekelen 62 7 3 Working of the test dashboard aaan ai a a e aaa oiai raie heini 63 7 3 1 Structure of the Dashboard excel file a 63 7 3 2 Adjustable parameters RUE a Wis oe Gidea ene ee ead 64 7 3 3 Inputs of the test dashboard annae ennen enseneer versen enansenssenserennenensneneseens 66 7 3 4 Working of the dashboard update 70 TA Postsanalysis stenen inenten er et t de ben Ee 75 TAA Perftormance TGport bte t de e e eke 75 7 4 2 Route efficiency check ssi eR ME dep RERUM AER 75 8 Expansion to a real life tasedi hanorina inaran rriari En e a nnne eene uA u niri nnns 77 8 1 Differences between a real life case and the test setup 77 8 2 Changes needed to the test dashboard 78 97 So IV ELL m T E 79 10 acne ee 81 11 Appendices Ea eve me dee ite tte utu ied ee tu tees teas 11 1 Appendix A Completed Metrics Development Matrix II Appendix B Flowchart of the Status update Hl 11 2 11 3 Appendix C Test dashboard output
27. El Houssaine Aghezzaf II II II Faculteit Ingenieurswetenschappen en Architectuur UNIVERSITEIT Academiejaar 2010 2011 GENT Dynamische analyse van interne logistiek op basis van RFID data Simon De Buyser Promotor s H Van Landeghem V Lim re Abstract In deze thesis worden nieuwe dynamische analysemethodes voorgesteld om de prestatie van interne logistieke voertuigen te meten Eerst wordt er gebruik gemaakt van het Performance Measurement Development System om een set metrics te ontwerpen Vervolgens worden de ontworpen metrics visueel gepresenteerd in een dynamisch dashboard De nauwkeurigheid en correctheid van de metrics en hun voorstelling in het dashboard worden nadien getest alsook de mogelijkheid om het dashboard in real time te updaten Hiervoor wordt een testopstelling gebruikt die bestaat uit een speelgoedtrein die door middel van RFID technologie kan worden gevolgd Keywords In plant logistics Performance measurement system Dashboard design RFID data Real time analysis I INLEIDING In de laatste jaren zijn meer en meer bedrijven begonnen met het implementeren van lean principles op hun bedrijfs processen Door het bevorderen van de flow van producten doorheen de supply chain en het continue verbeteren van de gebruikte productieprocessen kunnen deze bedrijven hun positie in de markt verstevigen Naast het optimaliseren van de productieprocessen moet er ook aandacht besteed worden aan de interne logistiek O
28. Step 2 Define what we do In this step the unit of analysis is determined Also the mission and the vision why do we exist are defined Here the unit of analysis is the in plant logistics In plant logistics covers all the movement of goods through the plant and their storage underway Activities of the internal logistics include the storage of the supplied goods in warehouses supply to the border of line and reversed logistics within the plant More specific in this thesis the performance of the logistics vehicles is analysed The mission of the in plant logistics defines why it exists The mission here could be To provide logistic services when needed deliver on time and correct and maintain an efficient flow of goods through the company The vision of the internal logistics defines its desired future Here the vision could be To become the best in internal logistics by obtaining a stable and predictable flow of goods through the plant In this the efficiency and reliability of the internal logistics will be improved continually 3 1 3 Step 3 Define what we must excel at This step is important for the development of good metrics These metrics have to be aimed at the most important areas of the in plant logistics First there needs to be determined what these areas are or what the internal logistics department has to excel at These important points are the Key Performance Areas or KPAs This is basically an area
29. The other Defects are caused by stops that lasted too long as mentioned before On the Error Event Manager some other Off track errors can also seen These errors are however not caused by the behavior of the vehicle but by the inaccuracy of the RFID data This can be detected quickly as the errors due to inaccuracy only last one or two seconds which is too short for a real error 57 Error Event Manager Nbr Problem Starttime Duration Location Zone X Y 1 Off track 14 11 20 0 00 01 1 459 5 171 Lane 6 2 Defect 14 14 28 0 00 17 5 394 4 998 Lane 4 3 Off track 14 14 47 0 00 01 5 831 3 977 Lane 4 4 Off track 14 15 08 0 00 01 4 847 4 050 Lane 2 5 Defect 14 17 31 0 00 21 2 995 4 547 Lane 1 6 Defect 14 19 14 0 00 01 3 856 5 730 Lane 5 7 Defect 14 19 16 0 00 28 3 612 5 652 Lane 5 8 Off track 14 20 03 0 00 01 0 977 4 702 Lane 7 9 Off track 14 20 12 0 00 01 1 315 3 772 Lane 7 10 Off track 11 Off track 14 30 03 0 00 05 14 30 17 0 00 01 3 462 3 282 5 362 Lane 5 5 492 Lane 5 12 Defect 14 30 18 0 00 25 3 279 5 525 Lane 5 13 Defect 14 31 14 0 03 04 2 500 5 642 Lane 5 14 Off track 14 34 31 0 00 01 2 460 5 482 Lane 5 15 Defect 14 35 26 0 00 26 3 108 0 126 Lane 18 16 Defect 14 36 31 0 00 30 2 999 3 532 Lane 9 17 Defect 14 38 09 0 00 39 3 027 0 147 Lane 18 18 Defect 14 39 38 0 00 01 3 150 3 409 Lane 9 19 Defect 14 39 57 0 11 07 3 172 3 404 Lane 9 2 20 Defect 14 51 07 0 00
30. discussed further in this chapter 5 2 3 Screen 1 Summary view Summary view Vehicle location in the plant Amount of Lost time during the last day B Working idle Defect 1 3 Figure 21 Screen 1 Summary view Forklift 3 Legend Current Status Visibility w Visibility 2 Visibility w Working o Current Location Hall 2 Warehouse Hall 2 wW Idle e Current order 156032 156038 156028 w Defect o 5 oce I i sa progress Nbr of orders Ginie 6 15 5 15 6 15 50 50 50 2 5 2 5 2 5 Utilization 7 during the last day 9 oo 0 oo o 64 85 96 80 97 96 m a 5 4 Efficiency Efficiency Efficiency 50 50 50 a 2 5 2 2 5 c Overall Efficiency a during the last day 0 0 00 0 00 34 3 75 96 60 24 Forklift 1 E Forklift 2 h Forklift 3 38 The summary view presents the performance for every used logistics vehicle In the example above three vehicles all forklifts are used For each vehicle a couple of metrics are displayed The performance can then be compared amongst the three vehicles The used metrics fit into two timeframes current time and last day As the purpose of the summary view is to allow a quick overview of the performance per day the last day timeframe is best fitted for this However in case of errors or just to see the progress in the order list it is also useful to displa
31. electromagnetic field If the product is however not paid for the RFID readers at the exit of the store will pick up the signal and trigger an alarm 10 Baggage handling Instead of using bar coded labels RFID stick on labels can be used The RFID readers are then placed at various locations on the conveyor belts The advantage is that many tags can be read at one time and human error can be eliminated This method of baggage handling is already applied at the San Francisco International Airport amongst others Real Time Location Tracking Systems RLTS RFID tags can be used to track people vehicles or products within a company This can be done by placing RFID readers around the company that record all the tags in their vicinity When a tag is detected in a certain area this information is transferred to a database so the location of the tag is always known 2 3 Applied methods for real time analysis of in plant logistics In this thesis RFID tags will be attached to logistics vehicles to measure their performance Apart from the examples given before RFID technology also has some applications in the real time analysis of in plant logistics Some of the found research will briefly be discussed here 2 3 1 Benefits of the use of RFID technology within in plant logistics First of all some potential benefits of the usage of RFID technology in in plant logistics are given Tajima 2007 e Reduced Loss of products Products
32. finished faster than the goal time a red bar means that the order was late 43 5 2 6 Screen 4 Reliability Forklift 1 Reliability Return to Summary Error event manager Nbr Problem Time Duration Location Cause Defect 8 15 35 Warehouse Fill this in Off route 9 01 06 Hall 1 after the Off route 9 04 54 Hall 1 error occures Defect 9 12 42 Warehouse Pickingerror 10 45 05 Warehouse Off route 11 01 23 Warehouse Off route 11 35 41 Hall 1 Amountof lost time Causes of lost time Sidle E Defect B Working E Off route E Losttime E Picking error Total lost time 24 min per 8 hours Lost time per hour Location of error W Defect m off route B Pickingerror Figure 26 Screen 4 Reliability The used timeframe in this screen is that of the last day All the metrics used here are aggregated to present the data measured during an entire day This large timeframe was chosen because of the fact that normally errors shouldn t occur often Because of this low occurrence the smaller timeframes current time and last hour would not be interesting to present the lost time The amount of lost time is presented in a pie chart as a percentage of the total time A second pie chart zooms in on the lost time and indentifies the causes and their respective share of the lost time In the top right corner of the screen the Error Event Manager is displayed This table lis
33. information about the performance of the vehicle and can do so in real time Keywords In plant logistics Performance measurement system Dashboard design RFID data Real time analysis I INTRODUCTION During the last years more and more companies started applying lean principles on their business processes By constantly improving the used production methods companies can maintain their competitive advance Apart from optimizing the production methods the in plant logistics also needs to be improved to increase the general flow of products through the factory To be able to remove all the waste and find the improvement possibilities the in plant logistics process first needs to be visualized For this reason a performance measurement system needs to be set up Because of the increasing complexity and fast paced behavior of the in plant logistics real time analysis will be used to correctly capture the performance This real time analysis can be made possible by equipping the logistics vehicles and products with RFID tags and monitoring their movement through the factory The remainder of this paper is structured as following First a short literature study is given in which some similar researches are briefly discussed Secondly the development of the metrics is described Next the design of a dashboard is presented after which the designed metrics are tested and the results are given The paper ends with the conclusion of the re
34. is important that the dashboard works properly and correctly displays all the important metrics This is manually checked during the test The table underneath summarizes the results of the test Table 8 Performance of the dashboard at different update rates Update rate Nbrofupdatedrows Duration of the update Dashboard is displayed correctly 00 30 00 1800 29 seconds Yes 00 20 00 1200 18 seconds Yes 00 10 00 600 9 seconds Yes 00 05 00 300 4 seconds Yes 00 01 00 60 1 second Yes 00 00 30 30 1 second Yes 00 00 10 10 Less than 1 second Yes 00 00 02 2 Less than 1 second Yes 00 00 01 1 Less than 1 second No In this table also the number of the updated rows is given This number is equal to the total amount of time in seconds that has passed since the last update which is logical because the cleaned RFID data has one data point per second From this test it could be concluded that the duration of the update is directly proportional to the number of updated rows or to the update rate This relationship is shown in the graph underneath 35 30 25 20 15 10 Duration of the update sec 5 0 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Nbr of updated rows Figure 39 Relationship between the number of updated rows and the duration of the update From the tests it can also be concluded that the dashboard update can be displayed correctly until an update rate of one update every two seconds When th
35. logistics vehicles Eventually 19 metrics are designed which fit under 4 Key Performance Areas KPAs These KPAs and the accompanying metrics are displayed in the table underneath Table 1 Designed metrics Nbr Metric KPA Visibility 1 Location of the vehicle 2 Status of the vehicle 3 Order list progress KPA Efficiency Overall efficiency Transportation efficiency Route efficiency 4 5 6 Average speed 7 8 9 Loading efficiency Unloading efficiency 10 Load Unload time variance 11 Overall Utilization 12 Percentage loaded KPA Reliability 13 Lost time percentage 14 Vehicle reliability 15 Mean time to repair 16 Route reliability vi 17 Picking reliability KPA Productivity 18 orders picked hour 19 orders picked Km A detailed explanation of the PMDS and the further development of the mentioned metrics can be found in the accompanying thesis Dynamic analysis of in plant logistics based on RFID data IV PRESENTATION OF THE METRICS The designed metrics alone are not enough to form an effective performance measurement system The metrics need to be presented in a clear and concise manner so the performance of the in plant logistics vehicles can be easily read Because of the real time aspect that needs to be captured in this metric presentation a dynamic dashboard is designed This dashboa
36. plant logistics based on RFID data Simon De Buyser Summary The purpose of this thesis is to design and develop new dynamic analysis methods to monitor the performance of the in plant logistics vehicles First a literature study is done in which the in plant logistics and RFID technology are looked at Some similar researches are also discussed in this literature study Then by using the Performance Measurement Development System a set of metrics is defined These metrics are then be used to design a dynamic dashboard which displays the real time performance of the logistics vehicles Next a test is done to evaluate the working and real time aspect of the designed dashboard From the test it is concluded that the performance of the logistics vehicle can be displayed correctly and in real time a maximum update rate of one update every two seconds is obtained which is more than satisfying The thesis concludes by presenting some adjustments that can be made to the developed dashboard to make it accessible for real life cases Keywords In plant logistics Performance measurement system Dashboard design RFID data real time analysis Promotor prof dr ir Hendrik Van Landeghem Begeleider ir Ihsan Arkan Masterproef ingediend tot het behalen van de academische graad van Master in de ingenieurswetenschappen bedrijfskundige systeemtechnieken en operationeel onderzoek Vakgroep Technische Bedrijfsvoering ca S Voorzitter prof dr
37. priority the AGV s shortest path will be recalculated but with the blocked lane removed from the possible lanes Kootbally et al 2009 The collision avoidance described here could possibly be used as an expansion on this thesis when multiple logistics vehicles are deployed in one area The applications that are described here will be used as a guideline for the development of performance measurement system in this thesis In these three applications there is always a direct 13 performance feedback to the logistics vehicle The performance is reported to the driver of the vehicle and the instructions to perform the tasks are also given e g the route that needs to be taken the order in which the products need to be loaded or unloaded However in the thesis there will be no direct feedback to the logistics vehicle the real time performance will only be communicated to the plant manager through a specially designed dashboard 14 3 Design of a performance measurement system To be able to measure the performance of the in plant logistics first a Performance Measurement System needs to be defined This is a set of processes and tools that can be used to monitor the working of a unit of interest such as the internal logistics department allowing the user to draw conclusions and if necessary take action s based upon this An effective performance measurement system provides actionable information on a focused set of me
38. stops will be considered as errors by the dashboard The resulted dataset contained 45 minutes of RFID data This is of course a much smaller time interval than would be expected in a real life situation where the duration of a shift is 8 hours For this reason some of the used metrics have been adjusted especially for the test setup The details of these adjustments are discussed in the following chapter under 7 1 Used metrics and adaptations After the RFID data is gathered and cleaned it is loaded into the dashboard Here two things will be tested First the accuracy of the dashboard will be reviewed Here it is checked if the dashboard correctly displays the performance of the train during the test Secondly the real time capabilities of the dashboard will be tested This will be done by dynamically loading the RFID data into the dashboard 6 6 Results Accuracy of the analysis In the first test the entire gathered RFID dataset is loaded into the dashboard The data is thus not analysed in real time but afterwards The goal is to find out if the situation is correctly analysed by the dashboard The results of this test will be depicted in the test dashboard and also in a performance report that can be generated after the test this is one of the functions that is included in the developed dashboard file 6 6 1 Dashboard output The resulted view can be seen on figure 35 However due to the limited space here the dashboar
39. test During the last 10 minutes of the test the utilization had dropped to 33 The status pie chart verifies this the vehicle was in error for 67 of the time while there was no idle time during the last 10 minutes The low utilization is thus caused by an unusually high error rate and not because the vehicle was idle The result that is depicted here is not incorrect The huge amount of lost time had also been noticed during the test This lost time was mainly caused by stops between the different orders in the test As mentioned before at certain moments the sidetracks needed to be adjusted manually when switching from one section to another section of the track during which the train was standing still but not in the parking zone As the vehicle was standing still in places it is not normally supposed to stand still the vehicle was presumed to be defect In the upper right corner of the dashboard a spaghetti chart of the last 10 minutes is displayed Here also the locations of the errors are visible As can be seen on the map most errors occurred in either Lane 5 at the top of the map or Lane 9 in the middle of the map These were also the locations were the vehicle was standing still while the sidetracks were being adjusted The status bar chart presents a clear trend of the utilization of the train In the beginning right side of the bar chart the utilization was high and the lost time due to errors was low As the test progress
40. that is required for success These KPAs can t be measured directly but are supported by several metrics which will be designed in the next step Further a good KPA should be aligned with the mission and vision of the business unit and only a limited focused set of KPAs should be utilized Below the chosen KPAs in this thesis are given e Efficiency Increase the efficiency and keep the cost of the internal logistics to a minimum Reliability Increase the reliability of the internal logistics and analyze the causes of lost time e Productivity Increase the productivity of the internal logistics e Visibility Improve the visibility of the internal logistics process Though increasing the visibility is not really a KPA in itself it is needed because it supports the other KPAs When the logistics process is better visualized problems or inefficiencies can be detected much faster and improvements can be implemented better 17 3 1 4 Step 4 Define how we know if we re successful In this step metrics are designed to measure the performance in each KPA Here the complete specifications of the metrics are determined such as the way the metrics are portrayed and how the data for the metrics is acquired For this the Metrics Development Matrix MDM can be used The MDM is a table in which the chosen metrics and their details need to be entered At the same time it can be seen as a checklist to determine if the metrics are use
41. the dataset 3 As mentioned before the smallest time interval needed for the used metrics and their accompanying functions is equal to one second However in the raw data multiple coordinates per second are given to counter the imprecise scaling frequency A good approximation of the tag s location for each second can now be obtained by taking the median value of all the coordinates measured during that second 4 Further due to the used technology Ubisense s active RFID tags gaps in time can appear in the raw data as shown in the table below Table 5 Gaps in time in the logged RFID data Date x y 14 31 18 2377452 5 626453 14 31 19 2 365974 5 628467 14 33 47 2 360700 5 626945 14 33 48 2 351197 5 621039 These gaps are caused by the fact that the active RFID tags will go into sleep mode once they have stopped moving for a certain amount of time a built in motion sensor can trigger the sleep mode The gaps can be detected easily and the data can be cleaned by simply adding the data points in between the gaps the last known coordinates are copied 5 The resulted data is now in the correct format The speed is however further smoothed by applying statistical control limits to get a better result The exact method to obtain the corrected speed will not be discussed here as it is not a part of the thesis and it would lead us too far After the corrected speed is calculated the data is reversed
42. under 4 2 Efficiency The resulted efficiencies are then also added in the table Table 22 Order list with filled in efficiencies Order list details Efficiencies Order Load Unload Route1 Route2 Total Route Overall NBR U Efficiency Efficiency Efficiency Efficiency Efficiency Efficiency 1 ae 111 83 15 86 33 42 7 3 4 7 Step 7 Update of the visual presentations In this last step the graphs on the dashboard are updated The results of all the previous steps are now used as inputs for the graphs so the newest data can be displayed 74 7 4 Post analysis The dashboard offers the user real time information about the performance of the internal logistics vehicles However after the data is gathered it can be important to analyse this information so that valid conclusions can be drawn and possible improvements can be worked out Here two forms of post analysis are included in the dashboard 7 4 1 Performance report After the tests are run the user can press the Print report button on the dashboard Then a report is generated which contains all the important information about the performance during the entire test An example of this generated report is added in the appendices appendix D Performance Report 7 4 2 Route efficiency check The route efficiency is calculated for every completed order This data is not displayed in the dashboard but it is presented in the performance repo
43. 11 Off track 14 30 17 00 9 5 492 Lane 5 12 Defect 14 30 18 00 5 525 Lane 5 13 Defect 14 31 14 03 4 5 642 Lane 5 14 Off track 14 34 31 00 i 5 482 Lane 5 15 Defect 14 35 26 00 0 126 Lane 18 16 Defect 14 36 31 00 A 3 532 Lane 9 17 Defect 14 38 09 00 0 147 Lane 18 18 Defect 14 39 38 00 f 3 409 Lane 9 19 Defect 14 39 57 11 5 3 404 Lane 9 20 Defect 14 51 07 00 E 3 498 Lane 9 21 Off track 14 51 45 00 2 808 Lane 16 22 Off track 14 53 27 00 3 2 951 Lane 16 NBR zone Paca ica starttime endtime iia eine LaneS 14 11 48 14 11 57 14 12 11 14 12 23 Lane3 14 12 33 14 12 41 14 12 57 14 13 10 Lane8 14 13 14 14 13 22 14 13 41 14 13 52 Lane3 14 14 05 14 14 09 14 14 49 14 15 07 Lane5 14 15 11 14 15 20 14 15 46 14 15 55 Lane3 14 17 17 14 17 23 14 18 35 14 18 49 Lane7 14 18 56 14 19 13 14 19 55 14 20 02 Lane5 14 24 05 14 24 15 14 24 26 14 24 38 Lane3 14 24 48 14 24 59 14 25 14 14 25 25 Lane5 14 26 27 14 26 35 14 26 46 14 26 57 Lane5 14 27 17 14 27 28 14 27 40 14 27 50 Lane5 14 28 10 14 28 20 14 28 31 14 28 42 Lane5 14 29 32 14 29 41 14 29 52 14 30 03 Lane9 14 35 08 14 35 25 14 36 13 14 36 31 Lane9 14 37 51 14 38 08 14 39 20 14 39 38 Lane 13 14 39 39 14 39 56 14 53 38 14 53 48 Lane 13 14 54 04 14 54 14 14 54 30 14 54 39 A Ui b UN P Vl
44. 32 3 200 3 498 Lane 9 21 Off track 14 51 45 0 00 01 6 345 2 808 Lane 16 22 Off track 14 53 27 0 00 02 5 583 2 951 Lane 16 s 2 3 s Figure 38 Errors that occurred during the test 6 7 Results Real time performance of the dashboard The purpose of this second test is to see if the dashboard can keep up with the dynamic RFID dataflow and can still present the metrics correctly Here the same dataset is used as in the first test but instead of loading it into the dashboard all in one time the data is dynamically loaded into the dashboard To be able to perform this test the entire RFID dataset is copied to another excel file Datafile xlsx one row at a time by a programmed macro This new excel file is thus constantly updated as if the RFID data is directly logged to this file and it will serve as input file for the dashboard To check the behaviour of the dashboard while analyzing this real time data the update rate of the dashboard was varied between updating every second and updating every half hour For each update rate two things were examined e Duration of the update The calculations and functions that need to be done to make the dashboard work require a certain amount of time This time is measured by programming the dashboard so it fills in the start time in excel and when it is finished with the calculations it fills in the end time of the update 58 e Is the dashboard displayed correctly It
45. BA Visual Basics for Applications The VBA code will not be added in this thesis but some of the more important operational functions will be explained by flowcharts First the structure of the used excel file will be explained Then the adjustable parameters and the needed inputs will be discussed Finally the updating of the dashboard is described step by step 7 3 1 Structure of the Dashboard excel file Though the actual test dashboard only consists of one screen multiple sheets are used in the excel file The function of each sheet will be explained briefly as this helps to clarify the working of the dashboard TestDashboard xlsm e Dashboard This sheet is as the name says the actual dashboard It contains the presentations of the used metrics e CheckRoute As part of the post analysis the efficiency of the used routes can be checked here The efficiency is visually presented in this sheet by comparing the actual route to the shortest route on the layout of the train track e Report After the test is done a report can be generated to display a summary of the occurred events and the overall performance during the test When the report is generated it is added as a new sheet with as name Report the current date and time e DashboardCalculation Here the data for the pie charts the utilization gauge meter and the bar chart is stored The necessary calculations are done in VBA and th
46. D data The first step of the dashboard update is to check if there is new data available in the used RFID data file as entered in the Parameters sheet The timestamp of the last data point in the RFID data file is compared to the last timestamp in the dashboard file If these timestamps differ from each other it means that new data is available in the RFID data file All the new data points are then added to the dashboard file in the RFID data sheet and the distance between each consecutive point is calculated as described under 6 4 3 1 Cleaned RFID data If no new data was detected the dashboard generates a warning for the user and ends the update cycle The first three steps of the dashboard update affect the data in the RFID data sheet To visualize the progress during these steps the result is displayed in a small example at the end of the explanation of each step Table 14 RFID data sheet result after step 1 Timestamp Coordinates Distance Speed Area Purpose Status hh mm ss x y m m s of area 14 12 07 4 401 5 757 0 143 0 143 7 3 4 2 Step 2 Calculation of the area of location The area of location zone of each new data point is now calculated These calculations are done in the Zones sheet For every zone four statements are checked as described under 4 4 1 Location of the vehicle If the vehicles location is not within any of the defined zones it is considered to be
47. Details of the order eee Order list progress Current order wh Process within the order Route selection Transportation efficiency i Overall Efficiency per Average speed efficiency of the d order vehicle Loading and Overall Unloading cud Efficiency utilization of the efficiency DUUM MUSS vehicle Percentage loaded Details about each defect Vehicle reliability MTTR Reliability Percentage of Causes of lost CETTE NETS Details about lost time time each error Eb pono orders Loading accurac picked per hour E Y Productivity Picking accuracy Nbr of orders r Unloading me ee ey accuracy driven dist Figure 12 Hierarchy of the designed metrics 20 4 Metric development and programming In the previous chapter the development of a performance measurement system was described Based upon MSDP a set of clear metrics was designed which will be used to monitor the performance of the internal logistics Here the proposed metrics will be described more in detail and the required functions and formulas will be explained 4 1 Visibility 4 1 1 Location of the vehicle The location of the vehicle is considered on two aggregation levels e Detailed location Current coordinates of the location of the vehicle e Area of location The area where the vehicle is driving in e g aisle 1 in the warehouse hall x department y etc The detailed lo
48. Driving lane Locations which are only used as roads but have no other function Parking This location is meant as a parking place for idle vehicles e Offtrack Locations in a factory where the vehicle should not be Here it should be noted that the loading and unloading stations are not fixed They are dependent of the current order and the progress in that order and will be updated each time an order is completed and a new one is begun The parking is a predetermined area which will normally not be changed and as mentioned before the vehicle will be considered Off track when it is not within any of the defined zones Here the purpose of this zone will also be described as Off track 4 1 2 Status of the vehicle One of the most important metrics here for the real time analysis of the in plant logistics is the status update of the logistics vehicle This metric looks at the current status of the vehicle and also forms the basis for a lot of other used metrics and their accompanying formulas functions Here the possible status updates are Driving Driving around the plant in order to perform a given task e Loading Performing a manoeuvre to load an item on the vehicle e Unloading Performing a manoeuvre to unload an item from the vehicle e Waiting Standing still in the plant for a short period e g to let another vehicle pass e Idle Standing still on the parking because no work is given or during bre
49. Dynamic analysis of in plant logistics based on RFID data Simon De Buyser Promotor prof dr ir Hendrik Van Landeghem Begeleider ir Ihsan Arkan Masterproef ingediend tot het behalen van de academische graad van Master in de ingenieurswetenschappen bedrijfskundige systeemtechnieken en operationeel onderzoek Vakgroep Technische Bedrijfsvoering Voorzitter prof dr El Houssaine Aghezzaf l l Faculteit Ingenieurswetenschappen en Architectuur UNIVERSITEIT Academiejaar 2010 2011 GENT Dynamic analysis of in plant logistics based on RFID data Simon De Buyser Promotor prof dr ir Hendrik Van Landeghem Begeleider ir Ihsan Arkan Masterproef ingediend tot het behalen van de academische graad van Master in de ingenieurswetenschappen bedrijfskundige systeemtechnieken en operationeel onderzoek Vakgroep Technische Bedrijfsvoering E Voorzitter prof dr El Houssaine Aghezzaf II II Faculteit Ingenieurswetenschappen en Architectuur UNIVERSITEIT Academiejaar 2010 2011 ENT Acknowledgements During this thesis have received help and guidance from some people who would like to thank would like to thank my promoters prof dr ir Hendrik Van Landeghem and ir Ihsan Arkan for their time and patience Especially ir Ihsan Arkan who in spite of his own very busy schedule still found the time to assist me and guide me in finding the right approach for this thesis and for motivating me to work it out Further wo
50. FID tag right courtesy of Ubisense Both the RFID tags and the readers are products made by Ubisense Also the software used is provided by Ubisense This software calculates the location of the used RFID tags by performing triangulation Further the software is needed to configure the RFID system and to select the appropriate filtering options The filtering options that were used in this test will be discussed later in this chapter The Ubisense software allows the user to visually track the movement of the used RFID tags Not only the current location of the tags but also their previous locations can be displayed in the form of a trail similar to a spaghetti chart 46 id art wd p i RFID reader TM 7 93 Master Location of the tag Figure 31 Trails of multiple tags in the Ubisense software By using software written by De Jaeger Automation the acquired RFID data can be logged to an excel file However this is limited to one RFID tag at a time so if multiple tags are being used at the same time only one of them can be logged to an excel file This excel log file will be used in the test as an input for the dashboard but this will be discussed later in this chapter The dashboard itself and all the calculations for the metrics will be done in excel For this a version of Microsoft Office Excel 2007 is used 47 6 1 2 Layout The layout of the train track is displayed i
51. Flowchart of the status update Once the status is calculated it is added in the RFID data sheet Table 17 RFID data sheet result after step 4 hh mm ss x y m m s of area 7 3 4 5 Step 5 Update of the Error Event Manager Timestamp Coordinates Distance Speed Area Purpose Status 14 12 07 4 401 5 757 0 143 0 143 LaneS Unloading station Based upon the current status of the vehicle the Error Event Manager can be updated Every time the status changes to Error Defect or Error Off track the error and its start time are added to the table The coordinates of the error s location and the zone are also added Table 18 Update of the Error Event Manager when the status changes to Error Error Event Manager Nbr Problem Start time Duration Location Zone X Y 1 Off track 14 11 20 1 459 5 171 Lane 6 When the status changes again the problem is resolved the end time of the error is known and the duration of the error can be entered Table 19 Update of the Error Event Manager when the error problem is resolved Error Event Manager Nbr Problem Start time Duration Location Zone 1 Off track 14 11 20 0 00 01 1 459 5 171 Lane 6 73 7 3 4 6 Step 6 Update of the Order List This step is also based on the current status An order consists of a specific progression in the status The progress of an order being fulfilled is typically Driving gt
52. P GL PS PS PS So a D J SA D 2 D SSS Time Figure 25 Transportation efficiency Situation C Next to the information about the average speed of the vehicle the graph also gives an indication about the route efficiency If the blue line goes higher than the maximum value of the red line on the y axis it means that the vehicle has taken a route that is longer than the shortest path situation C On the other hand it will be impossible for the vehicle to find a route that is shorter than the shortest path so the blue line should always reach at least the same height as the red line To clarify this graph also a gauge meter is added which displays the average speed during the current order and also indicates the ideal average speed Further the travelled path and the proposed shortest path are displayed on the layout of the plant to indicate the route efficiency The load unload efficiency is displayed in a bar chart on which the standard times are also indicated as a green line Underneath this bar chart the load unload variance is also given as a percentage The efficiency of the previous orders is situated in the last timeframe last day The performance of each order is stored in a table and is visualized in a bar chart The bar chart shows the duration of each order and also their time goal To make sure that the graph is easily read the colour of the bars is adjusted to the performance a green bar means that the order was
53. The working of Dijkstra s algorithm is described below Aghezzaf 2009 0 Declarations of the parameters and variables Let G be a graph with known vertices v in V G and known edge weights w w uv weight of the edge connecting vertices u and v e s EV G Source vertex s v Length of the shortest path between vertices s and v div Estimate of 6 s v and d v gt 8 s v riv Predecessor of vertex v e K set of vertices whose shortest path is known 1 Initialization d v Vv vin V G m v Nil Vv vin V G gt d s 0 The distance to the source vertex is zero K 2 Build priority queue Q from V G K The vertices are ordered by distance d v u Min Q e K K U u Add vertex u to the set K 3 Relax u v w for each v in Adj u Perform relaxation for each vertex v adjacent to u if d ul w u v lt d v then d v d u w u v Update the estimation of the distance to vertex v n v u Remember the predecessor of vertex v 4 While Q is not empty repeat steps 2 and 3 A small example has been added to illustrate this algorithm figure 19 A graph with five vertices is given with the most left vertex s as the source In the first step second graph the source vertex is added to K indicated by colouring the vertex grey and the distances d v and predecessors r v for each node have been added Then the vertex with the shortest distance is added to the set K and the entire process i
54. Zuipeorm SunreM E SulAugm saznurw OT 3se u ouedna22Q sn3e3s os saznurw OT ase ut UOHEZIIIN peojun uonepodsuesy peo uonejs8uipeoj 0 aAug 143p10 1ue4un u0I3820 Ue AIN sn3e3s Juang lppdn 1501 IV 11 3 Appendix D Performance report Internal Logistics Report Date and time of printout 27 05 2011 21 01 Starttime of the RFID data 14 10 57 Endtime of the RFID data 14 54 56 Efficiency and Utilization Average Efficiency of the vehicle 52 Total Utilization of the vehicle Driving Loading Unloading Overall Occupancy Occupancy per 10 minutes Lane 1 E lane 2 Lane 3 Lane 4 Lane 5 Lane 6 Lane 7 Lane 8 c Lane 9 Lane 10 Lane 11 Lane 12 Lane 13 Lane 14 Lane 15 Lane 16 Lane 17 Lane 18 Coordinates Mm Errors Lost time Occured Errors Cause of errors Total lost time 18 1 min 17 85 min 0 25 min Error Event Manager Problem Starttime Duration Location Zone x Y 1 Off track 14 11 20 00 5 171 Lane 6 2 Defect 14 14 28 00 E 4 998 Lane 4 3 Off track 14 14 47 00 1 3 977 Lane 4 4 Off track 14 15 08 00 3 4 050 Lane 2 5 Defect 14 17 31 00 1 4 547 Lane 1 6 Defect 14 19 14 00 5 730 Lane 5 7 Defect 14 19 16 00 5 5 652 Lane 5 8 Off track 14 20 03 00 4 702 Lane 7 9 Off track 14 20 12 00 j 3 772 Lane 7 10 Off track 14 30 03 00 a 5 362 Lane 5
55. ained by the following example Figure 14 Example of a zone To determine if a point is located within the zone presented above four statements will be checked If each of the statements is correct the point is within the zone If however one or more statements are incorrect the point is not in the zone In this example the equations of the lines are 1 2 3 4 y 3x 2 2 8 G y 5x 19 y 1 These equations lead to the following statements which must be fulfilled for a point with coordinates x1 y to lie within the zone 2 3 4 2 X gt utt y 21 This can now be extended to multiple zones in which each of the statements will be checked If the data point is not within any of the defined zones the point will be considered to be Off track This will further be explained later 22 Each zone is identified by its descriptive name This name is chosen so that it allows a quick interpretation of the current location e g aisle 1 hall 3 inbound etc Next to the descriptive names the zones will also be given functional names These names say something about the purpose of each zones and will later be used as an input for the status update of the vehicle The used functional zones here are e Loading station Locations where the loading of the logistics vehicle will occur e Unloading station Locations where the unloading of the logistics vehicle will occur e
56. aks e Defect Error A problem occurs causing the vehicle to stand still e Off Track A problem occurs causing the vehicle to go off track The status of the vehicle is determined based upon the following parameters e Speed The speed of the vehicle can be calculated from the RFID data e Direction This could be calculated if two RFID tags were attached to the vehicle one in the front and the other in the back of the vehicle e Duration of standstill How long has the vehicle been standing still e Area of location This is the zone in which the vehicle is operating e Order progress The current order that is being done and the progress within that order 23 The first three parameters are easily found by looking at the RFID data The last two parameters are a bit more complicated The area of location and more important the purpose of that location can be calculated by using the functions described under 4 1 1 Location of the vehicle The order progress will be determined by looking at the order list Next to the fact that the status update depends on the order progress the order progress itself is also derived from the status update and will also be updated accordingly but this will be explained later On the following figure a schematic representation of the interaction between the status update and the needed data RFID data predefined data and order list is given RFID data Timestamp Speed Dir
57. auruedap sonsiSo 3uaunjedap ueu luleuu au ui jue ursAo1dui Ji aas eoueusyureur jo ilenb sajiyan pasn aui Jo Aujenb mous 150 Jo suosea1 moys pue A uN e1o1 s u All p Buoum o anp oun 1501 own 30 2e gO Suro3 o3 anp w 1501 yoajap e jo uoneunp 3es ny uun 301 323Jag 01 anp euim 1501 16301 BULL 350 01 Ayiaeroa Suppld Auliqerja anoy areda 0 aw ue W a8equaaiad uun1sO1 waysAs sonsi80 ay3 JO AyIqel a4 ayy aseasou e sy 3ueuuloju d o peo Noy up Ajuaioyjaul st J3IN A ay J J8A0 papaau ase s p y a 40 Jl 40 ap aie saj21yan Auew 001 Ji ponoaduu aq ues uoneziinn aueuw mous ease Buipeojun Jo Buipeoy aui 3e swargoad 33430 spoyyawy Buipeojun qua oyjau Suo 9430 spoyjaw 8ulpeo 1u l3UJ u Suo 322180 aanos Buinup au jo Aujewndo ayy azijensip Jouapyja paads au Aeldsi paads aderane Ieapi 1e ssauZoud paysadxa ayy 0 paueduioo sapso ypee jo ssasBosd ayy azijensin wayshs Suy USAID Jo 4qu e301 papeo lIuA u Aup Suuy JO JON eun eIOL SUH YOM eolun aun Buipeo oun Buipeojun jenpy ew Buipeojun puepuers oui Buipeon jenpy aus upeo puepuers eno eme ayy jo iua ano lqissod 1s uuous aui jo y13u 7 Buug sni amp 3s e luw paads aBesony eun uonexodsuea jemoy awy uoneuodsuea pep edx3 JapJo 9 10 Aouayyyga jjesano ayr JO sjoen daay sad awn p 12 dx Japso Jad aun 270 Jeny winwiuiw e 0 350 34 daay pue A3u l31g
58. be made that contains the EPC the location and the In time and Out time Further reduction can be obtained by grouping information of items that move together through the plant Instead of having a single EPC in a tuple a set of EPCs could be stored EPC list location time By logically reducing the number of needed records all the important information remains intact and the useless data is removed Gonzalez et al 2006 In this thesis the general guidelines will be followed however because the thesis focuses on the logistics vehicles rather than the flow of products through the supply chain the research of Gonzalez et al cannot be applied directly If in an expansion the products would also be tagged and followed then this filtering technique would become more interesting 2 2 3 Dealing with the unreliability of RFID data As mentioned before one of the challenges RFID systems face is the reliability of the system Because of the fact that RFID data is inherently unreliable widespread adaptation of the RFID technology is tackled However most RFID middleware systems take this into account and employ a smoothing filter to correct these errors as much as possible This smoothing filter can be defined as a sliding window aggregate that interpolates for lost readings Jeffery 2006 The choice of the window size is however not a trivial task If the window it chosen too small the systems reacts very heavily on err
59. c and real time data retrieval which increases the overall accuracy by eliminating human errors In their research they propose the use of an RFID case based logistics resource management system R LRMS This system will improve the efficiency and effectiveness of order picking operations in a warehouse by using the real time aspect of RFID technology and case based reasoning CBR as a decision support system The R LRMS can be used to select the most appropriate material handling equipment for a certain order by using the CBR and looking at previous cases and can then determine the shortest path to fulfil this order The logistics vehicle driver will then be directed to the correct picking route and will also be alerted when a wrong route is taken or when his vehicle is standing still for too long In their research they also presented a case study in the company GSL Here all forklifts and all Stock Keeping Units SKUs were equipped with passive RFID tags By placing multiple RFID readers and antennae all over the plant the location of each forklift and SKU could be determined Then for each order picking operation the closest forklift was automatically selected and the most efficient route to pick up the needed SKUs was calculated This information was then given to the forklift driver so he could follow this route From this case study they concluded that the accuracy of the inventory data had increased significantly now the exact inve
60. can get lost within the supply chain for example by misplacement spoilage or by theft The capabilities of RFID technology can reduce this loss significantly The goods can be tracked through the company countering misplacement and theft and warnings can be given when spoilable goods reach their expiration date e Increased data accuracy By reducing the human errors the inventory data will be much more accurate when using RFID data Not just the inventory data but also the shipment data can be more accurate by using RFID This in turn can improve the demand forecast and the production planning e More efficient material handling RFID can effectively decrease the overall handling time needed Its ability to identify multiple products at once decreases the inventory counting time and receiving time Further RFID data can be used to automatically determine a loading or unloading point for a logistics vehicle and even choose the most efficient route 11 e Timely exception management By using RFID technology unusual situations can be detected faster and responsive action can be taken before the problem escalates This aspect can also be automated by generating warnings or alerts when an error or problem is detected This is based upon the real time aspect of RFID technology 2 3 2 Applications Chow et al researched the integration of RFID technology into existing Warehouse Management Systems WMSs This would allow automati
61. cation of the vehicle is always known due to the RFID tag attached to it The RFID coordinates are calculated by the used RFID software or Real Time Location System and are then cleaned and smoothed before they are used as input for the dashboard The X and Y coordinates can directly be plotted in a scatter plot in excel displaying the current detailed location of the logistics vehicle in the plant layout For the area of location zones are defined on the plant layout These zones can either be defined in the RFID software itself or in excel If the zones are defined in excel the following method will be used to construct the zones and to determine in which zone the vehicle is located First of all some conditions are specified to keep the calculations simple Here the format of a zone is constrained as described below e Every zone is defined by four coordinates e These four corners are connected to each other by straight lines B lt Figure 13 Format of the zones the 3th shape is not convex thus not allowed e Every zone has to be convex V 21 When the coordinates of the corners are given the equations of the bordering lines can be calculated The basic equation of a line through two points is given below Y2 7 y y y x 2 1 After the equation of the each line is calculated one needs to determine on which side of a line a point needs to be placed to be in the defined zone This calculation is expl
62. cific products Figure 2 Standard forklift Courtesy of Mitsubishi Figure 3 Crane Attachment Courtesy of handlinggear com Next to the standard forklifts also many specialized forklifts exist to handle heavy or very big products As space is always a limiting factor some warehouses use very narrow aisles To handle products in these warehouses special forklift variants are also available for operating in narrow aisles Figure 4 Heavy duty forklift Courtesy of Toyota Figure 5 Narrow aisle forklift Courtesy of AisleMaster 2 1 1 2 Tugger trains Tugger trains are operator driven vehicles that tow carts through the company As opposed to the forklifts they are not able to pick products from racks They are however more suited for transporting prepared kits from the supermarket to the border of line Mostly the tugger trains will be used as mizusumashi as described above Figure 6 Tugger train courtesy of K Tec 2 1 1 3 AGVs Automated Guided Vehicles or AGVs can also be used to perform logistics tasks As the name says these vehicles are not driven by operators but are automatically routed through the manufacturing plant to pick up or drop off products AGVs can be subdivided according to the way they navigate through the plant The guidance system of an AGV can make use of some different technologies In most cases navigation is made possible by using an optical magnetic or wireless radio guidance sys
63. cs Development Matrix and now need to be implemented This implementation of the measurement system can be done by designing a clear dashboard portrayal tool which presents the chosen metrics The design of this dashboard is quite extensive and will be discussed in chapter 5 Metric presentation 3 1 6 Step 6 Utilize the MS The last step is of course the utilization of the measurement system The measurement system will now be used to analyse the current performance and take actions based upon this to further improve the working of the in plant logistics The measurement system will be used in a test setup which is discussed later in chapter 6 Test setup HoWest 19 3 2 Hierarchy of the designed metrics The designed metrics can be calculated and displayed on different aggregation levels Further a certain hierarchy can be formed amongst the metrics This hierarchy needs to be kept in mind when designing the dashboard and is displayed in figure 12 At the left side of the figure the four KPAs are presented These KPAs are measured by the End Result Metrics which are placed directly next to the KPAs The End Result Metrics can then also be linked to Driver Metrics which are on the right side of the figure Area of location Detailed location Change in utilization Status division per hour Change in lost time Visibility Status division during the last Current status hour
64. d This can be configured by the Quality of Service QoS parameter as can be seen in figure 33 This parameter is split into two rates a slower QoS and a faster QoS which are separated by a speed threshold If an RFID tag is moving with a speed higher than the defined threshold the faster QoS will be used otherwise the slower QoS will be used The reason for this separation between a faster and a slower QoS is that a faster moving object requires more data points to correctly describe it Tag Range Parameters From Id 020 000 000 000 Filter Default no filtering v To Id 020 000 000 255 Slower Qo5 one update every 32 time slots h Faster QoS ane update every 16 time slots v Threshold 4 mis Cancel Figure 33 Tag range parameters Selection of the QoS For this test both rates have been given the same value One update every 8 timeslots One timeslot is equal to 0 027 seconds so the RFID tag location will be measured every 0 216 seconds With this setting almost five points are generated per second Ubisense 2009 Though only one point per second is needed for the correct functioning of the metrics this value for the QoS setting was chosen because of the imprecise scaling frequency When the speed is calculated between two consecutive data points very high fluctuation can be seen because of this imprecise scaling frequency To make the resulted RFID data more accurate and r
65. d screen has been cropped to fit on this page A larger and more clear figure of this output is added in the appendices appendix C Test dashboard output 54 TRAIN 1 Update Stop zum Last Update 14 54 56 16 13 57 Dashboard Current Status e Status Occupancy per 10 minutes Current Location Print Report Spaghettichart last 10 minutes Error location Current Order Lane1 Progress Drive to loadingstation Bene eee Load x Didie Sees Transportation Lanes B Waiting Lanes ie r Liane7 B Unloading LX IH ili i i V Q k amp Lane 9 50 Bl Loading Lane 10 m mE Lane 11 B Driving ae Lane 12 Lane 13 1 L 4 i b Lane 14 Lane15 Lane16 Lane 17 14354 1425 14 15 14 05 Lane18 Coordinates Time CurrentLocation 7 W Errors Change in Utilization Error Event Manager Change in Lost time Nbr Problem Starttime Duration Location Zone Cause Causes of lost time in last 10 minutes 14 11 20 14 14 28 14 14 47 14 15 08 14 17 31 14 19 14 14 19 16 14 20 03 14 20 12 14 30 03 14 30 17 14 30 18 14 31 14 Woriving waiting 1 BlLoading Unloading Error Offtrack Didie WBrror Defect Error Total lost time 402 sec in last 10 minutes BREE D S n m m Figure 35 Cropped test dashboard output The first thing that can be noticed on the dashboard is that a lot of lost time occurred during the
66. d variance Optimal route Distance travelled at ideal speed miei ma EE Previous orders Previous performance Order Efficiency Actual time vs Standard time Efficiency Avg Speed Route Loading Unloading Load Unload time efficiency efficiency variance B On time m Late m Goal 15362 15363 15364 15365 Order Figure 23 Screen 3 Efficiency The efficiency screen is divided into two parts e Efficiency of the current order e Efficiency of the previous orders 41 The efficiency of the current order is situated in the first two timeframes current time and last hour However the second timeframe is not strictly followed as the performance is not measured over the last hour but rather over the duration of the current order immediate past Besides the efficiency this screen also displays some information about the current order progress This is displayed in the upper left corner As described under 4 2 1 Overall efficiency an order can be divided into four steps Here the progress in the four steps is indicated by displaying the percentage of completion or a green checkmark when a step is completed 100 The progress of the entire order is also displayed as a meter filling up This summarizes the progress that is depicted above The overall efficiency of an order can be calculated as the weighted average of the transportation efficiency and the load unload efficienc
67. d with a signal that identified the aircraft as friendly Over the years extensive research was done in the field of RFID technology but there were not many commercial applications until the late nineties because of the high cost of RFID systems In 1999 the Auto ID Center at the Massachusetts Institute of Technology MIT was founded Here research was done to make low cost RFID tags and place these tags on products to track them through the supply chain The tags could be produced at a lower cost because they only needed to store a serial number and thus the needed memory capacity was very small The data associated with the serial number name of the product lot number etc would now not be stored in the tag but rather in an internet database This research changed the meaning of RFID technology in the supply chain Because of the cost that was now significantly reduced the RFID technology became much more interesting Three major organisations that can be considered as the pioneers in the large scale adoption of RFID technology are Wal Mart Tesco and the US Department of Defense Now an increasing number of companies use RFID technology to track goods in their supply chain Want 2006 Roberti 2007 2 2 1 3 Working An RFID system consists of RFID tags that are fixed to the object that needs to be identified or tracked and RFID readers that read the data from the tags The readers communicate with the tags through inductive coupling
68. dynamic performance measurement system is described followed by the detailed working of the designed metrics and their presentation in a dashboard Then the designed metrics are tested by implementing a dashboard on a test setup at HoWest Finally the results of this test are discussed and some remarks are given about the expansion to a real life situation 2 Literature study First a brief introduction to in plant logistics is given Here the purpose of in plant logistics and also the used logistics vehicles are described In the second part of the literature study the working of RFID technology is briefly explained and it is described how RFID data can be cleaned and filtered so it becomes useful for real life applications In the last part of the literature study a summary is given of current researches that use RFID data to provide real time monitoring of the in plant logistics vehicles and to improve their performance 2 1 In plant logistics In plant logistics is the part of logistics that takes place within the walls of a company Here the transport and storage of products in the warehouse the supermarket and at the border of line is the subject of interest Though transportation and storage are considered to be non value adding processes they are still needed for the correct functioning of the plant necessary non value adding processes Not only is the in plant logistics necessary it is in fact an important factor in the sup
69. e In a normal manufacturing plant multiple logistics vehicles will be driving around to fulfil orders Special attention should be paid to make sure that the vehicles do not interfere with each other s work too much 8 2 Changes needed to the test dashboard The test dashboard that was designed in this thesis can also be used to analyse data from real life situations As mentioned above some differences exist between the test setup and a typical real life situation Here a small summary is given of the needed changes to the test dashboard for the use on a real life case e Timeframes The timeframes need to be changed to more meaningful durations current time last hour last day e Inputs The inputs need to be adjusted to the situation The zones and the network data need to be adapted to the layout of the plant The order list will of course change depending on the tasks that need to be fulfilled e Parameters The defined parameters are dependent of the user s preferences For example if the user wants to detect defects very quickly he should select a low Error Threshold if the vehicle however needs to make frequent stops and the duration can be long the Error Threshold should be higher to avoid an overly sensitive reaction e Status update The way the status is updated and which statuses exist should be changed according to the used vehicle If the dashboard is used to monitor AGVs or Tugger trains the status
70. e As mentioned before under 4 2 1 1 Transportation efficiency this ideal average speed is based upon the speed capacity of the vehicle and on the safety rules within the plant the vehicle should drive at a safe speed to avoid accidents 65 7 3 3 Inputs of the test dashboard When the dashboard is used it dynamically extracts the cleaned RFID data from another excel file the RFID data file as described under 6 4 2 Adjustable parameters Apart from the cleaned RFID data three more inputs are needed to assure that the situation is interpreted correctly These extra inputs are the order list the used zones on the plant layout and the information about the available routes and intersections The RFID data is loaded in the file while the dashboard is being used As opposed to the RFID data the other three inputs need to be loaded into the dashboard before it can be used All four inputs are needed for the calculations of the metrics and have to be loaded correctly into the dashboard file Here the correct format of these inputs is described as well as the way they are implemented into the dashboard file 7 3 3 1 Cleaned RFID data The cleaned RFID data consists of a timestamp X and Y coordinates and the corrected speed This data should be provided in an input file in the following format Table 10 Format of the Cleaned RFID data Time X Y Speed 14 54 56 4 038 3 364 0 590 14 54 55 3 448 3 359 0 166 14 54
71. e results are placed in this sheet e ErrorEventManager In this sheet all the occurred errors are stored with all the important information about each error e ReportTemplate This sheet contains the template that is used to generate the report 63 e Parameters Various parameters in the dashboard can be adjusted by the user These parameters are stored in this sheet e RFID data The cleaned RFID data is loaded into this sheet and the details of each data point are added when they are calculated area of location purpose of area status and current order e OrderList interesting information about the executed orders Input sheets e Zones The used order list has to be loaded into this sheet After the performance of each order is calculated the results are also added here The OrderList sheet thus offers all the The specific zones of the plant are declared in this sheet Also the calculations for the area of location are done here e ShortestPathCalculations The necessary data for the shortest path calculation is stored in this sheet It contains all the required information about the plant s layout coordinates of the intersections and the connections between them 7 3 2 Adjustable parameters Some parameters of the dashboard are adjustable and can be entered by the user These parameters are presented in the Parameters sheet as explained before Be
72. e update rate is increased to one update every second the dashboard update fails because the RFID data cannot be provided fast enough Due to the design of the update function the dashboard gives a warning that no new data is available and automatically stops updating 59 Though it is possible to update the dashboard every two seconds this is not recommended During the test it was noticed that excel was very loaded and seemed to be a bit unstable A larger update rate is thus recommended Further it should also be noted that in this test it was assumed that the raw RFID data could be cleaned immediately However if this cleaning requires a certain amount of time this should also be kept in mind when determining the best possible update rate 60 7 Design of the test dashboard The test that was described in the previous chapter is done with an own developed test dashboard In this chapter the design and the working of this dashboard is discussed First the used metrics and the needed adaptations for the test are described Then the graphic design and the working of the dashboard are explained Finally the post analysis options are presented 7 1 Used metrics and adaptations The test dashboard was already used in the previous chapter From the used figures it could be seen that not all the designed metrics have been integrated in this dashboard Here the used metrics are described Some metrics have been adjusted
73. ection Duration of standstill X Y Coordinates Area of Current Progress in Location status Order Predefined data Defined Zones Purpose of area Order list Current Update Order Current Order Figure 15 Interaction between the status update and the needed information The exact calculation of the current status also depends on the used vehicle and some other defined parameters For example later in this thesis a test dashboard will be designed to use on a setup consisting of a toy train on a track In this test setup a loading or unloading action can be seen as the vehicle will be standing still at a loading or unloading station for a specific amount of time Similar behaviour will also be seen when an AGV or a tugger train is analysed However as opposed to the toy train a forklift in a plant will not simply be standing still when it needs to load or unload an item It will instead perform a specific manoeuvre In the appendices a flowchart of the status update for the test setup can be found appendix B Flowchart of the status update 24 Once the status of the vehicle is determined this metric can be displayed on multiple aggregation levels e The current status can be displayed so errors or defects are quickly discovered e The status division in the last hour can be displayed to indicate the occupancy and pe
74. ed the utilization dropped and the reason is clearly the increasing amount of lost time due to defects In a real life situation this would be an indication that there are too many problems with the logistics vehicle and that an intervention will be needed to increase the utilization again 55 Further the dashboard also displays the current order Here the current order is Order 18 However only 17 orders were given in the order list This means that all 17 orders have been correctly fulfilled 6 6 2 Performance report After the test a performance report can be printed out This report summarizes all the important events during the test and also presents the total utilization and efficiency of the vehicle The performance report is quite large as is thus added as an appendix appendix D Performance report The performance report verifies the results in the dashboard and also displays some extra information From the dashboard it could be seen that the utilization of the vehicle in the last 10 minutes was only 33 However over the entire duration of the test the utilization of the train was 48 according to the performance report which is logical as the bar chart already indicated that the utilization was higher in the beginning of the test The utilization over the entire test is very low and this is mainly caused by lost time due to errors 41 from which 99 is caused by defects 17 85 minutes of the total time
75. educe the high fluctuations in the calculated speed a higher update rate QoS has be chosen so the median value can be calculated over each second Arkan 2010 This high update rate can be more straining for the system and excel but will give a more accurate location and speed of the RFID tag Once the parameters are decided the data can be logged to an excel file with the software written by De Jaeger Automation The format of the logged data is displayed in table 3 49 Table 3 Format of the logged raw RFID data Date x y z Cell TagID Tagname BatteryLevel 14 11 00 3 682304 4 681683 0 513881 020 000 118 099 Treini 0 14 11 00 3 694071 4 677925 0 507248 020 000 118 099 _ Trein1 0 14 11 00 3 943779 4 620492 0 620687 020 000 118 099 _ Trein1 0 14 11 00 4 072446 4 606907 0 944648 020 000 118 099 _ Trein1 0 14 11 00 4 073717 4 610288 0 950814 020 000 118 099 Treinl 0 14 11 01 4 080714 461588 0 95363 020 000 118 099 Trein1 0 14 11 01 4 211967 4 616945 0 724454 020 000 118 099 Treinl 0 14 11 01 4 219471 4 617785 0 709438 020 000 118 099 Trein1 0 14 11 01 4 237974 4 619946 0 688029 020 000 118 099 Treinl 0 14 11 02 4 266108 4 617431 0 667755 020 000 118 099 Trein1 0 14 11 02 4 302099 4 610484 0 65017 020 000 118 099 _ Trein1 0 14 11 02 4 344167 4 595333 0 63886 020 000 118 099 Trein1 0 The acquired data as presented here is still raw and needs to be cleaned before it can be used as input for the dashboard The clea
76. efined duration that is added in the order list This standard time will be calculated according to the handlings that need to done to fulfil the loading 31 or unloading at the standard pace However the exact calculation of this standard time is not part of the thesis and will not be described further It will be assumed that this standard time is given It should be noted that this metric is constrained in value Because of the way the status update is calculated a loading or unloading action can only be considered to be valid when the duration of the action is at least a given percentage of the standard time Under this lower bound the action will be considered as waiting instead of loading unloading Next to this lower bound also an upper bound is determined When this upper bound is reached the status changes to error thus limiting the loading or unloading in duration Figure 20 illustrates the working of this status update Standard time for loading 10 seconds LowerThreshold 50 of the standard time 5 seconds UpperThreshold 180 of the standard time 18 seconds Duration 4 lt LowerThreshold Timestamp Coordinates Distance Speed Area Purpose Status Order x y of area 14 37 54 3 029 0 148 0 016 0 016Lane18 Loading station 35 14 37 53 3 013 0 148 0 016 0 016 Lane 18 Loading station 15 14 37 52 3 027 0 138 0 024 0 024 Lane 18 Loading station 15 14 37 51 3 051 0 143 0 027 0 027 Lane18 Loading stat
77. eit voor een bepaalde order en bepaalt de kortste route om dit order te vervullen 1 Het tweede onderzoek van Ludwig en Goomas stelt een systeem voor waarin de prestatie van vorklift chauffeurs gemeten wordt en direct als feedback wordt gegeven zodat ze hun prestatie rap kunnen bijsturen Hiervoor wordt de werkelijke duur van een order vergeleken met de standard time De standard time wordt hier berekend op basis van de kortste route die kan worden genomen aan de ideale snelheid en de berekende tijden voor de meest effici nte laad en ontlaadprocedures 2 In het laatste onderzoek stellen Kootbally et al het PRIDE algoritme voor om AGVs te navigeren doorheen een dynamische werkomgeving Door bestaande shortest path algoritmes te combineren met een actieve collision avoidance botsing vermijding kan de betrouwbaarheid en ook de totale prestatie van de interne logistiek verbeterd worden 3 III ONTWIKKELING VAN DE METRICS Door gebruik te maken van het Performance Measurement Development System PMDS kan een performance measurement system opgesteld worden Dit proces wordt gebruikt om een set van metrics te ontwerpen voor de dynamische analyse van de interne logistieke voertuigen Uiteindelijk worden via deze methode 19 metrics ontworpen die logisch passen onder 4 Key Performance Areas KPAs Deze KPAs en de bijhorende metrics zijn voorgesteld in de onderstaande tabel Tabel 1 Ontworpen metrics
78. er oir Ee verbe Pres Qv aede er a e vedan EXE NES UN 41 5 2 6 Screen Reliability rhet eee eim eet ter nn ertet 44 5 2 7 Dashboard navigation iret ee A dan e TERR ee RR mannen 45 Test Setup HOWEST Su aisle Puto eoe ai el eo to kac ue te a INE ve Ce eye dee Q eels 46 6 1 Details of the test Setup eret i eter P Iq P ner ERE 46 6 1 1 Equipment used ns senen etten nd tetti diete ti arte ves 46 6 2 Selection of the parameters and cleaning of the data 49 6 3 Defined zones eee pe e de tree aun teres t vae Eee raa pero i e rl ees 52 6 4 Used order list oti re ete RET REPOS dirae Corne i de 53 6 5 Progression of tlie test ote ree RR DR Y hua rr REP RYE d 54 6 6 Results Accuracy of the analysis 54 6 6 1 D shboard Outpul acte aten eet tr tle oret er ae RE ve nete d d e ds 54 6 6 2 Performance T6epolt tereti ive entree 56 6 7 Results Real time performance of the dashboard 58 7 Design of the test dashboard 61 7 1 Used metrics and adaptations 61 L2 Graphic A amp SIEN ween kenner inden daneen ak des eea ae aen kal q Na ave delen
79. erarchy of the designed metrics esses 0 nnne nnn nnn sa 20 Format of the zones the 3th shape is not convex thus not allowed 21 Example of a zone eo ote reet ate eere t iste es 22 Interaction between the status update and the needed information 24 Efficiency loss cii e e a E B II ea eaaa 26 Average speed vs ideal average speed 28 Plant layout left and the accompanying network right 29 Example of the use of Dijkstra s algorithm 31 Working of the status update in case of loading or unloading 32 Screen 1 Summary M eW u u umn sana lan aan a aulia 38 Screen 28 Visibility etse euet u i a ee 40 Screen 3 XETICISHI CV Sirti ROO RITE 41 Transportation efficiency Situation A left and B right 42 Transportation efficiency Situation C 43 Screem 4 Reliability itt tt ett ee eei etes ee vn ER oes 44 Connection between the four screens of the dashboard 45 Button Return to Summary view
80. erent steps of the order the transportation efficiency for step 1 and 3 and the loading and unloading efficiency for step 2 and 4 These metrics are explained further in this chapter In figure 16 underneath the actual total time is compared to the expected total time The difference between these times is the efficiency loss indicated in red Actual total time M gt Actual transportation time to Actual loading Actual transportation time to Actual unloading loading point time unloading point time Expected total time lt pe EL Expected transportation time Standard Expected transportation time Standard to loading point loading time to unloading point unloading time Expected transportation time Distance of the shortest path m Ideal average speed m s Figure 16 Efficiency loss 26 4 2 11 Transportation efficiency The efficiency of the transportation of the logistics vehicle is measured here This metric compares the actual time it took to get from one point to another with the expected transportation time The expected transportation time is here calculated as the time needed to reach the endpoint while driving at the ideal average speed and choosing the shortest possible route As mentioned before an order usually consists of four steps As there are two transportation steps transportation from current location to loading point and transportation from loading point to unloading point this metric will c
81. f the network data Node X Y Name Nbr Of Edges 1stconnection 2nd connection 3th connection 0 1 8 5 7 2 1 15 1 3 1 5 7 LaneS 2 0 2 2 45 5 7 2 1 3 3 5 1 5 3 Lane4 2 2 4 4 6 47 3 3 5 22 Some explanation is given to clarify this format Each used node in the network is placed in a different row of the table and is uniquely identified by the number in the first column The coordinates of every node are added in the table and a name can also be given if the node represents a zone of interest In the table above node 1 represents Lane 5 This node is situated in the middle of Lane 5 and can be used as a loading or unloading point Further the number of connected edges is given and the nodes connected to those edges For example node 1 has two connecting edges that lead to node 0 and node 2 68 Figure 42 Node 1 and its connected nodes 0 and 2 on the train track layout In this test the network has been simplified a bit Not every intersection and zone was translated into a node only the needed nodes were defined to show the capacities of the shortest path calculation In the figure underneath the layout of the train track and the locations of every used node is shown Figure 43 Locations of the used nodes 69 7 3 4 Working of the dashboard update Once the order list the used zones and the network data are loaded into the dashboard it can be used for dynamic analysis of the train s pe
82. fe cases The test dashboard can be applied with only minor adjustments 79 when the performance of AGVs or tugger trains is measured If the test dashboard would however be used to measure the performance of forklifts further adjustments would be needed to better fit the behaviour of a forklift the status update needs to be reprogrammed To increase the value of these designed metrics and the dashboard an expansion is proposed here By providing direct feedback of the measured performance to the driver of the logistics vehicle the performance can be actively improved The locations of the load and unload points could be communicated and the shortest path could be visually proposed to the driver This would increase the overall efficiency of the in plant logistics By warning the driver about errors e g the wrong product has been loaded or the wrong route is taken these errors can be resolved more quickly and the reliability will also increase Overall it can be concluded that the proposed analysis methods are useful to visualize the performance of the logistics vehicles A correct application on real life cases could provide some interesting information about the current situation and could help improve the overall performance of the in plant logistics 80 10 References Aghezzaf E H 2009 Introduction to Operations Research Industrial Management Faculteit Ingenieurswetenschappen UGent Chapter Network problems
83. ful by asking for the purpose of the metric how they can be presented to the users how the required data can be collected etc The details that need to be entered in the MDM are listed below e Metric Specification Metric name Operational definition and or formula Purpose of the Metric Metric owner e Portrayal Design Portrayal frequency Type of data Portrayal tool e Data Collection Plan Tracking tool Data available Data collection responsibility Data collection tool s Data collection frequency e Utilization Implementation date Metric goal Some of these details are not really important for the right development of the metrics in this thesis so will not entered into the MDM Metric owner implementation date etc For the four KPAs 19 metrics have been designed These metrics were then entered into the MDM and their details were added Because of its large size the Metrics Development Matrix cannot be placed in this chapter it is added in the appendices Appendix A Completed Metrics Development Matrix However an overview of the 19 designed metrics is presented in table 1 on the next page 18 Table 1 Developed metrics Nbr Metric Operational Definition or Formula KPA Visibility 1 Location of the vehicle Coordinates Area of location zone 2 Status of the vehicle The status can be Driving Waiting Un Loading Idle or Error 3 Order list progress Display the c
84. g for the problem 5 2 Dashboard design Here the performance of in plant logistics vehicles will be the subject of the dashboard The user of the dashboard is likely to be a plant manager who wants to see how the logistics is currently performing and how it can be improved The dashboard should thus be designed to best fit the needs and interests of the plant manager 5 2 1 Used timeframes The data that is displayed in the dashboard is aggregated according to three different timeframes e Current time e Last Hour e Last day The Current Time frame allow a quick interpretation of the current situation Errors or problems can be detected timely the location of the logistic vehicles is always known and the work progress is visualized Whereas the first timeframe only takes a snapshot of the current situation the second timeframe shows some more information about the immediate past This can be useful to examine the occupancy of the vehicle how high is the utilization and what tasks has the vehicle performed in the last hour and to examine its travelled path The third timeframe offers a summary of the performance during the entire day In this timeframe also the trends for the occupancy of the vehicle and the occurrence of defects are given These trends depict the performance of the internal logistics in timeslots of one hour From it the user can place the data given in the other timeframes in a context e g is the bad perf
85. gegaan worden hoe het dashboard reageert Bij elke geteste update rate wordt er gekeken hoe lang de update duurt de update heeft een zekere tijd nodig om alle noodzakelijke berekeningen en functies uit te voeren en of het dashboard nog correct de data weergeeft Uit de test kan geconcludeerd worden dat een maximale update rate van n update elke twee seconden nog steeds een goed resultaat weergeeft Deze update rate is zeker hoog genoeg om de prestaties van de interne logistieke voertuigen correct weer te geven VI CONCLUSIE De voorgestelde metrics en hun presentatie in het dashboard kunnen nuttige informatie verstrekken over de geleverde prestaties van de logistieke voertuigen Uit de tests blijkt dat de metrics weldegelijk een correct beeld geven van de realiteit en dat real time analyse mogelijk is Voor verder onderzoek kunnen deze metrics en het dashboard worden toegepast op re le bedrijfssituaties om de prestatie van echte logistieke voertuigen zoals vorkliften tugger trains en AGVs te meten REFERENTIES 1 Chow H Choy K L Lee W B and Lau K C 2005 Design of a RFID case based resource management system for warehouse operations Expert Systems with Applications 30 2006 p 561 576 2 Ludwig T and Goomas D 2009 Real time performance monitoring goal setting and feedback for forklift drivers in a distribution centre Journal of Occupational and Organizational Psychology 82 2009 p 391
86. goal setting and feedback for forklift drivers in a distribution centre Journal of Occupational and Organizational Psychology 82 2009 p 391 403 81 Palmer M 2004 Seven Principles of Effective RFID Data Management www objectstore com docs articles 7 principles rfid mgmnt pdf Polniak S 2007 The RFID case study book RFID application stories from around the globe Abhisam Software Poon T C Choy K L Chow H Lau H Chan F and Ho K C 2009 A RFID case based logistics resource management system for managing order picking operations in warehouses Expert Systems with Applications 36 2009 p 8277 8301 Pureshare Metrics dashboard design designing effective metrics management dashboards http www pureshare com resources resource_files PureShare_Dashboard_Design pdf Roberti M 2007 The history of RFID technology RF D Journal www rfidjournal com article print 1338 Tajima M 2007 Strategic value of RFID in supply chain management Journal of purchasing amp Supply management 13 2007 p 261 273 Ubisense RTLS training 2009 Location engine config user manual Van Goubergen D 2010 Design of Manufacturing and Service Operations Industrial Management Faculteit Ingenieurswetenschappen UGent p 59 85 p 113 132 Van Landeghem H 2008 Advanced Methods in Production and Logistics Industrial Management Faculteit Ingenieurswetenschappen UGent Want R 2006 A
87. hnology would be more regulated to protect the customers rights the general acceptance would be higher Want 2006 2 2 2 Dealing with massive RFID data sets RFID systems are capable to generate huge amounts of data a modest RFID system can generate gigabytes of data per day Though it can be useful to have as much data as possible in most cases not all the acquired data is needed Further this flood of data is very stressing for the used technology infrastructure and can even exceed its capacity Data filtering will be needed to extract the useful information and remove the useless or redundant data To effectively deal with these huge data streams some general guidelines can be followed Palmer 2004 e To reduce the stress on the technology infrastructure as much as possible the incoming RFID data should be digested close to the source All the unnecessary data should be removed and only the meaningful information should be sent to the central IT systems e The incoming information needs to be turned into meaningful events By analysing the incoming data specific patterns can be detected from which meaningful events can be deducted e g a vehicle drives to a loading point stops for an amount of time and then drives away again gt the vehicle is performing a loading action By changing the gathered RFID data in events the data stream can be reduced e The RFID data needs to be aged correctly The gathered data doesn t need
88. ination quicker but this could also cause safety issues The speed of the vehicle should thus be moderated in order not to cause any accidents 27 AVERAGE SPEED OF THE VEHICLE 2 Km h 8 Km h The average speed is lower than the ideal The average speed is higher than the ideal average speed 5 km h average speed 5 km h The transportation efficiency will be low The transportation efficiency will be better There can be safety issues Figure 17 Average speed vs ideal average speed The average speed can be easily calculated with the following formula t 2d x Xen Xt Via Average speed tstop tstart With tstart and tstop representing the respective start and stop times of the period over which the average speed is calculated 4 2 1 1 2 Route efficiency The route efficiency metric can be calculated for every order on the order list from which the loading and unloading points are known As indicated by the name this metric measures the efficiency of the chosen route while fulfilling an order Shortest distance between start and endpoint Route Selection O V FP o V OBnFh Actual driven distance To be able to calculate this metric the following information will be needed e Actual driven distance between start and endpoint gt RFID data gt Order list progress e Shortest distance shortest path between start and endpoint gt Current location
89. ion 15 14 37 50 3 049 0 170 0 381 0 381 Lane 18 Loading station 15 Duration 5 2 LowerThreshold Timestamp Coordinates Distance Speed Area Purpose Status Order x y of area 14 37 55 3 027 0 147 0 002 0 002 Lane 18 Loading station 15 14 37 54 3 029 0 148 0 016 0 016 Lane 18 Loading station 15 14 37 53 3 013 0 148 0 016 0 016 Lane 18 Loading station 15 14 37 52 3 027 0 138 0 024 0 024 Lane18 Loading station 15 14 37 51 3 051 0 143 0 027 0 027 Lane18 Loading station 15 14 37 50 3 049 0 170 0 381 0 381 Lane 18 Loading station 15 Duration 19 gt UpperThreshold Timestamp Coordinates Distance Speed Area Purpose Status Order x y of area 14 38 09 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 08 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 07 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 06 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 05 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 04 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 03 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 02 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 01 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 38 00 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 37 59 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 37 58 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 37 57 3 027 0 147 0 000 0 000 Lane 18 Loading station 15 14 37 56 3 027 0 147 0 000 0 000 Lane 18 Loading s
90. istances in comparison with passive tags As opposed to active tags passive tags don t require a power source In a passive system the readers also send out electromagnetic waves that can induce a current in the passive tag s antenna This current is then used to power the tag so it can transfer its EPC back to the reader The last class uses semi passive tags These tags require an own power supply to power the internal circuits but use a combination of the EM waves sent by the reader and its own power to broadcast its data Angeles 2005 Want 2006 Figure 10 From left to right a passive RFID tag an active RFID tag and an RFID reader courtesy to Ubisense Next to the fact that RFID readers can receive the data stored on the tags in their proximity they can also determine the exact location of the RFID tags By using multiple readers the location of an active tag can be calculated by using triangulation Further modern RFID tags have increasing possibilities attached to them One very interesting aspect of modern RFID tags is that they can also convey information from onboard sensors they contain next to their usual identification For example a passive force sensor can be incorporated into an RFID tag When the product with the RFID tag attached to it is dropped on the floor and the impact could have damaged the product the passive force sensor will supply one single bit alerting the system about the problem An
91. iting Overall utilization r Total time 4 2 3 Percentage loaded This metric looks at the percentage of the travelled distance that the logistics vehicle is loaded This gives an indication of how efficient the order list is set up to be The more a vehicle is loaded the more efficient the order list and thus the in plant logistics P t m Distance travelled while loaded namen E The distance travelled while loaded can be calculated as the sum of the travelled distances between each loading and unloading point The total travelled distance is then calculated as the sum of all the travelled distances 4 3 Reliability 4 3 1 Lost time percentage This metric gives an indication of how much time is lost due to various errors that could occur in the in plant logistics It compares the lost time with the total time and obviously the lower the value of this metric the better the reliability of the system te eee _ Lost Time ost Time Percentage Total Time 33 Though this metric already says something about the reliability of the system it does not yet give a clear indication about the real problem To find the underlying cause of the lost time the reliability of the in plant logistics is further measured with more specific metrics on a lower aggregation level These metrics each focus on a more specific type of error and are described further 4 3 1 1 Vehicle reliability Vehicle reliability focuses on the porti
92. l to configure zones in the dashboard file 67 Node 1 and its connected nodes 0 and 2 on the train track layout 69 Locations of the used nodes 69 Buttons on the dashboard sciinte iae auqa b ua 70 Flowchart of the dashboard update 71 Route selection tool and input box 76 Result in the route selection tool nennen enne 76 xiii List of Tables Tableatr Developedumetries n ene I ee derden des 19 Table 2 Indication of the progress in the order list 25 Table 3 Format of the logged raw RFID data a 50 Table4 Useful part of the raw data znne ien innen eoe ie eorr ees 50 Table 5 Gaps in time in the logged RFID data 51 Table 6 Format of the cleaned and smoothed RFID data eene 51 Table 7 Used order list in the test aasan u mms aaa i ua naa asa 53 Table 8 Performance of the dashboard at different update rates
93. lines and design tips is given which will be followed when designing the dashboard 5 1 Definition and Key features of a dashboard A dashboard could be defined as an graphic interface which organizes and presents important business information in a way so that it can easily be interpreted It should have an intuitive design and be tuned to fit the needs and expectations of the user To make the dashboard as clear as possible a couple of guidelines have to be kept in mind Van Goubergen 2010 Pureshare e Use intuitive colours and symbols to represent the data The goal of the dashboard is to provide as much information as possible at a glance By using intuitive colours e g green for good situations red for bad situations data can be interpreted more quickly Further symbols or indicators such as arrows can be used to show the link between various graphs and bring structure in the entire dashboard e Use the same kind of presentation for all related metrics The graphs or metric presentations should remain consistent If a pie chart is used to display the overall occupation of the internal logistics then on another aggregate level occupation of one vehicle in the last hour the metric should also be depicted with a pie chart This should be done to avoid confusion e Visualize trends By displaying the performance over a few periods instead of just one a trend can be visualized This trend can provide information abou
94. lockage and interference of the RFID tags will be more likely The vehicle will now be driving around in a real manufacturing plant between racks and machines This can cause blockage or interference of the RFID signals Because of the bigger size of the area more RFID readers will have to be used to maintain a strong signal over the entire plant e Inaccuracy of the location and speed will be less significant During the test it was seen that the raw RFID could contain some inaccuracies which caused the vehicle to appear Off track Not all these inaccuracies could be fixed by cleaning and smoothing the data These inaccuracies were significant in the test because of the small scale of the train and the train track However in a real life situation this inaccuracy is relatively small in comparison to the big distances that have to be travelled by the vehicle The second difference between a real life situation and the test setup is the time over which the performance is measured During the test 45 minutes of RFID data was acquired This is a rather limited dataset in comparison to the data that would be gathered in a real life situation If the dashboard would be used in real internal logistics department the performance would be measured over an entire shift 8 hours This is more than ten times the duration of the test Because of this huge amount of data the dashboard would be more strained than during the test and the performance of
95. low these parameters are displayed as in the dashboard excel file and their purpose is described Table 9 Adjustable parameters of the dashboard USED PARAMETERS Location of the cleaned RFID data file path C Users Simon Desktop Datafile xlsx Name of the cleaned RFID data file Datafile xlsx Update rate hh mm ss 0 00 10 THRESHOLDS Speed Threshold m s 0 11 Lower Handling Time Threshold 96 0 5 Upper Handling Time Threshold 96 1 8 Error Threshold hh mm ss 0 00 15 Ideal average speed m s 0 45 64 The first two parameters Location of the cleaned RFID data file and Name of the cleaned RFID data file are used to determine where the dashboard file gets its RFID data Here the path and the name of the excel file that contains the cleaned RFID data have to be entered The third parameter is the Update rate With this parameter the user can determine how frequently the dashboard should be updated In the example above the dashboard will be updated automatically every 10 seconds The next four parameters are the thresholds that are needed to correctly calculate the status of the logistics vehicle Due to the inaccuracy of the gathered RFID data it is not always clear when the vehicle is stationary or when it is moving small fluctuations in the measured position can make it appear as if the vehicle is moving very slowly in random directions To counter this a
96. m de interne logistiek zo effici nt mogelijk te laten verlopen is het noodzakelijk dat dit proces eerst gevisualiseerd wordt Vervolgens kunnen de verschillende vormen van waste gevonden worden en nadien ge limineerd worden Om de interne logistiek te visualiseren wordt een performance measurement system opgesteld Er wordt gebruik gemaakt van real time analyse om zo goed mogelijk het complexe gedrag van de interne logistiek te kunnen weergeven Deze real time analyse wordt mogelijk gemaakt door het gebruik van RFID technologie Elk gebruikt logistiek voertuig wordt hierbij voorzien van een RFID tag die toelaat hun beweging doorheen het bedrijf te volgen De rest van dit artikel is als volgt ingedeeld Eerst wordt een korte literatuurstudie gegeven waarin gelijkaardige onderzoeken kort worden besproken Vervolgens wordt de ontwikkeling van de metrics voorgesteld Daarna wordt een dashboard ontworpen die de metrics visueel presenteert Uiteindelijk worden de ontworpen metrics en het dashboard onderworpen aan twee testen Het artikel wordt afgesloten met een algemeen besluit over het onderzoek II LITERATUURSTUDIE Drie gelijkaardige onderzoeken worden hier kort uitgelegd Chow et al stelden een RFID case based logistics resource management system R LRMS voor om de effici ntie van order picking operaties in een warehouse the verhogen Het voorgestelde systeem berekent telkens het beste logistieke voertuig qua locatie en capacit
97. maginary loading or unloading action To obtain more divers data the vehicle was also sent to the parking zone Lane 6 at different times to simulate planned breaks Idle time These breaks occurred after orders 5 7 and 9 Further two defects were simulated by stopping the train for roughly 20 and 30 seconds during orders 6 and 7 Finally also an Off track error was simulated by taking the train of the track after order 13 Further in this chapter these events are indicated in the results 53 6 5 Progression of the test During the test at HoWest the order list was fulfilled while the RFID data was recorded However the dashboard was not used at that time so there was no dynamic analysis of the data The reason for this is that the acquired data was still raw at that moment and needed to be cleaned before it could be used While the test was being done the correct parameters were determined for the cleaning of the data Afterwards the entire dataset was cleaned so it could be used for the dashboard Because of the fact that the data was analyzed to determine the right cleaning parameters and because of small problems with some of the sidetracks of the train track sometimes the sidetrack could not be switched automatically and needed to be adjusted manually the order list was not followed exactly All the orders have been fulfilled but during some of the orders long unexpected stops had occurred These long
98. me dashboard The second test makes use of the same RFID data that is gathered in the first test However instead of loading the data in the dashboard all at one time the data is dynamically inserted into the dashboard By changing the update rate of the dashboard its performance can be checked for each possible update rate For each update rate it is checked how long the update takes for calculating and executing the necessary functions to adjust the graphs and if the dashboard is still displayed correctly capabilities of the test The test concludes that a maximum update rate of one update every two seconds is still possible which is more than satisfying VI CONCLUSION The presented metrics and their presentation in the dashboard can provide useful information about the performance of the in plant logistics vehicles From the tests it can be concluded that the dashboard offers a correct representation of the actual situation and that it can be done in real time For further research these metrics and their presentation can be applied on real life cases to see the performance of actual logistics vehicles e g AGVs forklifts and tugger trains REFERENCES 1 Chow H Choy K L Lee W B and Lau K C 2005 Design of a RFID case based resource management system for warehouse operations Expert Systems with Applications 30 2006 p 561 576 2 Ludwig T and Goomas D 2009 Real time performance m
99. me due to picking errors Picking reliability metdTime 4 4 Productivity 4 4 1 Number of orders picked hour As the name says this metric simply looks at the number of orders picked per hour This can give an indication of how well the internal logistics is working However this metric can be misleading as not every order has the same work content Some orders require more time due to the greater distance that has to be travelled or the longer time that is needed to perform the un loading of a certain item If multiple orders with large work content have to be fulfilled after each other the value of this metric will be low but this does not mean that the overall performance is worse than before 4 4 2 Number of orders picked driven km As the previous metric this metric can also be misleading as some orders require greater distances to be travelled This metric does however give an indication of how well the layout of the plant is If this metric has a persistent high value this could mean that the warehouse is situated too far from the border of line or the infrastructure is not efficient 35 5 Metric presentation The metrics discussed in the previous chapters now need to be presented in a way that is clear to the user and that allows a fast interpretation of the performance of the in plant logistics Here the design of a dashboard is described First the concept of dashboards is explained and a set of guide
100. n a test dashboard as displayed below It has a more compact design than the multi screen dashboard that was described earlier as it only consists of one single screen TRAIN 1 ie Empty Last Update 14 54 56 16 13 57 Dashboard mieren eeen Current Order 18 Lane 1 Progress Drive to loadingstation 1 Error 4 ume Load Didie sii Transportation K Lane 5 B Waitin Ex Unload E Lane6 Lane7 Print Report B unloading Lanes Lane 9 Lane 10 Lane 11 9 Lane 12 Lane 13 Lane 14 Lane 15 Lane 16 B Loading B Driving F T Lane 17 14 15 14 05 Lane 18 Coordinates CurrentLocation m Errors Status Occupancy in last 10 minutes i il ion Error Event Manager Change in Lost time Nbr Problem Starttime Duration Location Causes of lost time in last 10 minutes 141120 14 14 28 waiting 1 14 14 47 Loading 14 15 08 14 17 31 unloading Error Offtrack TS me Defect Didle rror Defect rn En 14 20 12 14 30 03 14 30 17 14 44 57 Total lost time 402 sec 14 30 18 14 54 56 in last 10 minutes 14 31 14 Driving Figure 40 Test dashboard 62 Display Post analysis Supporting sheets 7 3 Working of the test dashboard Behind the dashboard s screen the necessary calculations and functions are executed to assure that the data is processed correctly and the presented graphs are updated accordingly The programming of these functions has been done in V
101. n figure 32 The train can reach every section of the track by a series of sidetracks which are centrally controlled but can also be switched manually The RFID readers are placed just outside the four corners of the train track indicated in green and thus cover the entire area where the train can move under normal circumstances The size of this area is approximately 25 m2 For most of the test only the upper section of the track is used Later also some other parts of the track are used to examine the route efficiency of the vehicle Figure 32 Layout of the train track with the four RFID readers To interpret the movement of the train the track is divided into several zones However the declaration of these zones will be described later in this chapter 48 6 2 Selection of the parameters and cleaning of the data To get an accurate result for the metrics the RFID data has to be filtered and cleaned before it can be used The Ubisense software allows the user to adjust some parameters to obtain the desired accuracy of the acquired data Here not all these parameters will be discussed as the cleaning and filtering is not really part of the thesis However a brief description will be given to demonstrate how the raw RFID data is transformed into cleaned data The Ubisense software allows the user to choose the update rate of the measurements This rate determines how much data points coordinates are measured each secon
102. n introduction to RFID technology EEE Pervasive Computing Magazine 6 p 25 33 82 11Appendices e Appendix A Completed Metrics Development Matrix e Appendix B Flowchart of the status update e Appendix C Test dashboard output e Appendix D Performance report Development Matrix Completed Metrics 11 1 Appendix A WPN poxajduioo st Japio ue auum peg paxo duuoo sr Japio ue auum upeg puooas A193 puooas A1anq puooas A193 puooas A193 puooas sang puooas Aang puooas uana puoaes Asang puooes A1ang puooes A1ang puooas Asang puooas Asang puooas uad puooas Asang payejdusos Japso ue aw Yoe puooas sang puooas A193 leo ayequone Aouanbasy ueuejdug uono lo2 exea sij JopJ0 ezep QW Japio eyep Gide andu jenuew eyep aide e1ep aldy anduy jenuew exp qid ep gis e ep DIN ep aldy I 3x eyep aij Japio eyep did sewn psepuess Japuo eep Ise sawn paepueys Jopao eyep ads I 3x uuuo3je ed 1se10us eep Aldy ep Old ja0xa uutnuoS e yred ysayoys ezep Aldy Japio ezep Idy axe Isil 19pJO jooxo ezep qiu ep OIJY s loo1 uon ajo exea utiqisuodsoy uon llo2 ezea suy uoAup oui 1sureSe poxpid ssapso Jo JqN ayy Suiwous udeuo sunou x Jano pu n au 8 iounseou 01 Moy ON ena aouewsopsad smoys tpluM ajeos inojoo e pue s lu AIl p Buoum Jo JN aui Buimoys a3neo eyo meuyBeds au ul Jensa smess u Buju
103. ne 2 Lane 5 10 10 This order list has to be manually placed into the OrderList sheet before the dashboard is used 7 3 3 3 Zones The third required input for the dashboard is connected to the layout of the manufacturing plant or in this case the layout of the train track The track is divided into different zones which can then be used to indicate the location of the train and of the loading and unloading points As mentioned under 4 1 1 Location of the vehicle each zone is defined by four coordinates the corners of the zone that enclose a convex area To insert new zones into the dashboard a tool was programmed in VBA This tool helps the user add a new zone by asking for the coordinates of the corners and the position of the four lines The zone is then automatically configured and can be used in the dashboard This tool is displayed below Add A New Zone x rx Add A New Zone e J Insert Coordinates Line 1 Line 2 Line 3 Line 4 Insert Coordinates Line 3 Line 4 Insert name of zone Zone 1 Insert Coordinates Location of the zone vs line C lt smaller than X Y 1 8 5 5 18 6 0 4 6 VERTICAL OR SKEWED lt to the LEFT of the line to the RIGHT of the line HORIZONTAL Next below the line above the line Figure 41 Tool to configure zones in the dashboard file 67
104. needed because all the needed items can be reached more easily The items in the supermarket can be picked and transported to the border of line by a so called mizusumashi This is a logistics vehicle that supplies the border of line by driving around in a fixed route with a fixed cycle time Mostly tugger trains or AGVs automatic guided vehicles are used for this These vehicles can tow multiple carts with items so they don t need to make as many trips as would be needed with forklifts Typically a mizusumashi will pick up items at the supermarket and then make multiple stops at different places to resupply the border of line Coimbra 2009 Ww qM LE JEJE Emmm OIO L CI Y JY JY Figure 1 Forklift supply vs mizusumashi supply 2 1 1 Logistics vehicles Underneath a brief description is given about the most commonly used logistics vehicles Van Landeghem 2008 2 1 1 1 Forklifts Though forklifts are not the most efficient means of transportation they are still often used in the in plant logistics They are well suited for the handling of materials because of their flexible design With their lift mechanism they can pick products out of high racks difficult to reach locations and also stack products on top of each other Further the spacing between the forks can be adjusted so the forklift is able to handle a greater variety of products and several attachments can be added to the forklift to handle more spe
105. ning because the shipments have arrived then Underneath the bar chart an indication is given about the change in utilization and lost time during the last hour in the example above the utilization and lost time are compared between 11 00 and 10 00 A positive change will be indicated in green the lost time was reduced with 3 which is good and a negative change will be indicated in red the utilization has gone down by 3 which is bad By using these colour indications the overall legibility of the dashboard improves The location of the vehicle is presented in the second timeframe last hour It is displayed in the form of a spaghetti chart which consists of a line connecting all the locations the vehicle has been to in the last hour This spaghetti chart can be useful because it shows all the paths that have been taken and also the frequency in which a certain area has been visited this can be seen because more lines will overlap 5 2 5 Screen 3 Efficiency Forklift 1 Efficiency Return to Summary Current order Route selection Transportation efficiency Load Unload Drive to loadingpoint V efficiency Loading 0 00 22 Transportation 0 00 17 Unloading okas 0 00 09 0 00 00 Distance m mioading M Unloading mGoal ow o 2 Pot OU WY Qe DD D D X9 QU QU X DU P ob QM af DD D F D SS SSS SS SS SS ST HK HS Load Unload 27 NN Actual distance travelle
106. ning of this data is described here in a couple of steps 1 The first thing that needs to be done to clean the data is to remove all the unnecessary information In the test the train moves on a flat platform Therefore the z coordinate is not useful Further the raw RFID data contains a Cell column This column is empty because the test setup only consists of one RFID system one cell This column can thus also be removed The TaglD and Tagname are also not needed because only one train will be used in the test Though if multiple vehicles are monitored this information would be needed The last column contains the BatteryLevel of the active RFID tag Though this value is equal to zero in the entire column the real battery level was good it was tested before and after the test This column also has no meaning for this test so will be removed as well The remaining raw data is displayed in table 4 Table 4 Useful part of the raw data Date x y 14 11 00 3 694071 4 677925 14 11 00 3 943779 4 620492 14 11 00 4 072446 4 606907 14 11 00 4 073717 4 610288 14 11 01 4 080714 4 61588 14 11 01 4 211967 4 616945 14 11 01 4 219471 4 617785 14 11 01 4 237974 4 619946 14 11 02 4 266108 4 617431 14 11 02 4 302099 4 610484 14 11 02 4 344167 4 595333 50 2 As can also be seen in the logged data an empty row is placed after every eleven rows indicated in red in table 4 These empty rows need to be removed from
107. ntories were known as opposed to the previous situation in which inventory data was recorded manually The visibility of the warehouse was also improved the exact locations of the SKUs and the material handling equipment the forklifts are now known Further the job assignment process was now automatic and could be done in a much smaller amount of time Chow et al 2006 Poon et al 2009 Ludwig and Goomas researched the effect of real time performance monitoring and feedback on the efficiency of forklift drivers Though this research is more about the psychological effects of the real time performance measurement the implemented system is still interesting for this thesis subject 12 In this research the performance of the forklift drivers was measured by comparing the actual time they needed to fulfil an order to the standard time This standard time could be calculated by subdividing a task into its basic elements By then performing work measurement techniques the needed time for each element could be calculated These times were then added together and an allowance for fatigue was also taken into account The standard times were calculated for the travel time and the loading and unloading times of the vehicle The travel time is based on the shortest distance between the start and stop point taking into account corners and passageways The loading and unloading times can be divided into arm lift and drop times based on how high
108. oints would be Off track not in any of the defined zones Finally it was also noticed that the calculations intensify when more zones are defined which causes the dashboard to react slower 52 6 4 Used order list To simulate a logistics vehicle that is performing tasks a small order list was made up During each order the vehicle had to pick up an object on one location and drop it off on another location However the train in the test setup could not really perform loading or unloading actions so to simulate this the train was simply stopped at the loading and unloading point for a certain amount of time The used order list during the test is given in the table underneath Table 7 Used order list in the test Order Load zone Unload zone Standard time Loading Standard time Unloading NBR sec sec 1 Lane 2 Lane5 10 10 2 Lane 7 Lane 3 10 10 3 Lane 4 Lane 8 10 10 4 Lane 7 Lane 3 10 10 5 Lane 2 Lane 5 10 10 6 Lane 7 Lane 3 10 10 7 Lane 5 Lane 7 10 10 8 Lane 2 Lane 5 10 10 9 Lane 7 Lane 3 10 10 10 Lane 2 Lane 5 10 10 11 Lane 2 Lane 5 10 10 12 Lane 2 Lane 5 10 10 13 Lane 2 Lane 5 10 10 14 Lane 18 Lane 9 10 10 15 Lane 18 Lane 9 10 10 16 Lane 9 Lane 13 10 10 17 Lane 9 Lane 13 10 10 As can be seen in the table the loading and unloading zones are given for each order Further a standard time is given for each action This standard time represents the time that is needed to perform the in this case i
109. on of lost time that is caused by defects of the vehicle Though it is hard to determine whether a vehicle is defect based only upon RFID data a few assumptions can be made that allow the system to detect abnormal situations When a vehicle stops in a place where it is not supposed to stop e g in the middle of a hallway or corridor and stands still for a long time this can be considered to be abnormal Either the vehicle has become non responsive due to a defect of the vehicle or another problem related to the driver or the environment has occurred causing the vehicle to stop Lost Time due to defects of the vehicle Vehicl liability ehicle reliability Total Time The vehicle reliability depends on two factors the quality and condition of the used logistics vehicles and the efficiency of the maintenance department The efficiency of the performed maintenance can also be measured in the following metric Mean Time To Repair 4 3 1 1 1 Mean Time To Repair The Mean Time To Repair MTTR measures the average time a vehicle remains out of order when a defect occurs This gives an indication of how fast maintenance can be done or how serious the defect is A high value on this metric would indicate that improvement in the maintenance is needed MTTR can be determined by taking the average of the duration of every single vehicle defect This duration can be determined by looking at the status of the vehicle 4 3 1 2 Route reliability
110. onitoring goal setting and feedback for forklift drivers in a distribution centre Journal of Occupational and Organizational Psychology 82 2009 p 391 403 3 Kootbally Z Schlenoff C Madhavan R 2009 Performance assessment of PRIDE in manufacturing environments http info ornl gov sites publications files Pub2 1 558 pdf Last Update 14 54 56 _ 16 13 57 TRAIN 1 Update stop Eoo Print Report Current Status Current Location Current Order Progress Status Occupancy per 10 minutes Lane 9 18 mer Drive to loadingstation ad Load Transportation Unload x Didie E Waiting Utilization in last 10 minutes B Loading B Driving Change Change in Lost tii Status Occupancy in last 10 minutes Causes of lost time in last 10 minutes Driving waiting 1 WiLoading Unloading Error Off track Error Defect Didie 3 2 Start time End time 14 44 57 14 54 56 Total lost time in last 10 minutes B Unloading Spaghettichart last 10 minutes Error location inate CurrentLocation W Errors Error Event Manager Nbr Problem Starttime Duration Location Zone Cause x Y 1 Off track 14 11 20 0 00 01 1 459 5 171 Lane6 2 Defect 14 14 28 0 00 17 5 394 4998 lane4 3 Off track 14 14 47 0 00 01 5 831 3 977 lane4 4 Off track 14 15 08 0 00 01 4
111. onsider both steps together and thus present the transportation efficiency for the entire order m Expected transp time transp erfieeney Actual transp time Expected transp timestep 1 Expected transp tiMestep 3 Actual transp timestep 1 Actual transp timestep 3 The actual transportation time can be obtained from the RFID data and the status update by simply subtracting the transportation start time from the end time For the value of the expected transportation time two things need to be known the ideal average speed and the shortest possible route The ideal average speed will be determined beforehand based upon the capacity of the logistics vehicle and speed rules within the plant The shortest path will be determined by using shortest path algorithms 4 2 1 1 1 Average speed The actual transportation time depends on the chosen route distance that has to be travelled and the average speed of the vehicle The average speed can be considered as a metric because it gives an indication of how well the vehicle is driving This average speed can be compared to the ideal average speed and conclusions can be drawn from this If the vehicle drives much slower than the ideal average speed the transportation efficiency will always be low The vehicle should thus drive faster If the vehicle drives much faster than the ideal average speed this can be good for the transportation efficiency as the vehicle will always reach its dest
112. or unacceptable red This allows the dashboard user to quickly react when the utilization or efficiency goes into the danger zone yellow The lost time is shown in a pie chart This pie chart displays the working time green the idle time yellow and the lost time red Underneath this chart the percentage of lost time is displayed to make it more clear 39 5 2 4 Screen 2 Visibility Forklift 1 Visibility Return to Summary Status Status Occupancy per hour Occupancy in the last hour E o wo o u o B o WiDefect Time min Sidle w o EB Waiting B Unloading BLoading N o m o B Driving o 4 Hour Utilization Spaghettichart of the last hour Figure 22 Screen 2 Visibility This screen zooms in on the status and the location of the logistics vehicle The status of the vehicle is presented in two different timeframes Occupancy during the ast hour and during the ast day In the pie chart the occupancy of the last hour is given This indicates which tasks the vehicle has performed in the immediate past and how busy it was how much time the vehicle was actually driving loading or unloading The bar chart presents the occupancy per hour for the entire day This view can be useful to detect certain trends in the performance of the vehicle e g the vehicle could have a higher utilization in 40 the mor
113. ormance in the 37 last hour an indication that a coincidental error occurred or is the performance always at this low level As the purpose of this dashboard is to present the real time performance the used timeframes are rather limited in length The performance can be seen over each individual day but in the dashboard the performance per day cannot be compared This can however still be done by printing out a performance report each day and comparing these reports with each other This performance report is basically a summary of the performance during one day 5 2 2 Multi screen design The proposed dashboard consists of multiple screens One central summary view will be used to present some basic metrics for the overall performance of the internal logistics Here the information is given about every used logistics vehicle To provide a more in depth view of the performance of each individual vehicle links are added in the summary view to other more detailed views These extra views are subdivided in Visibility Efficiency and Reliability As can be noticed here there is no separate screen that gives a more detailed view of the Productivity of the in plant logistics The reason for this is that the productivity metrics can easily be misinterpreted as described under 4 4 Productivity By displaying them on the dashboard wrong conclusions could be drawn Each screen of the dashboard is
114. ors If the window is chosen too large the system reacts sluggish and might miss transitions of the tag going in or out of the range of the reader The window size should thus be adapted according to the behaviour of the data For this problem Researchers at UC Berkeley presented an approach to adapt the window size dynamically based on statistical sampling SMURF or Statistical sMoothing for Unreliable RFid data Jeffery 2006 This technique could be used for cleaning and smoothing the raw RFID data that will be used in this thesis but as the cleaning is not part of the thesis itself it will not be discussed further 2 2 4 Current applications RFID technology has been proven useful in many fields Some examples of applications are given below Polniak 2007 Animal identification One of the first applications of RFID technology was to identify cattle by implanting RFID tags under their skin Next to the identification other data could be included on the tags such as the age and vaccination records Now RFID tags are also being used to identify pets in case they have been lost Anti theft systems A well know application is the use of RFID tags as anti theft system Here a tag is attached to the product and RFID readers are placed at the exit of a store When a product is brought to the checkout counter the tag is either removed from the product and reused or in case of disposable passive tags destroyed by subjecting it to a strong
115. other example is an RFID temperature sensor which can be attached to perishable goods such as meat or dairy products Once a certain temperature is exceeded the tag will give a warning that the product is not fit for consumption anymore Want 2006 2 2 1 4 Challenges Despite the advantages that RFID technology has to offer some issues remain that are holding back the widespread adoption of this technology The three main issues here are the cost design and acceptance Currently RFID tags are available at low prices around 13 cents per tag but are still much more expensive than printed labels These prices are still decreasing as time goes by but are not yet at a price tipping point and therefore the use of RFID tags instead of bar codes is not yet profitable The second important issue that holds back the adoption of RFID technology is the design of the tags and readers Currently RFID data can be fairly accurate but it still has a certain in some cases unacceptable error margin The reliability of the identification should be further improved before widespread adoption will be obtained The third issue is the acceptance There are some general concerns about the privacy of the consumers If every product were to be equipped with RFID tags the consumer could be tracked by the tags on their bought products In theory vendors could use this information to analyse the consumption behaviour of their customers If the usage of RFID tec
116. ply chain In plant logistics requires a lot of man hours and materials which brings a great cost with it In order to limit these costs it is important that the in plant logistics is performed as efficient and effective as possible Excellence would be obtained here by assuring that the logistics process is not a limiting factor in the production and that every order is completed on time In the past mainly forklifts were used to pick pallets from the warehouse and transport them directly to the border of line where they were needed traditional supply Though a forklift is able to pick pallets out of the high racks in the warehouse these handlings take a lot of time and the load capacity number of pallets that can be carried at a time of the forklift is rather limited Because of this limited load capacity the forklift would need to make many trips between the warehouse and the border of line to supply the necessary items With the increasing popularity of lean implementation and Total Flow Management TFM the tendencies in internal logistics have changed The use of forklifts to supply the border of line has decreased because of its inefficiency as described above Instead of picking the orders out of warehouses with high racks supermarkets are now used flow supply These supermarkets are designed to allow quick and easy order picking All the needed items are now placed in low flow racks or on wheeled bases Forklifts are no longer
117. rd consists of graphs bar charts pie charts and other graphic representations of the metrics which are updated dynamically In the accompanying thesis the design of a multi screen dashboard is proposed For the exact design of this dashboard I refer to the thesis itself To perform the tests as described in chapter V a test dashboard was developed in MS Excel This dashboard is less extensive as it only consists of one screen Figure 1 presents the design of this test dashboard V TESTING OF THE DASHBOARD AND RESULTS Two tests are done to check the accuracy and correctness of the displayed metrics and to check the real time aspect of the test dashboard For these tests a test setup at HoWest in Kortrijk is used The test setup consists of a toy train riding on a track An RFID tag is attached to the toy train and its movements can be followed by four RFID readers that are placed around the track A Test 1 Accuracy and correctness of the displayed metrics In the first test the train is ordered to fulfill a predetermined order list The RFID data is measured during the test and is cleaned afterwards The cleaned RFID data is then loaded into the dashboard and the results are compared to the actual performance of the train during the test In this test it is concluded that the monitored data in the test dashboard is indeed accurate and correct in comparison with the real measured performance of the trainB Test 2 Real ti
118. rformance Below the working of the dashboard update will be explained step by step First an overall view is given of how the update cycle works and then the used steps are discussed more in detail The dashboard contains four buttons that are required for its functioning Berekenen Update Stop tis d Print Report ashboar Figure 44 Buttons on the dashboard e Update This button starts the automatic updating of the dashboard When the button is pressed the dashboard is updated immediately and the timed update cycle will begin The dashboard will then be further updated at the update rate that was specified in the Parameters sheet e Stop The automatic update cycle can be stopped by pressing this button e Empty Dashboard This button allows the user to empty and reset the entire dashboard file in order for the test to be restarted or a new test to be started e Print Report After the test is done the user can choose to print a report that summarizes the performance of the system during the test By pressing this button the report is generated in a new excel sheet 70 When the Update button is pressed the update cycle of the dashboard starts The general working of this update cycle is displayed in the flowchart underneath Start Update Restart every x seconds dependant of the update rate Figure 45 Flowchart of the dashboard update 71 7 3 4 1 Step 1 Update of the RFI
119. rformance of the vehicle e The hourly occupancy status division per hour can be displayed to show possible trends e g changes in utilization or increase decrease of defects 4 1 3 Order list progress This metric visualizes the progress in the order list Not only the current order can be displayed but also the progress within the order itself As mentioned before under 4 1 1 Location of the vehicle the order list contains the zones which will be used as loading and unloading points Further it also contains the standard times for the loading and unloading which will be needed when calculating the loading and unloading efficiency As seen under 4 1 2 Status of the vehicle the status is partly determined by the information in the order list because the purpose of the area depends on the location of the loading and unloading points see figure 15 which is determined by the order list However the order list in its turn will also be determined by the status update For example when the status changes to Loading the order list will be updated and the progress within the current order will be changed to Loading Whenever the status changes again the order list will be further updated and when the Unloading status ends the current order will be considered as completed and a new order will begin The format of a used order list is displayed below Table 2 Indication of the progress in the order list
120. rt If the route efficiency is very low this would mean that the logistics vehicle has taken a much longer route than needed and that a great improvement could be obtained if this shorter route was found For this reason a tool was added to the dashboard file to allow the user to review the actual taken route and the shortest possible route This is explained here by an example From the performance report appendix D Performance Report it can be seen that the route efficiency of order 6 is only 29 The selected route of this order will be compared against the most efficient route In figure 46 next page the route selection tool is displayed This sheet consists of two screens indicating the layout of the train track On the left screen the route to the loading point will be presented On the right screen the route between the loading point and the unloading point will be presented 75 Route selection From previous location to loading point From loading point to unloading point FROM TO FROM Anito A Sorten teus Choice of order Which order do you want to check Figure 46 Route selection tool and input box Route selection check order Order 6 From previous location to loading point From loading point to unloading point FROM LaneS TO lane7 FROM lan7 TO lane3 K ActualToLoad Shortest ToLoad lt ActualLoadToUnload Shortest LoadToUnload
121. s repeated until every vertex is added to the set K and the shortest distance to every vertex is known 30 diu u YON x S a ge e 5 4 e 0 71 7 1 d 422 Ald 12 2 Figure 19 Example of the use of Dijkstra s algorithm In the original algorithm as described above the shortest path to each vertex is calculated However not every distance needs to be known for the application here When the shortest path to the stop point unloading point is found the algorithm can be ended so no extra calculation time is lost The output of this algorithm gives the shortest possible distance between the loading and unloading point and also gives the path that should be taken This path can then also be displayed on the dashboard and compared against the actual path that was taken 4 2 1 2 Loading and unloading efficiency This metric compares the actual time it takes to load or unload an item with a predefined standard time for each order Mt Standard loading time dai s Actual loading time i En Standard unloading time enk adi Actual unloading time The actual loading or unloading time can be derived by looking at the status update and the order list progress The start and stop time for a loading or unloading action is stored for the order list progress By taking the difference between these two times the actual loading or unloading time is found The standard loading or unloading time is a pred
122. search IL LITERATURE STUDY Three similar researches are mentioned in this literature study Chow et al propose an RFID case based logistics resource management system R LRMS to improve the efficiency and effectiveness of order picking operations in a warehouse This system automatically calculates the most appropriate material handling equipment logistics vehicle for a certain order and determines the shortest path to fulfill this order 1 Ludwig and Goomas propose another system in which the performance of forklift drivers is measured and feedback is directly given to the driver so he can adjust his behavior The performance is here measured by comparing the actual time it takes to complete an order with the standard time This standard time is calculated based on the shortest path that can be taken at the ideal speed and the most efficient loading and unloading procedures 2 Lastly Kootbally et al present the PRIDE algorithm which can be used to navigate AGVs in a dynamic manufacturing environment By integrating existing shortest path algorithms with active collision avoidance the reliability and the overall performance of the in plant logistics can be improved 3 III DEVELOPMENT OF THE NEEDED METRICS By using the Performance Measurement Development System PMDS a performance measurement system can be defined This process is used to design a set of metrics for the dynamic analysis of the in plant
123. sem 1924 HO Aa aauewsopad smoys uotum eas inojoo pue pen JJO sao lpiu A s tun Jo JqN ayy Buimoys 8neo sa sS19 J p x1se au jO 2 Buimoys ydeu Ean oN Jap X ISe ay jo YLW Y ys udeuo pey nieu8eds ul Jensy joox3 snaers u Busem 393439 I A I ueuuuop d swous uorum ajeas 1nojoa e pue s 2aJap Jo JqN 2u1 Buimoys sayaw 8neo awm 150 Jo 98e3u2219d y Buimoys HEUP ald sod azijen oys yeyo Jeg puan aj nq sa TNIE 01 sinou x 104 papeo aSeyueoied ayy 8 pu n so 7 azijensin o1 sinou x sano uonezijn aui Suwous yeu Jeg pu n azijensin o1 siopio x JOJ 32u J jJp aui Buimoys ueup jeg pu x azijensia 0 sinoy X JOJ own paepuess ayy sn eum Sulpeolun eme Suiwous yey seg puaJj azijensia 0 sunoy X JOJ awn puepueys ay s aum Juipeoj emoe Bumoys xeu Jeg yred ysayous au SA anos pamojoj jee au Bumoys ueu3 Mayseds paads 3es ne leap oua asuleSe paseduioo paads aSeJane ayy Buimoys a3neo S9A S A SOA SOA SOA oun zsute3e aoueysip uanup payadxa Y 5 eoueistp UaAUp eme eu Buimoys Jew udejo papaau s Suipeojun Buipeo 304 awn psepuers puay e azijensia 03 ssapuo x 40 aun 1210 pi2adxo 1sure8e uun jeyoa jenpe Sutueduuo2 yeys Jeg Jap o 3u24no u sseuSoud Aejdsip xa se Japso 1ue uno Aejdsiq pua aui 3uimoys noy jeaxaf sad Aouedna2o u jo yey Jeg nou 3sej aui J AO eun JOM JO UOISIAID Y smoys yolym 18 u atd 1X 1 se sme1s 1u un2 39 se pI
124. sitive change is displayed in green a negative change is displayed in red Current order The number of the current order is displayed Process within the order The process within the current order is displayed in several steps A green checkmark means the step is completed a yellow exclamation point means the step is being executed and a red cross means the step has not yet been started 61 e Efficiency Overall utilization The utilization is displayed by a gauge meter with a colour scale showing the performance of the system The utilization is measured as the useful time driving loading unloading waiting divided by the total time Reliability Lost time The lost time is already displayed in the Status pie chart as described above Causes of lost time A second pie chart shows how much time is lost by Defects and by the vehicle going Off track Lost time caused by Picking errors has not been included in this test dashboard these errors are also considered to be Defects Details about the errors An Error Event Manager shows the list of errors that occurred together with the time they occurred the duration of the error and the cause of the error which can also be further specified by the user of the dashboard e Productivity No useful metrics that display the productivity of the system have been included in this test dashboard 7 2 Graphic design These metrics are now presented i
125. t how the performance evolves through time e Give descriptive and clear titles and labels to all graphs A graph is only complete when it is accompanied by a clear title and labels The title should clearly describe the metric which is depicted underneath it Labels can be added to give extra information about the metric and its presentation e Provide timestamps It is important that the user of the dashboard is aware of the update rate The time of the last update should be mentioned on the dashboard This is to make sure that the user 36 bases his decisions or interventions on correct data e g the user should not rush to the site of an defect when this defect occurred more than an hour ago e Display goals and thresholds If the goal of a metric is determined it can also be indicated on the dashboard This improves the legibility of the metric The user can then clearly see if the performance is good the goal is reached or if it is not good the goal is not reached The indication of the goal can even be improved by using conditional make up of the graphs For example if a metric is presented by a bar chart the colour of the bars could change from red to goal is not reached to green the goal is reached Next to the goal other thresholds can also be placed on the graph These thresholds can for example indicate danger zones when the performance is too low and could also be used to trigger specific messages as warnin
126. t of the last hour Figure 27 Connection between the four screens of the dashboard To return to the summary view a button is added in the upper left corner of the three detailed screens Amount of lost time Causes of lost time Return to Summary Figure 28 Button Return to Summary view 45 6 Test Setup HOWEST To analyse the working and the real time aspect of the designed metrics and their presentation a test will be performed For this research a test setup at HoWest in Kortrijk will be utilized This setup consists of a toy train riding on a track with multiple sections The train has an RFID tag attached to it which allows its movement to be monitored A test dashboard has been designed that will be used to monitor the performance of the train In this chapter the test setup and the obtained results will be discussed The design and working of the used test dashboard is described in chapter 7 6 1 Details of the test setup 6 1 1 Equipment used As mentioned before in this test setup a toy train is monitored around a track by using RFID technology Here the equipment used in the test setup will be described The following hardware is used to measure the movement of the toy train e One active RFID tag attached to the train e Four sensors placed around the track forming approximately an area of 16m Figure 29 Used equipment RFID reader left and active R
127. t ook een beknopter test dashboard ontwikkeld in MS Excel Dit dashboard wordt verder gebruikt in de testen die hierna worden besproken Het test dashboard is te zien op figuur 1 V TESTEN VAN HET DASHBOARD EN BEKOMEN RESULTATEN Twee testen worden gedaan om de nauwkeurigheid en correctheid van de metrics te bepalen en om de real time capaciteiten van het dashboard na te gaan Voor deze testen wordt er gebruik gemaakt van een testopstelling die bestaat uit een speelgoedtrein die gevolgd wordt door middel van RFID technologie A Test 1 Nauwkeurigheid en correctheid van de voorgestelde metrics Gedurende de eerste test wordt de trein aangestuurd om een vooropgestelde order lijst te vervullen Tijdens de test wordt de locatie van de trein continu gemeten door middel van RFID technologie De verzamelde RFID data wordt nadien gekuist en in het juist formaat geplaatst om dan ingevoerd te worden in het dashboard De resultaten op het dashboard worden vervolgens vergeleken met de werkelijke situatie Uit de test kan geconcludeerd worden dat het dashboard de situatie correct analyseert en relevante data presenteert B Test 2 Real time capaciteiten van het test dashboard In deze test wordt nog steeds gebruik gemaakt van dezelfde dataset als in de eerste test In plaats van de data in n keer in het dashboard te laden wordt deze hier dynamisch ingeladen Door hierbij de update rate van het dashboard te laten vari ren kan er na
128. t them Underneath an example of a plant is given which is then translated into a graph amp gt amp m 0 G 3 2 3 3 3 4 5 10 10 10 5 z diii d VN ER e e 5 3 6 3 7 3 8 r gt c 10 10 10 o 5 j m amp o 9 3 10 3 11 3 2 ER ER ER di w w uU Figure 18 Plant layout left and the accompanying network right On the plant layout above a loading point L and unloading point U are marked Between these two points the shortest path will be calculated To make sure that this path corresponds to a physically possible path the vehicle can only move on the therefore intended roads extra nodes are placed on every intersection and corner of the layout nodes 1 to 12 The vehicle can now move on straight lines between the nodes to reach its goal The physical layout is then removed 29 and the aisles or connections between the nodes are replaced by arcs right side of the figure Each arc is here defined by its connected nodes and its length or weight If the coordinates of all the nodes are known the lengths of the arcs connecting them can be calculated These lengths have been displayed on the arcs Once the graph has been constructed Dijkstra s algorithm can be run to find the shortest path between the start point L Loading point and the endpoint U Unloading point
129. tation 15 14 37 55 3 027 0 147 0 002 0 002 Lane 18 Loading station 15 14 37 54 3 029 0 148 0 016 0 016 Lane 18 Loading station 15 14 37 53 3 013 0 148 0 016 0 016 Lane 18 Loading station 15 14 37 52 3 027 0 138 0 024 0 024 Lane 18 Loading station 15 14 37 51 3 051 0 143 0 027 0 027 Lane 18 Loading station 15 14 37 50 3 049 0 170 0 381 0 381 Lane 18 Loading station 15 Figure 20 Working of the status update in case of loading or unloading 32 4 2 1 2 1 Load Unload time variance In this metric the loading and unloading times per order are compared In normal circumstances these times should be more or less the same When a great deviation between these two times is measured this can point to a problem in either the loading area e g the items are difficult to reach or the unloading area i I Actual loadimg time Load Unload time variance Actual unloading time 4 2 2 Overall utilization The overall utilization compares the time that the vehicle is working to the total time Here the vehicle is considered to be working when its status is either Driving Loading Unloading or Waiting The Waiting status is considered as necessary e g waiting for another vehicle to pass or waiting for a port or door to open and is thus still seen as working This metric also gives an indication to the total amount of idle and defect time ION Time spend driving loading unloading and wa
130. tem The optical guiding system requires lines to be placed on the warehouse or manufacturing floor These lines are then followed by an optical sensor in the AGV The magnetic guiding system requires a cable that is installed beneath the floor which is then followed similarly to the optical system The wireless radio guiding system uses high frequency transmission to direct the AGV to the right path Many different sorts of AGVs exist ranging from light to heavy duty AGVs AGVs can also be equipped with forks or can have a very specific task related design They can also be used to tow carts like a tugger train Figure 7 Tow AGV with cart optically guided courtesy Figure 8 Fork AGV of IntelliCart courtesy of Egemin 2 2 RFID technology 2 2 1 Introduction to RFID technology 2 2 1 1 Definition RFID or Radio Frequency Identification is a technology that uses radio waves to indentify objects It enables identification from a distance and does not require a line of sight as opposed to other similar technologies such as bar code scanning Angeles 2005 Want 2006 2 2 1 2 History This technology first appeared during World War Il and was used to identify approaching planes The British army developed an Identify Friend or Foe IFF system which works on the same basic concept as RFID technology today A transmitter was placed on each British plane and when it received signals from radar stations on the ground it responde
131. the dashboard could differ To deal with this the data needs to be aged as was described in the literature study As all the RFID data is changed into meaning events the 77 order list is entered and every error is noted in the Error Event Manager the more detailed information of the RFID tag s location can be removed from the dataset when it reaches a certain age The third difference is the overall behaviour of the vehicle The behaviour of the used train in the test could best be compared with that of an AGV or perhaps even a tugger train The train drives on fixed tracks and a loading action or unloading action can simply be recognized when the train is stopped at a loading or unloading point When this is however compared to the behaviour of a forklift it can be noticed that they differ entirely A forklift does not drive on fixed paths and has a higher mobility than the used train Further to load or unload the forklift does not simply stop and wait at the right location until the items are picked up or dropped off Instead the forklift performs a specific manoeuvre to load or unload an item To recognize this the status update as used during the test would need to be reprogrammed to fit the behaviour of the forklift A fourth difference that can be noted is the presence of multiple vehicles During the test only one train was used the reason for this was that only one RFID tag could be logged at a time due to the used softwar
132. the fork needs to be lifted for the action and pallet retrieval and put away times based on the type of product handled and the storage location The actual times were measured by the time stamps provided by the wireless units attached to each forklift The forklifts were then equipped with screens that displayed the time goal and also the performance of the driver By communicating this performance to the drivers immediately the performance increased according to a test case Ludwig and Goomas 2009 Another interesting application of RFID technology in the performance measurement of internal logistics is the PRIDE framework as proposed by Kootbally et a PRIDE or Prediction In Dynamic Environments provides an autonomous vehicle path planning system with collision avoidance In this research PRIDE is used to navigate AGVs in a dynamic manufacturing environment These AGVs have to move around the plant between various loading and unloading points To minimize the time needed for this shortest path algorithms such as Dijkstra s algorithm can be used However when another vehicle blocks this shortest path the PRIDE algorithm will be used This algorithm calculates the shortest path but also provides collision avoidance between the different logistics vehicles First the importance or the role of the other vehicle is examined If this vehicle cannot be moved at that time e g because it is lifting items from a rack or has a higher
133. the newest data is placed at the top of the file and can now be used as input for the dashboard Table 6 Format of the cleaned and smoothed RFID data Time x y Speed 14 54 55 3 448 3 359 0 166 14 54 54 3 285 3 387 0 582 14 54 53 2 718 3 522 0 212 14 54 52 2 515 3 584 0 230 14 54 51 1 833 3 350 0 249 51 6 3 Defined zones The train track is divided into different zones that will be used to indicate the location of the vehicle and also the locations of loading and unloading points Underneath the layout of the train track with the defined zones is displayed i Lane 1 B Lane 2 hk Lane 3 Lane4 Lane 5 Lane 6 ane 7 Lane8 Lane 9 hk Lane 10 Lane 11 Lane 12 Lane 13 Lane 14 Lane 15 Lane 16 Lane 17 Lane 18 Figure 34 Layout of the train track with the defined zones Three things should be noted here First of all not every part of the track has been enclosed by zones This is because some parts of the track will not be used in the test It is thus unnecessary to place zones on these locations It can also be seen that the zones do not enclose the track tightly This is because the RFID data is still not entirely accurate due to second latency errors or vectorial acceleration errors Arkan 2010 By making the zones tighter around the track too many p
134. to be stored forever As the data gets older it will be less important especially when the data is used to measure the current performance and the dataset can be reduced The necessary data stays available but useless information is removed from the dataset Next to these guidelines other research has been done in the field of warehousing When the RFID technology is used to track items in the supply chain the generated volume of information can be enormous as each individual item leaves a trail of data as it moves through the plant However this data can be compressed while still preserving all the important object transitions This compression can be obtained based upon the observation that items tend to move and stay together in large groups through early stages in the supply chain e g items are moved in batches on pallets and are placed together on shelves and though RFID data is registered per item data analysis is usually done for a group of items instead of doing so for each individual item separately Suppose each item movement is recorded by the RFID system in a tuple of the form EPC location time where EPC is a unique identifier for each item then the number of records that needs to be stored can be significantly reduced by grouping information together For example if an item stands still on a shelf for an amount of time then it is not useful to maintain a tuple for each single second the item is there Instead a tuple could
135. to better fit the test setup and the used timeframes have been changed to Current time Last 10 minutes and Total time last day because of the relative short duration of the test 45 minutes Further the metrics are applied to one single toy train Though more trains and RFID tags were available it was impossible to log the data of multiple RFID tags to an excel file due to the used software Underneath the used metrics and their adjustments for the test setup are given e Visibility Area of location The track has been divided into different zones and the current zone is displayed in the dashboard Spaghetti chart The movement of the train during the last 10 minutes instead of during the last hour is displayed on a spaghetti chart Current status The current status is displayed by a traffic light Green Working driving loading unloading and waiting gt Yellow Idle at the parking gt Red An error occurred defect or off track Status occupancy in the last 10 minutes A pie chart displays the occupancy during the last 10 minutes instead of during the last hour Occupancy per 10 minutes A bar chart shows the occupancy per 10 minutes instead of per hour Change in utilization change in lost time This gives the change of utilization and lost time in percentage between the current period and the last period linked to the Occupancy per 10 minutes A po
136. trics to provide a balanced view of the performance of the organizational unit of interest and is used to make decisions to improve results Van Goubergen 2010 Here the Measurement System Development Process developed at the EERL at Virginia Tech will be used as a guide to design the performance measurement system 3 1 Measurement System Development Process The Measurement System Development Process MSDP is a tool to analyse currently used measurement systems and to improve these if needed or to develop and implement new ones By continually analysing the measurement system it is assured that the system remains in line with the business objectives mission and vision and that it is updated when the unit of analysis changes The MSDP consists of 6 steps that that have to be executed in order to obtain a good measurement system As mentioned before this process never ends so after step 6 the cycle repeats itself and the current measurement system is re evaluated and if needed updated The MSDP will be used as a guide to develop the measurement system for the real time analysis of the internal logistics Not every step of the process is useful for this thesis so some steps will not be discussed in detail 15 Step 2 Define what we do Step 3 Define what we must excel at p Define how we know if Step 1 Define the need for 4
137. ts all the errors and their details and is connected to the bottom left bar chart and the bottom right error map The bar chart is used to visualize trends during the day All the errors that are listed in the Error Event Manager contribute to this bar chart The lost time per hour is displayed to indicate the time of the errors and their duration The error map in the bottom right corner is a map of the plant layout on which every error is marked By marking the location of each error certain trends can be visualized e g if too many errors occur in one certain area there might be a problem with the infrastructure there 44 5 2 7 Dashboard navigation The navigation through the different screens of the dashboard can be done by the buttons that are added in the screens The connection between the four screens is displayed below Summary view Legend orent Status o 1 Visibility vw Working Current Location pm Warehouse wana Qa curema orter me me Q Detect Current Time 2 MENHNNEEE m lt E Nbr of orders ens completed y Xx sS Vehicle location in the plant Utilization 7 during the last day p i 85 2 80 97 x amassoy Overall Efficiency during the last day Last Day 95 75 60 Forki 1 Visibility Status Status Occupancy per hour Occupancy nme last hour is Em es umer Spaghettichar
138. u A e20 JO ease unou 3sej au JO ieu In uSeds corgejtenv ezed oo Suppe 1001 jeAesuog snonunuoo nn lqo snonunuo ennefqo snonunuoa ennoafqo snonunuoo annalqo snonunuoo anngalqo snonunuoo annalgo snonunuo xn fqo Snonunuoo ennafqo snonunuoo annoefao snonunuoo ennefqo snonunuoo ennafqo snonunuoo ana qo snonunuos anndalgo snonunuo3 annoefao snonunuoo ennafqo snonunuo xn fqo snonunuoa enndalqo snonunuoa aangalgo e3eq jo d 1 sunoo0 40118 upd e awn ue sumo Joua Bupnos e aun upea 51n220323Jop ue uin tpe3 51n220328j9p ue auim u3e3 AunoH AunoH auanbauy ssa 10 AunoH payajdwios s JopJo ue oun Arona poie duio si Jopao ue oun Aang paiojduio si sapo ue aun uaa Z 40 puooas Aang japjo Asang Z 40 puosas rang paiojduio2 si apo ue oun Aang paiajduio si apJo ue aun yeg snoy Asana puooas ia3 puoaas A1ang Aouanbaa4 jeAenuod ajalyan olduu onsiBo upe Jo AjAnonpoud ayy zllenSA apatyen olduu 3nsI8o yaea Jo AyANINposd eux ezijen wy uonup paxaid ssapso jo JAN anou p 321d siapio JO JAN wy payaid suapso any payaid ssap o 4 S3nsI8o jeuJa1ui u1 Jo AjAnonpoud eu eseaa2u easy B2UEWO ag Ady sjuiod Suipeojun pue Suipeo au 3noqe uoisnjuoo japuo Jeap ou suaddey axjerstw AYM mous anoo0 Aet asaym s sejd ayy pue suiejqoud ozijensin papaau si3u
139. uld also like to thank the people at HoWest that helped executing the tests and took the time to explain the used software Here would also like to thank Tim Bouttelgier for controlling the train and making the tests possible would also like to say thanks to my family and friends for supporting me this year and all the years before and for offering the needed friendship guidance and amusement Last want to thank my girlfriend for her patience and support during some of the busiest and most stressful weeks of this year Simon De Buyser De auteur en de promotoren geven de toelating deze masterproef voor consultatie beschikbaar te stellen en delen van de masterproef te kopi ren voor persoonlijk gebruik Elk ander gebruik valt onder de beperkingen van het auteursrecht in het bijzonder met betrekking tot de verplichting de bron uitdrukkelijk te vermelden bij het aanhalen van resultaten uit deze masterproef The author and the promoters give the permission make this master dissertation available for consultation and to copy parts of this master dissertation for personal use In case of any other use the limitations of the copyright have to be respected in particular with regard to the obligation to state expressly the source when quoting results from this master dissertation Gent juni 2011 The Promoters The Author Prof dr ir H Van Landeghem Arkan Simon De Buyser Overview Dynamic analysis of in
140. update can remain the same as in the test setup When a forklift is monitored the status update will need to be adapted to this 78 9 Conclusions The purpose of this thesis was to develop new dynamic analysis methods for the in plant logistics based upon RFID data In particular the performance of the in plant logistics vehicles was examined by equipping them with RFID tags and monitoring their movement through the plant or warehouse In first instance a literature study was done to examine similar cases where RFID technology was used to monitor or improve the performance of in plant logistics vehicles Then a performance measurement system was developed by using the Measurement System Development Process MSDP In this process 19 metrics were eventually designed which can be grouped in four Key Performance Areas KPAs Visibility Efficiency Reliability and Productivity The needed functions and formulas for the metrics were described and some presentation possibilities were proposed With the designed metrics a multi screen dashboard was developed This dashboard presents the important information about the performance of the logistics vehicles on a clear way so the user can quickly assess the situation and act upon it To verify the working of the designed metrics a test was executed The test setup consisted of a toy train riding on a track The train had an RFID tag attached to it and was monitored by four RFID readers that
141. urrent order and progress in the order KPA Efficiency 4 Overall efficiency Actual total time per order expected time per order 5 Transportation efficiency Expected transportation time Actual transportation time 6 Average speed Average speed while status Driving 7 Route efficiency Length of the shortest possible route Length of the actual route 8 Loading efficiency Standard Loading Time Actual Loading Time 9 Unloading efficiency Standard Unloading Time Actual Unloading Time 10 Load Unload time variance Loading time Unloading time 11 Overall Utilization Work time Total time 12 Percentage loaded Nbr of kms driven while loaded total nbr of driven Kms KPA Reliability 13 Lost time percentage Total lost Time Total time 14 Vehicle reliability 15 Mean time to repair 16 Route reliability 17 Picking reliability Lost time due to Defects Total time Average duration of a defect Lost time due to going Off track Total time Lost time due to wrong deliveries Total time KPA Productivity 18 orders picked hour 19 orders picked Km Nbr of orders picked hour Nbr of orders picked Km 3 1 5 Step 5 Implement the MS After the metrics have been chosen an implementation plan needs to be made Here the portrayal tools and data collection tools are further developed and put in place These tools were already designed or chosen during the previous step and then entered into the Metri
142. were placed around the track The measured RFID data was cleaned and put in the correct format Together with a made up order list and the needed information about the layout of the train track the RFID data was inserted into a specially designed test dashboard In the test the accuracy and correctness of the metrics were examined The results that were presented in the test dashboard and in the performance report were compared to the real situation From this test it could be concluded that the used metrics and the test dashboard give a good representation of the actual performance of the vehicle A second test was also done to examine the real time aspect of the dynamic analysis methods In this test the update rate of the test dashboard was varied and the performance of each update rate was inspected This test showed that a maximum update rate of 1 update every 2 seconds could be obtained while the dashboard still functions correctly It also proved that the needed time for the update due to the functions and calculations that need to be done is directly proportional to the chosen update rate or the time that has passed since the last update From the tests it can be concluded that the designed metrics and their presentation in the dashboard can offer a real time view of the performance of the logistics vehicles Finally some remarks can be made about the application of these metrics and the designed test dashboard on real li
143. y These metrics are displayed in the screen The transportation efficiency is displayed in a graph that presents the travelled distance against the time In this graph not only the actual travelled distance is shown blue line but also the ideal situation red line This ideal situation occurs when the vehicle takes the shortest possible path and drives at the ideal average speed This graph is updated every second so that the progress of the transportation can be seen Based on the position of the blue line actual distance time against the position of the red line ideal situation some conclusions can be drawn To better explain this the possible positions are displayed in the following figures o e 1 E 8 s s a a E N 0 Por a9 d d d d SN YW YM DP a o w oF OU OP P P AD AP 0D D 0D AD Do OU X oe SS SSS SS SS SS SSS od dv nr AP a DU ab ob 5 aD D D D a Sos er s SST S WR qq uu uq gw Time Time Figure 24 Transportation efficiency Situation A left and B right When the blue line lies above the red line situation A it means that the vehicle is driving faster than the ideal average speed and should reach its destination faster If the blue line lies underneath the red line situation B the vehicle is driving slower than the ideal average speed and will need more time to complete the transportation 42 er a d gd gd d o OG nr YO DP oM Ob oP AP APH AD aD o aD oD nD ee S
144. y the current time timeframe The summary can thus be divided into two parts The upper part of the screen displays the current information about the vehicles current status current location current order etc The lower part of the screen displays the total efficiency utilization and lost time over the entire day The current status of the vehicle is presented in the form of a green working yellow idle or red defect button This does not yet describe the specific status of the vehicle driving loading or unloading are all presented as a green button but does offer a clear and simple view on what the vehicle is doing The location of the vehicle is presented in two ways First of all the name of the current area is written underneath the status This already gives an indication of where the vehicle is active Next to this area of location also a map with the exact location and the direction of the vehicle is presented The direction of the vehicle is displayed because it can offer extra information about the current task e g if a vehicle is standing between racks and is oriented towards a rack it would be performing a loading or unloading action Further the current order is displayed as well as the progress in that order and in the order list The total utilization and efficiency are displayed in gauge meters The colour scale on the meter indicates if the utilization or efficiency is good green low yellow

Download Pdf Manuals

Related Search

Related Contents

取扱説明書( PDF: 1MB )  English - Fiery Help documents    

Copyright © All rights reserved. Failed to retrieve file