casting design with help of information fusion

Proceedings of the 2001 IEEE Systems, Man, and Cybernetics ConferenceCopyright 2001

Casting Design with Help of Information Fusion

Sylke Krötzsch1, Ines Hofmann2, Georg Paul1

1 Institute of Technical and Business Information Systems, Otto-von-Guericke-UniversityMagdeburg, Universitätsplatz 2, 39106 Magdeburg, Germany

2 Institute of Manufacturing Technology and Quality Management Otto-von-Guericke-University Magdeburg, Universitätsplatz 2, 39106 Magdeburg, Germany

Abstract

The computer aided support of several tasks isthe main focus in a lot of companies. Especially thework scheduling and the casting design in foundriesis an important area which is responsible for theeconomical survival on the world market. Thispaper contains proposals for a better support of thephases of the work scheduling and the castingdesign with help of modern computer techniques.At first a review of the research group “InformationFusion“ is given. Thereafter several aspects of thework scheduling are explained, like the integratedprocess of the workout of offers and scheduling ofthe manufacturing, the in- and outgoing conditionsat the workout of offers, checklists and databasesfor technical checks etc. In order to manage thismentioned main points of the work scheduling, aControl-System, like Workflow ManagementSystems, is introduced. The paper presents varioussolutions in order to support the workout of offers.

KeywordsComputer Aided Engineering, Offers for Castings,Information Fusion, Workflow Management

1 Introduction

Technologies, like information processing andcommunication, are currently developed very quicklyworldwide. The casting industry (e.g. automobileindustry or machine building industry) uses the newdevelopments with big success. Examples are CAD-Systems for casting construction and systems forstress and solidification simulations [1, 2]. The 50%quota of CAD-data in die casting plants is muchhigher than in steel and iron foundries. Here a quotaof 10% was ascertained [3, 4]. An analyse of thedevelopment in foundries results in the fact, that thereis no complex system which is able to integrateprogressive technologies for information processingand communication oriented on requirements of thecasting industry. In order to ensure a permanentcoexistence of the casting industry in ensemble withthe most efficient machine-building and equipment-building industry, an adaption to the progressivedevelopments is necessary. The continuous use of the

information and communication technology in allsections of foundries is the future goal. In order toachieve the goal, a scientific penetration of thecompany own technological process is necessary.

Presently one mainpoint of the DFG-ResearchGroup “Workbench for Information Fusion“ is thecasting design process.

2 Information Fusion and Work Scheduling

The research group “Workbench for InformationFusion“ combines the work of several part projects,in order to develop new techniques, concepts andmethods for produce information in a better quality.There are six part projects which develop conceptsfor the workbench. Another three part projectsestablish the applicaion areas. “Casting Technoloy“ isone application field, which analyzes the castingdesign process with the aim to get information in anew quality with new concepts and techniques ofcomputer science. Here some databases with foundryspecific data and other auxiliary means are used.During the processes of work scheduling the staffused the available databases directly. That means theemployee get only the stored information of aspecific database. With help of the concepts andtechniques of the workbench it is tryed to getinformation in a new high quality. This is possiblebecause all databases are used for decision makingand the workbench filter the relevant information forspecific tasks. An overview of the “Workbench ofInformation Fusion” is given in fig. 1. Theapplication profile “Casting Technology“ works witha lot of different data. For instance, normaldocuments (e.g. norms and pictures), CAD-Data, anddifferent databases are used during the casting designprocess.

These data support the decisions of the employeeduring the single worksteps. The data are distributedin heterogenous databases, technical manuals, or asexpert knowledge. Especially the last point is a veryimportant fact in the casting design process. Theexpert knowledge is available in a fuzzy kind and it isvery difficult to prepare the knowlegde for acomputer aided processing. Almost the completeknowledge of a company is normally combined at the


work scheduling (in extreme manner). For instance,information to economical questions, questions aboutmaterial, logistic, operations, and equipment, as wellas questions to the quality assurance, qualifications ofthe worker, and so on are accumulate in foundries.

Application ProfileFoundry Technology

Local DB Extern DB WWW

Toos for KnowledgeAcquisition

Federation and Integration Services

Visualizationof Data

Mining and Graphic

Interaction

Active Learning

MoreFusion

Modules

Casting Design Process

Casting Defect DB

SimDat DB Citim DB CAD-DataNorms

Pictures

Workbench

Fig. 1: Workbench for Information Fusion

An important part of the casting design process isthe work scheduling in foundries. Already here thefirst decisions about the casting are made. Normallythe work scheduling is executed in two steps. Theprocess begins with Phase 2 (fig. 2), the workout ofan exact offer, after an inquiry by a customer isreceived. In case of getting an order by the customerPhase 3 (fig. 2), the elaboration of the workdocuments, is executed. The steps of workout ofoffers and elaboration of work documents differ fromeach other in the level of detail. In context of thecasting design process the work scheduling isexpanded. This means, there is one phase more(Phase 1), the workout of a rough offer. Fig. 2 showsthe necessary work steps and clarifies the connectionbetween the first two phases. The new strategy startswith Phase 1 (fig. 2), if the foundry would likeparticipate in the product development processtogether with the customer. The decision must bedeliberate by the foundry because the productdevelopment needs time and causes costs. But on thisway the company tryed to get a good basic positionin order to get the order. The rough offercomprehends the necessary product developmentinformation for a specific casting. If the foundry getthe offer for product development an exact offer iscreate (Phase 2). The main focus is the cooperationwith the customer during the casting design process.On this way it is possible to consider all necessaryrequirements for casting technical view. Phase 3 isuncoupled from the workout of offers.

The practical experiences of the executed analysesin range of expedient Workflow ManagementSystems in foundries confirm the necessary insert of

new techniques [5, 8]. Both reliable company-intern(e.g. information flows between workout of offer,marketing, and manufacturing) and company-extern(e.g. information flows between engineering offices,customer, cooperation partner, etc.) informationflows are required, which are supported andwarranted by adequate multimedia techniques [6].The goal of information processing systems is thestorage and processing of different types ofinformation in only one system, like acoustic andvisual information. This is required because thequantity and quality of information is relates topermanently changes by the simultaneouscooccurrence of heterogeneous information.Furthermore a use of these information at any time byplanning engineers and by other competentemployees is required.

Phase 1:

workout of rough offer

Phase 2:

workout of exact offer

Phase 3:

elaboration of work documents

inquiry of the

customer

rough offer

offer for product

development final offer

order

releasedocuments

to production

Fig. 2: Phases of Work Scheduling in Foundries

Such new approach requires a demanding datamanagement [7] and a high qualification anddiscipline oriented on the warranty of the actuality ofdata. All modifications have to be gathered andstored consequently by authorized staff. The majorhurdle on this way is the insufficient scientificpenetration of the technological processes infoundries (e.g. the control of the cavity fill process, inparticular on complicated castings) and the economicevaluation (e.g. ascertainment of the limits ofeconomic manufacturing with available resources). Inthe following the main results of the analyses of workscheduling in foundries are introduced. They areoriented on contents of the research project “CastingTechnology“.

3 Reference Processes of the Work Scheduling

The precondition to a successfull existence of acompany on the world market are qualified offers,which are elaborated in a short term. Interviews infoundries results in an increasing number of deliveredoffers in the last years (35%). In the same time thesuccess rate came under 30%. The main workstepsfor workout of offers are “register the inquiry”, “testand rate the inquiry”, “set pre-definitions”, “calculateinquiry”, “calculate extern inquiry” and “create


offer”. These steps are executed more or lessdetailled in the companies dependent on inquiryvalence and the type of customer [13]. In case ofsmall inquiry volume no generation of a raw productis executed and the data of similar calculationdocuments are used. These documents are sorted insimilar and already produced castings. In foundriesauxiliary means for the workout of offers are used,like guiding principles, standards, several methodsfor checking the manufacturing oriented design,nomograms, casting classifications, regressionequations for ascertainment of allowed times,material sheets and so on. Very often offers ofalready produced castings are used, which are storedin file form or they are managed in specificproduction planning and control systems which arebased on concepts like Material Resource Planning(MRP I) and Manufacturing Resource Planning(MRP II). In order to support the work scheduling, atthe Otto-von-Guericke-University Magdeburg incollaboration with a well-known foundry a conceptfor an efficient computer-aided solution wasdeveloped [8].

The analyses during the research project containsthe development of auxiliary means [8, 11]. Theseare checklists for completeness check andverification of offers, the database “check theproducibility of inquiries“, and the algorithm for“generation the raw product“. These components areintroduced in the following.

documentsand reg-no

checkcompleteness

rejection

completedocuments

incompletedocuments

complementeddocuments

rating ofinquiry

classification

complementdocuments

check techn.feasability

ok

checkproducibility tested

documents

notproducible

casting

executechanges

changedcasting

K1

K2

K1 or K2K3

K4

K3 or K4c c

Fig. 3: Subsequence “test and rate the inquiry”

The process step “test and rate the inquiry” issubdivided in a subsequence (fig. 3). In order tomodel the identified process steps FunSoft-Nets areused [12]. Here all necessary steps for testing theinquiry are executed. The workout of offers startsnormally with a completeness check of the inquirydocuments. In order to represent the efficiency andthe service of the foundry, the goal is the reduction ofconsultations with the customer. In collaboration witha well-known foundry a checklist was developedwhich completes missing information in dialoguewith the employee.

The missing information are to differ in necessaryand desired characteristics. Necessary characteristics

influence the technology of castings, themanufacturing costs, and the quality. An elaborationof desired characteristics with help of necessarycharacteristics is possible. In the following anexample is introduced. The manufacturing means forcasting production (manual or automated) depends onthe piece number. Is the piece number not specifiedby the customer (5% of the inquiries) the foundrysubmit an alternative offer, provided that severalmechanized and automated form plans are availablein the foundry. The proposal of automated moldplants is dependent on the minimal economic piecenumber (manual mold = piece number of 1). Not ineach case the foundry elaborate an offer. It is alsopossible to eliminate an inquiry before the processstep “check completeness“ is executed. Reasons forthis can be a desiderative reliability of the customer,a small volume of trade, a small piece number (e.g. anew production line for the product is not advisable),and a cognizable compare of inquiries (e.g. thecustomer gets normally his products from anothersupplier).

Another very important point during the workoutof offers is the weighting of the inquiries. In thecontext of the research group “Workbench forInformation Fusion“ a rule-based program isdeveloped, which is adapted for a weighting of theinquiries in foundries. With help of the checklistseveral points were testet. Based on the tests theinquiry will evaluated and some solutions forpossible cost calculations are proposed. The points ofthe checklist are divided in fundamental, technicaland economical criteria. The evaluation of thefundamental criteria comprise the common relations,for instance customer-supplier-relations and therelations to the superior parent company. Thetechnical criteria are related to available technologies,limits for technical feasibility, economical lot sizesand experiences with similar parts. Economicalcriteria verify the reliability of the customer, requiredprice proposals of the customer etc.

Fig.4: Weighting of Inquiries in Foundries


Dependend on the results of the test, the programgive possible kinds of cost calculations (fig. 4). Thisare rough calculation, normal calculation, projectcalculation, alternative offers and disaffirmation.Some proposals of the program have supplements, sothat in some cases the agreement of the managementis required. Dependent on the result the nextworksteps are executed. The program based on ruleswhich are subdivided in superior and normal rules.An example for a superior rule is “Is it a comparativeinquiry THEN create a rough calculation“. The rulesare represented as decision tables. Tab. 1 shows anexample. There are 2 rules and 4 alternativedecisions. The complete checklist comprehends 10criteria. The consideration of all rule-combinationsleads to 210 possibilities. But not all rule-combinations are expedient and redundancies arepossible.

Tab. 1: Decision Table (Y- condition filled, N-condition not filled, X-action part will be executed)

R R R RB Technical Feasibility Y N Y NB Comparison Inquiry Y Y N NA Quick Calculation XA Normal Calculation XA Alternative Offer XA Rejection X

One part of the Research Group “Workbench forInformation Fusion“ comprises also analyses aboutinvolvement of the results. That means, the results ofthe rule evaluation will be considered in furthermoreweightings of future inquiries.

In exceptional case the documentation of theavailable equipment and conditions of use areutilized. In order to develop a software for check thetechnical feasibility the analysis comprehends thestructure of the equipment, the flow path of castingsthrough the single production units of the foundry,and the influence coefficients to choose from theequipment. Because of these facts the followingstatements are mentioned:

• Each company posses of a special structure of

equipment oriented on type and number of theequipments. Very often there are severaltechnological equipments (e.g. melting furnaces)for the same technological tasks (e.g. melting),which differ from each other in respect ofcapacitive and qualitative aspects.

• There is a special flow path through the foundrywhich depends on several constraints. Dependenton the forced-path the equipment is to choose.

• There are numerous influence coefficients, whichrestrict the choosing of the equipment. That aretechnical criteria (e.g. melting tools for specialmaterials), organizational criteria (e.g. deliverydate and available capacities), and economicalcriteria (e.g. piece number and quantity).

• A complete overview about the availableequipment in the company and the differenciatedconditions of use must be exist.

Within the scope of the investigations allsubstantially influence coefficients for the technicalfeasibility was identified. Based on these facts adatabase was designed with the goal to makeanalyses which are oriented on the decision processfor the equipment. Information about conditions ofuse depend on the influence coefficients areevaluated, too. It is possible to insert new data or todelete data of old equipments which are sorted out.As a result of the elaborated algorithm the availableequipments are showed which can used for producinga special casting. The result depends on castingcharacteristics.

4 Computer-aided Generation of Raw Products

A foundry generate a raw product if no documentsfor similar or equal castings are available and thevolume of trade is large. The generation of the rawproduct contains a scheme of allocation of sheets andthe draft of the raw casting (e.g. 3D-Data). The goalis an exact predefinition of manufacturing costs [9,10]. This contains a derivation of points, likeposition, number and shape of joint lines, of cores,and of the gating and feeding systems, allowancesand drafts.

In contrast to ancillary foundry industry for motorindustry, which use computer-aided construction(CAD-Systems, like CATIA, ProEngineer,UNIGRAPHIX, etc.), the small foundries work witha manual generation of raw products which based onexpert knowledge, like standards, guidelineprinciples, and some other methods and equations fordetermining of gating and feeding. Most expertknowledge oriented on the singularity of the foundryis normally not documented. The results of theresearches shall support the small foundries with helpof a partially realized solution. In preparation thealgorithm for generation of raw products allnecessary molding processes and the particularitiesare analyzed. Automated molding processes havemany constraints, e.g. default mold joints of themolding boxes (horizontal, vertical) and the defineddimensions. In contrast to the automated moldingprocess the manual molding process has nearly nolimits. At the use of different casting alloys, e.g. cast


iron and cast steel, the gating and feeding of thecasting is to consider. Other than the demarcation ofthe molding processes another task was the analysisof possible theoretical schemes of allocation ofsheets. Dependent on the approach of the employee 6basis schemes was identified. The basis schemes areconform to the 6 views of casting. Special variantsare possible, e.g. angular ram up of the casting.Because of the constraints an algorithm for aderivation of a raw product, including the position forram up, for a Seiatsu molding maschine wasdeveloped. During the execution of the algorithm theseveral positions for ram up the casting areeliminated because of illegitimate encores of material(allowance, drafts, feeder) on particular casting parts.The new strategy with help of the workbenchcontains the use of several databases. Each processstep uses special infomation of different databases.

technicalfeasibility

producibility

core molding

generation of raw products

2D/3D CAD-Data

casting geometry

test of technicalfeasibility

DB TR

description of castingdefects

DB Defects

foundry technology

DB Citim

materials

Module

molding processes

Module

design rules

Module

standard values

Module

Fig. 5: Databases for Casting Design Process

With help of the databases the casting designprocess is supported so much as possible. Fig. 5showes the connections between process steps anddatabases. It is possible that the employee needsinformation of several databases for one task. Inorder to support this, the techniques of the workbenchare used. During the check of producibility an accessto the databases DB Defects, DB CITIM – materials,DB CITIM – molding processes, and DB CITIM –design rules is necessary. These information arefiltered by the workbench. The arising new data helpsthe employer by the execution of the single processsteps.

5 Workflow Management and Work Scheduling

During the process of elaborating documents foroffers and manufacturing for castings a lot ofdifferent information (geometry, material, process

engineering, quality, deadline) have to be considered.The data shall be effective to gather, to archive, andto systematize. Thereby a cooperation of severalsoftware systems used in foundries is necessary.Under real conditions the documents for a castinghave to pass through several departments (workscheduling, construction, sale, manufacturing, ...),which are often locally separated. Different foundriesuse different concepts for the development ofcastings. That means, there are different informationand documents in different foundries and very oftenthe documents go bidirectionally between twosections. The consequence of exchanging documentsmultiple times is a loss of information and time. It isascertained, that the reason for the loss of informationand time are the idle periods of the documents in thesingle sections. In order to prevent the loss ofinformation and time a control system (e.g.Workflow Management System) for the workout ofoffers is suggested.

In the phase of workout an offer in foundries aControl-System has several tasks, like control of theidentified processes, management and archiving ofdocuments of the offer (2D/3D-Data, conventionaldrafts, pictures, motions, ...), management oftechnological data (material, requirements, tasks ofthe casting, ...), and elaboration of post calculations(used for offers of similar or equal parts). Thereduction of product development times, the reactionon the increasing product variety, a better quality ofthe products, and the adherence to a time limit (e.g.date of delivery) are requirements for the WorkflowManagement Systems. These points are conform withthe requirements of the foundry. The use of modernworkflow technology has several advantages. Theemployees are disburded by routine work and so theproductivity increases and the pass-through times andthe costs decrease. An improvement of quality ispossible through an integration of services andquality-saving measures. A flexible reaction ondesires of the customers and an effective monitoringof the sequences can be obtained, in order to analyzeand solve problems very quickly with the goal toincrease the satisfaction of the customer. Theimportant decision processes in a company areaccelarated. The reasons are the availability of thecurrent data, the optimized or reorganized processsequences of the company, and the use of anautomated control.

The base for the development of a control is ametamodell, which describe all necessary data. In theapplication field “Casting Technology“ exist twokinds of sequences. There are pre-modelledsequences, which are offered by the system. Thatmeans, these pre-defined sequences containsactivities which are repeated executed during the


casting design process. A sequence, a pre-modelledsequence too, consist of processes. A process can bea sequence again or a process step. A process step isthe smallest atomar unit which can be assigned to anactivity. It is also possible to assign a refined processsequence instead an activity [11]. Figure 6 showesthe metamodell.

The Workflow Management in the application fielddistinguish two layers, the technical/technologicaland the information technological layer. Thetechnical layer is used in order to model processesand sequences of the application field. Theinformation technical layer consist of activities forinformation technical view.

Process

ProcessStep StartProcess EndProcess Sequence

Activity

Role

User

Resource

Document Tool

idname

description

idname

id

idname

run_activity()

usedocument()

run_tool(Tool_ID)open_doc(Tool_ID, Doc_ID)

idname

getlayer()

predecessor

successor

*

*

1..* 1..*

*

*

3..*

1

pre_layer suc_layer

idname

Figure 6: Metamodell for Sequences

A control system, which support especially theapplication field “Casting Technology“ is developed.Based on this control system the techniques of theworkbench are used. These concepts and techniquesare assigned to special processes and offered forinstance as activities.

6 Conclusion

A multitude of analyses regarding efficient designof the workout of offers are executed. The aim is thereduction of the process time by increased quality.Some selected work steps of the workout of offerswere profoundly analyzed. Contents of future workswould be check the generality of the results incomparable companies. The paper contains anoverview of several developed auxiliary means forthe application field “Casting Technology” and is tounderstand as a case study. Furthermore workscomprehends the development of methods andtechniques, in order to support the InformationFusion from application side.

7 References

[1] Sirilertworakul, N.; Webster, P.; Dean, T.A.: Asoftware package for the design and production ofcastings. 60. Gießerei-Weltkongreß, The Hague1993.[2] Schmidt, D.; Law, T.: The benefits of PC-basedsolidification modelling of castings. Foundry TradeJournal, Band 167 (1993) H. 3479, p. 406 - 408.[3] Ambos, E.; Richter, U.; Soethe, M.; Salm, Th.:Computer Aided Construction and Production ofModels and Forms. Giesserei 79 (1992) 10,p. 395 - 401. (in german)[4] Grabowski, H.: Search System forInformationfinding. VDI-Berichte Nr. 13 (647),1987. (in german)[5] Krötzsch, S.; Hofmann, I:Process Control in theWork Scheduling. 2nd ICSC Symposium onEngineering of Intelligent Systems, Paisley,Scotlangd, 27.-30.06. 200[6] Ambos, E.; Brahmann, M.; du Maire, E.; Helm,B.: Multimedia and Data Communication forAccerelation and Qualification of the Developmentand Production Process for Castings. Giesserei 84(1997) 19, p. 27 - 31. (in german)[7] Mucksch, H.; Behme, W.: The Data-Warehouse-Concept. Gabler Verlag, Wiesbaden 1998. (ingerman)[8] Krötzsch, S.: Workout of Offers in Foundrieswith Help of a Structure Dependent Control.Disseration (eingereicht April 2001), Otto-von-Guericke-Universität Magdeburg, 2001. (in german)[9] Scheler, R.; Hofmann, I.; Krötzsch, S; Ambos,E.; Pfisterer W.: Computer Aided Search of SimilarCastings for the Workout of Offers. Giesserei,87 (2000) 8, p. 72 - 78. (in german)[10] Hofmann, I; Krötzsch, S.; Scheler, R.; Ambos,E.; Miersch, N.; Pfisterer, W.: Effectivity through theInsert of Computer Aided Solutions in the WorkScheduling. Giesserei, 87 (2000) 10, p. 39 - 45. (ingerman)[11] Scholz-Reiter, B.; Stahlmann, H.-D.; Nethe, A.:Process Modelling. Springer Verlag, 1999.[12] Gruhn, V.: Validation and Verification ofSoftware Process Models. Dissertation, UniversitätDortmund, 1991.[13] S. Krötzsch, I. Hofmann, G. Paul: Using ofProcess Modelling in the Work Scheduling. 12thInternational Conference on Computer Applicationsin Industry and Engineering (CAINE-99), Atlanta,Georgia USA, November 4-6, 1999.

casting design with help of information fusion

Documents