J Intell Manuf (2007) 18:467474 DOI 10.

1007/s10845-007-0049-1

Knowledge based manufacturing system (KBMS)

Published online: July 2007 Springer Science+Business Media, LLC 2007

Abstract Production management, in batch type manufacturing environment, is regarded by the current research community as a very complex task. This paper claims that the complexity is a result of the system approach where management performance relies on decisions made at a too early stage in the manufacturing process. Decisions are made and stored in company databases by engineers who are neither economists nor production planners experts. This paper presents a new method where engineers task is not to make decisions but rather to prepare a knowledge-based road map. The road map method does introduce exibility and dynamics in the manufacturing process and thus simplies the decision making process in production planning. Each user will generate a routine that meets his/her needs at the time of needs by using KBMS CAPP. Thereby this method increases dramatically manufacturing efciency. Keywords Manufacturing Scheduling Capacity planning Shop oor control

Introduction A typical characteristic of the manufacturing process is dynamic, where conditions are constantly changing and decisions have to be made within a short space in time. It
G. Halevi (B) Hal-Teck, 20a Dobnov St., Tel-Aviv, 64369, Israel e-mail: [email protected] K. Wang Department of Production and Quality Engineering, NTNU, 7491, Trondheim, Norway e-mail: [email protected]

is often preferable to make a decision at the right moment rather than to seek the optimum decision without any time limit. The better we have the available relevant data at the right time, the better decision we can reach. A computer is a tool that can be employed to narrow the gap between the conictions demands of "time" and "decision". A computer system can store and manipulate large quantity of data in a short period of time. Therefore the computer was accepted enthusiastically by industry as a data processing equipment. Most computerized applications use the methods and algorithms which are developed in a manual way. But performing them are usually more efcient and quicker. The real change in thinking and procedure in the manufacturing cycle was the MRPMaterial Requiring Planning/Material Resource Planning. The objective of MRP is to plan the activities to be performed in order to meet the goals of the master production schedule. The logic and mathematics which MRP is based on is very simple. The gross requirements of the end product for each specic order is compared against on-hand and on-order quantities and then offset by the lead-time to generate information as to when assembly should be started. All items or subassemblies required for the assembly should be available on that date, in the required quantity. Thus, the above computation establishes the gross requirements for the lower level items. The same computation is repeated level by level throughout the entire product structure. MRP (Liu & MacCarthy, 1996) represents an integrated communication and decision support system that supports management in a total manufacturing business. The use of optimizing techniques drawn from operation research and management science was adapted by MRP. One signicant reason why MRP adapted the technique was that it made use of the computers ability to centrally store and provide access to the large body of information that seemed necessary

123

468

J Intell Manuf (2007) 18:467474

to run a company. It helped coordinate the activities of various functions in the manufacturing rm such as engineering, production and materials, purchasing, inventory, etc. The attraction of MRP lays not only in its role as decisionmaking support, but more important in its integrative role within the manufacturing organization. Several areas of the plan are affected by MRP: production planning and scheduling, quality control, accounting, rst line supervision, production labor, and the use to develop plant-wide scheduling. A closed total working system can be based on MRP, starting with customer order, following by purchasing and subcontractors order, shop oor scheduling and control, inventory management and control, bookkeeping and accounting. All are dependent activities such that each one is validated by status of the manufacturing system. Engineering (design and process planning) processes are stand-alone activities that depend on the skill, creativity, imagination and knowledge of the engineer. Although they could be neither economists nor production planners experts, their activities (decisions) establishes the company level of competition. This paper presents a method where engineers task is not to make decisions but to prepare a knowledge-based road map. Each user will generate a routine that meets his/her needs at the time of need by using KBMS CAPP. Thereby this method increases dramatically manufacturing efciency. The paper is organized as the following: "Integrated MRPIMRP" describes an integrated MRP systemsIMRP; Manufacturing data collection is presented in "Manufacturing data"; A concepts of knowledge based manufacturingKBMS are introduced in "Knowledge based manufacturingKBMS"; "Examples of using the road map" demonstrate how the road map method can be implemented to solve manufacturing problems with some examples.

The logic of the inventory system is that issue and receipt are all predictable. No components are received unless they were ordered, no items are issue unless they serve a specic order, or are needed for assembly or processing of a specic order. No shop oor activity is performed unless it was scheduled by IMRP. No payment is done unless purchasing order was accepted and the payment terms are met. All activities validate one another and action decisions are made automatically without manual interversion. It sounds as a perfect system, but unfortunately, IMRP was not a success story. Probably because it was introduced before the IT technology and the datacollection systems were available to support its requirements. Therefore manufacturing turned to other venues. One philosophy believes that production planning and control is very complex, therefore the only way to make such a system effective is to simplify it (Halevi, 1999). This trend proposed systems such as: Just in Time, Lean manufacturing, Agile manufacturing, kanabn, Kaizen, Group Technology and cellular manufacturing. Another philosophy believes that production planning and control is very complex; therefore, there is no chance of developing a system that will solve production problems. Hence, the role of computers should be limited to supplying data while the decisions should be made by people (Fitzgerald, 2000). Several excellent systems are available. However these manufacturing philosophies do not pretend to be integrated systems. As people make decisions, they are the decision making responsibility, but. It is not the SYSTEM fault if a disruption occurs.

Manufacturing data No matter what manufacturing strategy is employed it is based on data stored in company computer databases. Databases contains many les, such as: Order le, Product structure, Routing le, Inventory le, Resource le, Production feedback data le. Data creation and storage is the responsibility of the appropriate discipline. For example, product structure le lists all items composing the product and the links between them. This data is created and stored by engineering department. Routing le species how each item and assembly is to be produced, created and stored by process planning department. Each of the appropriate departments will probably consider many options and optimize its decision before introducing the data to the company database. The data in those les is considered, by its various users, as a reliable, accurate and being what is stated beyond any possibility of doubt. It assumes that it is based on optimum values and validated in the process of inputting the data to the les.

Integrated MRPIMRP Integrated MRP is a working strategy where the computer is a working tool and not merely a data retrieval system (Halevi, 1980). It is an innovative strategy that covers all manufacturing stages to one comprehensive unique system. The logic of the system is that there is an objective, i.e. customer order that must be delivered. All activities of the manufacturing cycle are predicted and focused to meet that objective in the most efcient way and they can be performed automatically by using company databases, without human interventions. IMRP prescribes what should be purchased, what should be subcontracted and what should be processed inshop. Using company databases specications for purchasing the item on the list is prepared and sent to preferred suppliers for RFQ. Based on algorithm and suppliers rating data a decision is made.

123

J Intell Manuf (2007) 18:467474

469

Theoretical production planning and scheduling is actually very simple task. The plant gets orders which dene the product, the quantity and delivery dates. The resources of the plants are known, the product bill of material is known. The task of production scheduling is to make sure that the orders will be ready on time, thats all. IMRP and other manufacturing strategy can generate a good operation list. The complexity appears when disruptions encounters, disruptions in the planning stage and in shop oor processing. In many cases disruptions appears due to the nature of the manufacturing data in the company databases. Although the data is the optimum from the discipline that generated it, it must be remember that it is optimized by criteria of optimization of that of the specic discipline. Each discipline has different optimization criteria. For example: the design criteria are not the same as that of processing, and not that of shop oor control. The criteria of sales are not the same as those of the production management. The probability of having overall company optimization by using individual discipline optimization is very low and almost impossible. Especially when the optimization is different than any discipline optimization, it has prots. But none of the individual disciplines working with perfect local optimization assures company manufacturing meeting its objectives. Each discipline criteria of optimization must match the objective of the discipline. The main objectives are not negotiable and they must be met. If they are not met the task solution can not be acceptable. However there are several secondary objectives such as cost, weight, appearance, maintainability, environmental issues which is related to multiple-objective optimization solutions, where the optimization criterion is a compromise of all of them (Collette & Siarry, 2003; Wang, Hirpa, Wang, Yuan, & Fang, 2003) Compromise means that there are several alternative solutions and a decision for balancing the solutions should be made by the engineer that was assigned to the job. Furthermore, in process planning, for example, there are three criterion of optimization: minimum cost; maximum production; and maximum prot. The data does not specify which criterion was selected for the recommended routing. Process planning in spite of its importance to the manufacturing cycle is predominantly labor intensive, and it depends on experience, skill and intuition of human labor. Dependence on such methodologies often precludes a thorough analysis and optimization of the process and nearly always results in higher than necessary production cost, delays, errors and non-standardization of processes. There is time interval between the data creation and the data application. Optimized data is introduced into the databank a long time before its uses; therefore it cannot consider the present state of production. That is the main cause that production management becomes a complex task.

In production planning and scheduling resource overload are created and bottlenecks occur and must be resolved. As the data used was the optimum one, the only way to resolve the situation is to decide priority rules or develop a very complex capacity planning program.

Knowledge based manufacturing systemKBMS The complexity is not a must (Halevi, 2004; Tinham, 2000), it occurs due to the strategy taken; a Knowledge based manufacturing strategy will make manufacturing as a simple straight forward system. The objectives of the KBMS are to increase productivity and reduce manufacturing costs. These objectives are the same as those of many other systems. The difference lies in the approach and strategy employed (Guenov, 2002). The Knowledge based manufacturing strategy approach makes use of the following notions: There are innite ways of meeting design objectives. In any design about 75% of the dimensions (geometric shape) are nonfunctional (llers). These dimensions can vary considerably without affecting the design performance. There are innite ways of producing a product. The cost and lead time required to produce a component are functions of the process used. Transfer of knowledge between disciplines working to produces a product, should not be by transferring decisions, but rather by transferring alternatives, ideas, options considered, reasoning etc. Company database should be open and available to all disciplines. These notions are what any of us does in personal life. For example, if you wish to go from point A to point B you study the map and plan the optimum route to reach your destination. This is the present time decision. However, at another time when you have to move the same way, say, at night you might change the route. In winter you probably will look for a route with maximum shelter from the rain. In summer you might choose a route that protects you from the sun. In spring time you might choose a route with nice view. Despite the decision if you run into disruptions, such as a blocked road (bottleneck), red trafc light or with a crowded you might decide that instead of waiting it is better to consult the road map and change the route in order to nd a path with no obstacles. Such change is done at each junction. It might be a longer route but it will be faster. The original decision must not prevent one from adapting the new route. Similar strategy can be applied to production planning and management. The knowledge based manufacturing system

123

470

J Intell Manuf (2007) 18:467474

proposes to supply each manager with a road map which is stored in the company databases, and allow him to deviate from the original processing route while accomplishing the production program assign for a period. Naturally the production program must be practical. The problem is that by present day technology the manufacturing databases includes only decisions (without revealing why and how they were arrived at) and therefore the road map indicates only one path. Naturally the one who makes the decision should consider many options, alternatives, and optimization methods. To our sorrow all such considerations are lost, and merely decisions are transferred and stored in the company database. What should be done is to capture this available data and store it, instead of letting it go to waste. The task of the process planner is to select out of the tremendous number of alternatives and nd the most economical process comparing with those alternatives. The sequence of decisions affects the recommended process and thus can introduce many articial constrains on the selected process planning. However, the process planner is neither an economist nor a production planner expert. Therefore he should not make decisions that are beyond his eld of expertise. His task should be to prepare a road map, covering all feasible routings, and let the expert in production planning decide which route to use, in light of order size and plant load at the moment a decision is needed. A product tree network is used throughout all the planning steps. Figure 1 shows a working product tree for a spring return cylinder and Fig. 2 shows its road map The road map data can be achieved by the process planner; instead of throwing away his scratch book computation, let him write it down in the spread sheet. Or the selected routing is listed in the spread sheet format, and a computer program is used to ll in the operation time data for other resources. It takes a relative simple program to calculate the operation time based on the process planner computations. The time conversion can be based on resource specications (such as

Fig. 2 The road map for a spring return cylinder

Fig. 1 Product tree for a spring return cylinder

power); assembly operations can be converted by ratios of manual operation, machine operation or robotics, etc. The data (road map) can be used in many applications: in resource planning, costing, cash ow planning, prot forecasting, budget and management control, master production plan, capacity planning and shop oor scheduling and control, performance measuring decision support system setting delivery date as a function of cost. Each application uses the data in a method t its needs, but all use one main database. This paper concentrates on production planning applications. Alternative routings can be generated by KBMS CAPP program that transformed process planning from a technological into a mathematics decision. Any user can generate a routing, without the assistance of a professional process planner, by setting his parameters, and the computer program will generate the routing. The denition of the mathematical problem is as follows: Given a road map (Fig. 2) which lists operations to be performed; a list of available resources; and the time to process each operation on each resource. A decision is required as to which resource to use for each operation; in what sequence in order to generate an optimum routing. The operations in the list are not arranged according to any reasonable sequence. However, some operations must precede others or join with others. These constraints are stated in the road map in the PR (priority) column. Extra expenses and time should be added to cover extra when moving from one resource to another. These extra expenses are called penalty. Thus the penalty is for a batch, and is a function of the batch quantity to be produced.

123

J Intell Manuf (2007) 18:467474

471

Naturally, for different quantities and constraint a different routing will result.

Examples of using the road map The process planner recommends the best routing to produce a product. The questions are: what is best process? Is it best from maximum production or for minimum cost? Is it independent of quantity? Is it best to generate prot to the company? To examine those questions we used the road map KBMS CAPP program to generate 22 process alternatives of one item. The alternatives are shown in Fig. 3. As can be seen there is no general best process, it depends on the objection and user of the routing. For sure you cannot leave such a crucial decision to be taken by the process planner. Note: for max production use alt. #1; for min. cost use Alt. #22; for max. prot use alt. #9; for min. investment use alt. # 3; for No. of years to ROI use alt.13. To demonstrate the benets of employing the KBMS method in production scheduling, let examine the throughput of processing a product with the data given in the company road map. The objective of production management stage is to plan the activities on the shop oor in a manner that order delivery date will be met. Thus the planning must consider: the available resources, i.e. plan with nite capacity; the product structure. The use of the product structure is to make sure that all items required for assembly will be available when needed and not before. When an item is delayed, the planning of all other dependent items must be adjusted.

Fig. 4 System architecture for production planning and control

The KBMS planning method is with nite capacity and with the use of the product tree (Halevi, 1980). The detailed capacity plan is transformed into a real schedule for the purpose of job release to shop oor. Shop oor is still free to produce the product mix by any other routing which facilitates the solution of problems caused by disruption, with the restriction that the released product mix, as specied for the period, must be completed in time. The KBMS method of planning uses nite capacity and a product tree. Figure 4 shows the system architecture for production planning and control. The planning steps are as follows: 1. Determine stock allocation priorities. By interactive work with the product tree, a time-based bill of materials is created. 2. Stock allocation. The order with the low-level item in the time-based bill of materials must start at the earliest date and is given the highest priority for stock allocation. The time-based bill of materials is adjusted after each stock allocation, and priority and allocation are repeated until all orders are treated. 3. Adjust quantities by economic considerations. Reconstruct the working bill of materials according to economic lot size, using the KBMS CAPP method for economic order quantity or any other method. 4. Capacity planningmachine loading. The working bill of materials is transferred, refers to the KBMS CAPP for generating a routing, and constructs a time-based working bill of materials. The loading starts with the low-level item with the earliest start date. The loading is done by

Fig. 3 Alternate routing

123

472

J Intell Manuf (2007) 18:467474 Table 1 Comparison of scheduling strategies Criteria of routing optimization Maximum production Minimum cost Semi exible The KPMS strategy Number of periods to process orders 35 32 23 21 Unit cost 162 76.2 131 102

periods. Whenever the period to be loaded is already occupied (by a higher priority item), the road map is called to generate an alternative routing. Whenever it is economical, the alternate routing is used. Loading considers the bill of materials relationship of the items and subassemblies. After each item is loaded, the road map consults the time-based working bill of materials as to the next item to load. 5. Job release for execution. The few early periods in the capacity plan are released to the shop oor for execution, after checking that all auxiliary jobs are ready. The departments that prepare the auxiliary jobs are alerted in time to release the outgoing jobs. 6. Shop oor scheduling and control. Shop oor strategy is that whenever a resource is free, it searches for a free operation. If there is a match the operation is loaded. If the resource is not the optimum for any of the free operations, a look ahead is used to determine the waiting time for a free operation. The waiting time (idle time) is used to compute a new optimum for a free operation, and load it. It is the efciency of being in-efcient (Halevi & Cunha, 2006). The decision is made by scanning the operation raw of the road map (see Fig. 2) and comparing the free resource time to the optimum resource time. This scheduling strategy was
Fig. 5 Load prole for case KBMS planning strategy

tested on: two orders; 12 items, 15 resources. The results are shown in Table 1. Figures 5 and 6 show the two extreme detailed scheduling.

Conclusions Production managements job is to plan and produce products according to managements orders and policy. The traditional method for performing these tasks divides the manufacturing process into several stages. The orders are initially converted to the required manufacturing activities, which dene what should be produced in the shop and what should be purchased or outsourced. This stage is called material requirement planning/material resource planning. Any material or items that are produced outside the company are taken care

123

J Intell Manuf (2007) 18:467474 Fig. 6 Load prole for case maximum production planning strategy

473

of by purchasing. The items that are produced in the shop are handled in the next step: capacity planning. In this step, a long-range plan is made to plan the capacity required at each work center and to help allocate the machines and manpower required to meet production managements goals. The immediate short periods are scheduled in order to determine what jobs should be dispatched to the shop oor for execution in a certain time period. The job release step is the link between planning and execution. Planning and execution, in the traditional method as described above, regard the routing as static and unalterable. Thus, the planning is simple, but it robs the shop of exibility and efciency. The knowledge-based road map method can generate a process plan (routing) in a split second. Therefore a routing that takes the immediate constraints into consideration can be generated. This feature means that routing can be treated as a variable. This capability of the road map can be used to integrate process planning with production management. This paper presents a method for integrating all stages of the manufacturing process (i.e., Material Requirement Planning (MRP), Material Resource Planning (MRPII) and capacity planning) into one dynamic logistics program. A product tree (network) is used throughout all the planning

steps. The routing is dynamic; because it is generated at the instant it is needed. Using this dynamic planning method, lead time becomes much shorter (in the example lead time was reduced from 35 periods to 21 periods). The released plan is practical and can eliminate many disruptions on the shop oor that are due to the rigidity of the system. References
Collette, Y., & Siarry, P. (2003). Mulitobjective optimizationprinciple and case studies. Springer-Verlag. Fitzgerald, A. (2000). Enterprise resource planning (ERP)Breakthrough or buzzword? Third international conference on factory 2000. Competitive performance through advanced technology (Conf. Publ. No.359) (pp. 291297). London, UK: IEE. Guenov, M. D. (2002). Complexity and cost effectiveness measures for systems design. Second international conference of the manufacturing complexity network (pp. 455466). Halevi, G. (1980).The role of computers in the manufacturing processes. Weily-Interscience. Halevi, G. (1999). Re-structuring the manufacturing processapplying the matrix method. The St. Lucie/APICS Series on Resource Management. Halevi, G. (2004). Production planningsimple or complex. International Journal of Innovation and Technology Management, 1(4), 369372.

123

474 Halevi, G., & Cunha, P. F. (2006). Self organization shop oor control. International CIRP conference on digital enterprise technology, Setubal , Portugal, 1820 September, 2006. Liu, J., & MacCarthy, B. L. (1996). The classication of FMS scheduling problems. International Journal of Production Research, 34(3), 647656.

J Intell Manuf (2007) 18:467474 Tinham, B. (2000). What place MRP II in the new world? Manufacturing computer solutions, 6(1), 1418. Wang, K., Hirpa, L. G., Wang, Y., Yuan, Q., & Fang, M., (2003). A hybrid intelligent method for modelling the EDM process. International Journal of Machine Tools & Manufacturing, 43, 995999.

123