A product layout arranges machines or workers in a line according to the operations that need to be performed to assemble a particular product. From this description, it would seem the layout could be determined simply by following the order of assembly as contained in the bill of material for the product. To some extent, this is true. Precedence requirements specifying which operations must precede others, which can be done concurrently and which must wait until later are an important input to the product layout decision. But there are other factors that make the decision more complicated.
Product layouts or assembly lines are used for high-volume production. To attain the required output rate as efficiently as possible, jobs are broken down into their smallest indivisible portions, called work elements. Work elements are so small that they cannot be performed by more than one worker or at more than one workstation. But it is common for one worker to perform several work elements as the product passes through his or her workstation. Part of the layout decision is concerned with grouping these work elements into workstations so products flow through the assembly line smoothly. A workstation is any area along the assembly line that requires at least one worker or one machine. If each workstation on the assembly line takes the same amount of time to perform the work elements that have been assigned, then products will move successively from workstation to workstation with no need for a product to wait or a worker to be idle. The process of equalizing the amount of work at each workstation is called line balancing.
Assembly line balancing operates under two constraints, precedence requirements and cycle time restrictions.
Precedence requirements are physical restrictions on the order in which operations are performed on the assembly line. For example, we would not ask a worker to package a product before all the components were attached, even if he or she had the time to do so before passing the product to the next worker on the line. To facilitate line balancing, precedence requirements are often expressed in the form of a precedence diagram. The precedence diagram is a network, with work elements represented by circles or nodes and precedence relationships represented by directed line segments connecting the nodes. We will construct a precedence diagram later in Example 7.2.
Cycle time, the other restriction on line balancing, refers to the maximum amount of time the product is allowed to spend at each workstation if the targeted production rate is to be reached. Desired cycle time is calculated by dividing the time available for production by the number of units scheduled to be produced:
Suppose a company wanted to produce 120 units in an eight-hour day. The cycle time necessary to achieve that production quota, is
Cycle time can also be viewed as the time between completed items rolling off the assembly line. Consider the three-station assembly line shown here.
It takes 12 minutes (i.e., 4 + 4 + 4) for each item to pass completely through all three stations of the assembly line. The time required to complete an item is referred to as its flow time, or lead time. However, the assembly line does not work on only one item at a time. When fully operational, the line will be processing three items at a time, one at each workstation, in various stages of assembly. Every 4 minutes a new item enters the line at workstation 1, an item is passed from workstation 1 to workstation 2, another item is passed from workstation 2 to workstation 3, and a completed item leaves the assembly line. Thus, a completed item rolls off the assembly line every 4 minutes. This 4-minute interval is the actual cycle time of the line.
The actual cycle time, Ca, is the maximum workstation time on the line. It differs from the desired cycle time when the production quota does not match the maximum output attainable by the system. Sometimes the production quota cannot be achieved because the time required for one work element is too large. To correct the situation, the quota can be revised downward or parallel stations can be set up for the bottleneck element.
Line balancing is basically a trial and error process. We group elements into work stations recognizing time and precedence constraints. For simple problems, we can evaluate all feasible groupings of elements. For more complicated problems, we need to know when to stop trying different workstation configurations. The efficiency of the line can provide one type of guideline; the theoretical minimum number of workstations provides another. The formulas for efficiency, E, and minimum number of workstations, N, are
The total idle time of the line, called balance delay, is calculated as (1 - efficiency). Efficiency and balance delay are usually expressed as percentages. In practice, it may be difficult to attain the theoretical number of workstations or 100 percent efficiency.
The line balancing process can be summarized as follows:
Real Fruit Snack Strips are made from a mixture of dried fruit, food coloring, preservatives, and glucose. The mixture is pressed out into a thin sheet, imprinted with various shapes, rolled, and packaged. The precedence and time requirements for each step in the assembly process are given below. To meet demand, Real Fruit needs to produce 6,000 fruit strips every 40-hour week. Design an assembly line with the fewest number of workstations that will achieve the production quota without violating precedence constraints.
First, we draw a precedence diagram. Element A has no elements preceding it, so node A can be placed anywhere. Element A precedes element B, so the line segment that begins at node A must end at node B.
Element A precedes element C. Again, a line segment from node A must end at node C.
Elements B and C precede element D, so the line segments extending from nodes B and C must end at node D. The precedence diagram is completed by adding the time requirements beside each node.
Next, we calculate the desired cycle time and the theoretical minimum number of workstations:
Since we cannot have half a workstation (or any portion of a workstation), we round up to 3 workstations.
We must group elements into workstations so that the sum of the element time at each workstation is less than or equal to the desired cycle time of 0.4 minutes.
Examining the precedence diagram, let us begin with A since it is the only element that does not have a precedence. We assign A to workstation 1. B and C are now available for assignment. Cycle time is exceeded with A and C in the same workstation, so we assign B to workstation 1 and place C in a second workstation. No other element can be added to workstation 2, due to cycle time constraints. That leaves D for assignment to a third workstation. Elements grouped into workstations are circled on the precedence diagram.
Our assembly line consists of three workstations, arranged as follows:
Since the theoretical minimum number of workstations was three, we know we have balanced the line as efficiently as possible.
The assembly line has an efficiency of
Line balancing by hand becomes unwieldy as the problems grow in size. Fortunately, there are software packages that will balance large lines quickly. IBM's COMSOAL (Computer Method for Sequencing Operations for Assembly Lines) and GE's ASYBL (Assembly Line Configuration Program) can assign hundreds of work elements to workstations on an assembly line. These programs, and most that are commercially available, do not guarantee optimal solutions. They use various heuristics, or rules, to balance the line at an acceptable level of efficiency. The POM for Windows software lets the user select from five different heuristics: ranked positional weight, longest operation time, shortest operation time, most number of following tasks, and least number of following tasks. These heuristics specify the order in which work elements are considered for allocation to workstations. Elements are assigned to workstations in the order given until the cycle time is reached or until all tasks have been assigned.
|VW's Super Efficient Factory|
Volkswagen built a $250 million truck and bus plant in Resende, Brazil, that is likely to become the model for new-car factories around the world. Seven suppliers, each responsible for a single module, make components in the plant using their own equipment and attach the components to trucks and buses on the final assembly line. German manufacturer VDO Kienzle, for example, is in charge of the truck cab. Marked off from other suppliers' workspaces by yellow lines on the floor, VDO workers install everything from cab seats to instrument panels. Then they attach the completed cab to the final chassis as it moves down the assembly line through VDO's section of the factory. Inventory costs are down because parts are made only an hour or so before they are needed and schedules are tightly coordinated.
In traditional automotive plants, suppliers deliver parts, assemblies, and modules to the final assembly line, but they never assemble them themselves. In pilot runs, improvements by suppliers in final assembly have cut work hours by 12%. On-site supplier suggestions for improved designs are also expected to yield lower costs and improved quality.
The plant is run by a daily roundtable discussion between VW and its partners. "VW has to be part of the table, not its owner," says the plant manager. That's one big advantage of the new system--VW and its suppliers are all in it together. Individual capital investment is dramatically lowered with VW providing the building and assembly line conveyors, and the suppliers putting in their own tools and fixtures, and hiring their own workers. Only 200 of the 1,400 workers are VW employees. If sales of trucks and buses do not meet predictions, everyone takes a hit, not just VW.
|Source: David Woodruff, "VW's Factory of the Future," Business Week (October 7, 1996): 52-56.|
7-8. Describe several heuristic approaches to line balancing.