modeling a kanban production system -凯发k8网页登录

overview

this example shows a production system that uses kanbans to manage production activities. analysis of simulation results highlights problems in the system and suggests ways to improve its performance.

structure of the model

the modeled production system includes two part suppliers and an assembly line. the part suppliers use raw materials to manufacture parts. finished parts are transported to the assembly line to fabricate final products. completed products are shipped to distributors to fill production orders.

at the top level of the model:

kanban circulation

"kanban" comes from the japanese word for "signboard". a kanban production system is a pull system that determines its production according to the actual demand of the customers. these systems use kanbans as demand signals that propagate through the production system to trigger and regulate production activities, such as processing and storage.

this model simulates the circulation of two types of kanbans: withdrawal kanbans and work-in-process kanbans.

circulation of withdrawal kanbans for part a is modeled by the following blocks and subsystems:

the figures below show the part a supplier subsystem and assembly line subsystem.

during simulation, the block labeled obtain withdrawal kanban in the part a supplier subsystem must obtain a withdrawal kanban before a part a is transported and stored for final assembly. when a part a in storage is consumed in final assembly, the block labeled release withdrawal kanban a in the assembly line subsystem releases the withdrawal kanban. the kanban then returns to the block labeled obtain withdrawal kanban to allow replenishment of part a inventory.

circulation of work-in-process kanbans is modeled in the same fashion as withdrawal kanbans. for example, in the part a supplier subsystem, the block labeled obtain work-in-process kanban requests a work-in-process kanban upon producing a part a. after the part is completed and withdrawn, the block labeled release work-in-process kanban releases the work-in-process kanban. the kanban then returns to the block labeled obtain work-in-process kanban to allow the production of more of part a.

the model uses resource pools to model the group of kanbans. to learn about this technique, see .

dropped orders

a kanban production system reduces cost and waste by limiting inventories of work-in-process stock and finished products. however, when product demand fluctuates, lack of inventory may cause dropped orders.

this model simulates dropped orders caused by seasonal demand fluctuations. in the generate production orders subsystem, the output switch block labeled place orders uses first port that is not blocked as its switching criterion. during simulation, the block tries to send an order to the assembly line subsystem. if the inventory of finished product is empty, the block labeled fill production order in the assembly line subsystem does not accept this order. the block labeled place orders then drops this order by forwarding it to the entity sink block labeled dropped orders.

results and displays

during simulation, the data display subsystem displays these scopes to show the performance of the production system:

a display block at the root level of the model provides a numeric view of the number of orders completed and the number of orders dropped.

experimenting with the model

(for use with live model only)

using the model for performance analysis

the model with the original configuration represents a kanban production system with significant lost sales in months when demand is at a peak. analysis of simulation results suggests solutions to address this issue.

the following steps show how the solutions are developed.

step 1: run the simulation using the original configuration. as shown in the figures below, the scope labeled number of dropped orders indicates that the production system suffers significant lost sales between day 90 and day 150 of the year. comparing this result with the scope labeled product demand indicates that lost sales happen when product demand is at a peak.

step 2: comparing the demand in peak season with product supply indicates the assembly line does not provide sufficient production capability. according to the scope labeled product demand (see the figure above), 10 products are needed daily between day 90 and day 150. in contrast, as illustrated by the scope labeled number of products in final assembly (see the figure below), in the same period of time, only about 5 are in production every day - much less than the quantity in demand.

step 3: further observation of simulation results indicates that the inventory of part b is insufficient in the peak season. as illustrated by the scope labeled number of part b in storage (see the figure below), inventory is empty in the peak season. this explains the inadequacy in the production capability during final assembly - the assembly line is not provided with sufficient part b.

step 4: simulation results related to part b indicate that the use of withdrawal kanbans for part b is low in the peak season. this is displayed by the scope labeled number of part b withdrawal kanban in use shown in the figure below.

use of withdrawal kanbans is reduced when the assembly line requests a replenishment but the part supplier fails to respond to this request in time. this leads to an analysis of the production capability of part b in the peak season of the year.

step 5: the visual observations in the earlier steps suggest this quantitative analysis:

       part b demand = 10 /day

       maximal part b production rate = 12/1.5 = 8 /day

step 6: comparing demand and maximal production rate of part b indicates the inadequacy in production capacity. two possible solutions are:

step 7: to verify solution 1, reconfigure the kanban system by increasing the number of work-in-process kanbans for part b parameter to 15. simulation results indicate that with such an update, fifteen part bs are produced in parallel (see the scope labeled number of part b in process below). as indicated by the scope labeled number of dropped orders, the increase in part b supply eliminates the occurrence of dropped orders.

to verify solution 2, starting from the original configuration, reconfigure production capability by reducing the time it takes to produce a part b parameter to 1.2 day. with the increase in production capability, 10 final products are in assembly daily (see the scope labeled number of products in final assembly below). as illustrated in the scope labeled number of dropped orders below, such production capability can fully satisfy product demand, resulting in no loss of sales over the year.

the above steps explore the root cause of lost sales due to seasonal fluctuation in product demand. quantitative analysis suggests two solutions to respond to such demand fluctuations. simulation verifies that both solutions can indeed help the production system avoid seasonal lost sales.