Kanban and Visual Management
Kanban and Visual Management
Kanban is a lean method for identifying and managing the work and the flow, and it does so by creating a pull through environment. We will start by identifying the work we need to do. Each work item we need to do will end up on an individual card. The cards will be “pulled” as we have the time and talent to commence that specific work element. The card then moves from the backlog of work items, to the WiP or Work in Progress section of the visual management board. The visual board may have headings such as:
- Work in Progress (WiP)
- Work for Review
- Work Completed / Demonstrated
We will pull items from the backlog and place the card in the state that corresponds to the work, for example, the work moves from backlog to work in progress as the work commences. It is easy for anybody to see the status of the work. Thus, the name a pull through system is one in which the items to be worked, will be taken when there is appropriate talent (team member) and time to do so rather than schedule-based work. We will also have rules that define and limit the amount of open work in progress to maintain focus and not overwhelm the system. The work items are prioritized but that priority is not based upon a fixed time when a specific work item is to be undertaken as for example as may be seen in conventional projects in the Gantt chart (more later).
We have written on the connection between throughput and capacity, but we will write just a bit more, and we do so with some reference from Factory Physics (Spears and Hopp, 1996)
- Law of batches (move batching): Cycle times over a segment of a routing are roughly proportional to the transfer batch sizes used over that segment, provided there is no waiting for the conveyance device.
- Law of Variability corollary (placement): In a line with where releases are independent of completions, variability early in a routing increases cycle time more than equivalent variability later in the routing
- Law of Utilization: – If a station increases utilization without making any other changes, average cycle time will increase in a highly nonlinear fashion.
- Law of Variability: Increasing variability always degrades performance of a production system.
To control the work, we must understand what is presently in the work queue, and how much we can handle at any given time of open work items. The laws mentioned above, impact the work underway and how the work moves through the work process, and leads us to queuing theory.
Contrast Kanban with the Gantt chart for tracking and managing the project. On the Gantt chart, we record specific tasks, the dependencies of those tasks to other tasks both before and after a specific task, along with the start and stop date for the work. The approach generally means the work starts on the date identified for the work to commence though previous work elements may take longer than expected and we may need to adjust or schedule to accommodate. Our plan is to start the work per the schedule. With kanban, we have no such delusion that the specific work task has a specific date to start, it is not even implied.
We see the Gantt plan has dates associated as well as the responsible individual (not shown in this short example). The Gantt chart represents a schedule driven approach to the work. To be sure it is possible to take on the next task on the Gantt chart when an item is completed, but what is more noteworthy is when a specific work item runs long, and now the expectation from the project management often results in pressure to undertake the work item with other resources, or over load the system and disperse the attention of the team to work on this item within the schedule defined.
Like most choices, there are positive and negative ramifications for making the decisions that we do. There could even be an advantage to doing both, one at the operations level (kanban) the other at the planning and projection level (Gantt). In this way the system is not overwhelmed, but we have some way of visualizing the consequences of the rate of work being accomplished on the prospective project deliver or end date, facilitating a range of prediction as to the conclusion of the project.