Virtual Pull Systems
article written by Don Guild, Synchronous Management

In the first installment we discussed how to calculate (and recalculate) supermarkets. This installment discusses how to execute supermarket replenishment and make-to-order requirements with virtual pull.

Executing the Schedule with the "Fuel Gauge"
Now that we have a way to calculate supermarkets, let's look at how a visual Virtual Pull system might be executed. Open Customer Orders shows the current customer order backlog (i.e. sales orders), by future ship week. Even with MTS items, we want to make sure we have visibility of spike demands on their supermarkets. We also want to include MTO demand in the production schedule.
 

Fuel Gauge 1 shows the supermarkets and customer order requirements for each item in a Gantt chart format (weekly buckets). In order not to exceed the maximum allowable inventories for MTS items, the horizon shown for each MTS item is limited to the size of its supermarket in weeks (from Supermarket Calculations in the first installment). For MTS items, the larger of forecast and customer orders is displayed for each week, and for MTO items, total customer orders are displayed. Current on hand and in process inventories are also shown.

The red shaded weeks are covered by on hand inventory and the yellow shaded weeks are covered by in process inventories. Weeks with no shading are not yet covered. Our intent is to level load the machine for the next week at the forecasted rate of 14,200 units (see Forecast in the first installment), So, the scheduler's task is to determine the next 14,200 units to be released to production, while balancing both inventory and customer order requirements.

The scheduler generates the schedule for the next week by entering a value in the WEEKS TO COVER column (see Fuel Gauge 2). The Virtual Pull model then calculates and displays the number of units to make for each item to cover the specified number of weeks. Since the quantities in week 1 are already covered with on hand and on order inventory, let's begin by setting the WEEKS TO COVER to two weeks. Fuel Gauge 2 shows the resulting units to make, with coverage indicated by the green shading. However, the resulting schedule is for only 4,083 units, far below the level-loading requirement.

Let's extend the coverage to four weeks. Fuel Gauge 3 shows the result - over 27,000 items would need to be produced. This far exceeds the machine's capacity for one week - again not a level load.

Finally, let's cover three weeks, as shown in Fuel Gauge 4. The total units to make are now 14,250, approximately equal to one week of forecast - a level load - and a customer-driven mix! The final schedule for the next week is shown in Machine Schedule - One Week. And, Virtual Pull, the schedule can be easily reset at any time during the week if any of the input conditions change.

So, what have we accomplished? We have established maximum inventory levels required for stock items. We have quantified the constraints in both supply and demand. We have established a level-loaded, mixed-model schedule. This schedule fits our capacity and maximizes our customer service - on both stock and non-stock items - and it is easy to change. Finally, we have simplified the process to the point where we can reset supermarkets and generate schedules on demand for hundreds or thousands of items in minutes - without non-value-adding manual effort!

Virtual Pull Applications
Of course, these are only seven finished goods items. You will have to deal with many more, along with component items. Virtual Pull makes it easy. For example:
  One New England company buys raw goods from China in large batches and long lead times and
    finishes the products locally. Releases from China are managed via a monthly fuel gauge which is
    updated weekly and level-loaded by vendor (see Monthly Fuel Gauge screen shot).
  To schedule releases to one Midwest metal fabrication plant, the cell leaders and production
    schedulers collaborate with a weekly fuel gauge, updated daily (see Weekly Fuel Gauge).
  One southeast maker of consumer products posts the fuel gauge directly on the shop floor in the
    packaging cell; this fuel gauge and the cell are managed in real time by the cell leader - no
    scheduler is involved (see Daily Shop Floor Fuel Gauge).

With Virtual Pull all make-to-stock and make-to-order demand, at all levels of production and procurement, can be level-loaded and scheduled with the best possible mix and the least possible effort.

Other Issues
This introductory article has used a very simple set of data to illustrate the concepts of Virtual Pull. You will certainly face many more complicated supply and demand constraints.

Typical Demand Constraints
In our example, we have used weekly buckets. Daily or monthly buckets may be more appropriate for your environment. You will have to set your own definitions of make-to-stock vs. make-to-order. You may not have item forecasts available, and you may have to calculate or extrapolate them separately. For dependent demand items, you may have to use bills of materials, historical usage and future gross requirements instead of shipments and forecast. And, you may need to adjust your supermarkets over time to accommodate seasonal inventory builds or product introductions and phase-outs.

Typical Supply Constraints
On the supply side, you may need to adjust supermarkets for plant shutdowns, unreliable supply, or inventory inaccuracies. You may have dependent (major/minor) changeovers, fixed run sequences, or labor constraints. Out-of-pocket startup scrap or large changeover to cycle time ratios may limit how far batches can be reduced. Periods to cover may have to be adjusted on individual items to account for availability of raw material, capacity or family changeovers. And, larger batches and longer lead times could extend your scheduling horizon.

Finally...
Virtual Pull is not a proprietary software system; it is developed for your company using simple tools, such as Microsoft Excel and Access. The data for Virtual Pull is maintained in your current system - not on cumbersome kanban cards, and Virtual Pull is interfaced with your existing systems. Then, as changes are made to your data, the supermarkets are reset automatically.

Virtual Pull is also updated in real time as customer orders are received and shipped and as inventories are consumed and replenished. It provides scheduling and planning with the "control valves" they need to optimize flow and service. The Virtual Pull fuel gauge can be accessed or displayed anywhere in the plant (and to outside suppliers) to provide immediate visual control of all scheduling requirements. Since Virtual Pull does not require physical kanban cards or trigger boards, it can easily be applied to hundreds or thousands of items at once - at a small fraction of the effort required for manual systems. All of this, without making a single kanban card!