Ever since engineers and analysts have been crunching numbers there’s been this old adage – “Garbage In = Garbage Out”. What they’re saying is that data (metrics) is only as good as the methods used to measure it - if something is measured or calculated the “wrong way”, the results will be wrong. As obvious as this may be, the proper way to measure is not always obvious, or may even be in dispute.
For instance, many DCs have issues with low daily or hourly pick rates. Often, the pick rate is being measured based on a 420 minute work period (8 hrs minus lunch and break periods). However, if due to accumulative idle time the pickers are only picking between 240 and 360 minutes per period – that would translate into low pick rates on a spreadsheet, but it is a completely inaccurate statement of system picking capacity.
In a recently published article on this subject, Don Savage of Diamond Phoenix, said, “It is critical that all rates are measured against the time actually worked. This makes for a more complex measurement, but it gives you a much clearer picture of the true capabilities of the existing system. It is also a good indicator of wasted or available time throughout the day.”
The following (3) examples, taken in part from Savage’s article, help to explain how a systems performance can change based on how it is measured.
1. Inconsistent Flow of Work
Savage points out that many DC operations are unable to provide employees with a steady flow of work, which can be a serious problem for batch picking systems. The lack of constant work could be the result of IT issues, a lack of orders in process, or even a deficient routing system for passing orders.
Pickers cannot pick at high rates if they don’t have enough orders. This drought in orders creates lost time, which cannot be recovered, and daily pick rates plummet.
The best way to track this is to measure the number of orders present in each zone throughout the day in short time increments (5-minute intervals). Although it sounds tedious, this is an exceptionally important measurement because it directly verifies the effectiveness of the order routing and starting operation. If full batches are not available on a constant basis, pick rates will not reach their maximum.
Changes should then be made to move more orders into the zones.
2. Organization of Operations
According to Savage, many operations do not organize tasks in a way that fills available time; they are pushed into a linear process instead of a concurrent one. Because a separate measurement against time is necessary for each one, the time required for each operation is usually overstated.
Typically, DC systems operate in sporadic bursts of work with time left over in between. It’s a kind of hurry up and wait situation. This is most common in wave picking operations and is referred to as the bell curve effect. Sometimes this can be corrected by overlapping waves. However, you can’t fill that time appropriately if you don’t know when and how long it is. So the first step is to measure this over a period of multiple days so you can see those busy versus slack time windows.
If wave overlapping is not an option – try filling the gaps with other tasks, which may mean combining different tasks in the same time frame (picking and replenishment, for example).
3. Not Accounting For Peaks
In this case, Savage acknowledges that it is not uncommon for an entire day’s work to get compressed into a few hours due to response commitments or shipping requirements. In this situation, he says some managers still measure their efficiency rates over an entire day rather than actual time. In addition, when more workers are assigned to picking zones they are frequently not included in the rate measurements. This leads to over or understating rates for both the day and peak times.
To avoid this error, Savage suggests that if work is condensed into a specific period of the day, then measure the rate for that time period and the number of people working concurrently; this gives you your maximum rate per person. You can then use that rate and the peak window to plan the rest of the operations for the day, which may mean different labor assignments during peak and slack times.
|
