The long haul
Is your co-op's farm-to-plant milk hauling optimal?
This case study shows factors that can impact efficiency
By Peerapon Prasertsri, Graduate Research Assistant
Richard L. Kilmer, Professor University of Florida, Gainesville
Editor's Note: This article is based on
research conducted by the University of
Florida under a cooperative agreement
with the Rural Business-Cooperative Service
of USDA Rural Development.
n 1992, firms in food retailing became aware of a
major new competitor: Wal-Mart. Wal-Mart arrived on the food-retailing scene with a very cost-efficient inventory, warehouse and trucking system that allowed it to reduce operation costs by 5-percentage points below the food retailing industry average. In reaction to Wal-Mart, super-market chains (through their trade associations) started an initiative called "Efficient Consumer Response" (ECR), the objective of which was to design a more efficient food delivery system.
The efficiency of this system is now affecting milk marketing cooperatives. Milk processors, milk marketing cooperatives and dairy farmers need to improve their cooperation. They also need to be aware of how the action of each member of the vertical market influences the business operation of the others.
A major function of a milk-marketing cooperative is transferring and balancing the supply and demand of fluid milk from dairy farmers to milk processors. Farmers and processors want milk collected and delivered on a well timed schedule. It is the cooperative's task to ensure that the milk routing and scheduling are performed efficiently to improve coordination among farmers, the cooperative and milk processors.
This study focused on a state where most milk production is shipped to the fluid market, going directly from farm to private bottling plants in tanker trucks. The only storage the cooperative provides is on the tanker, and thus the logistics of operating a good tanker-transportation network is even more critical than in most other states. Delivery routes and schedules have been honed to a fine science here. But could they be doing even better?
A case study of a dairy cooperative's milk hauling operation was the basis for this article. Data was gathered for the farm-to-plant routing algorithm; the most efficient routing system for the cooperative was determined, as was the cost reduction from the most efficient routing system with the current routes operated by the cooperative. Sensitivity of the most efficient set of routes to imported milk procurement, changes in the dairy farm pick-up schedule and changes in the processing plant delivery schedule were also examined.
Routing complexity
Routing and scheduling are important activities for distributing highly perishable agricultural commodities in the vertical-market system. This is especially true for fluid milk, which requires virtually instantaneous transportation from producers to processing plants to maintain product quality.
The routing and scheduling problem is very complex. As the number of producers and processors increase, so do the possible ways to route and schedule trucks. The problem increases in complexity when farm and processing plant time windows are added. Routing and scheduling software, such as ArcLogistics Route 2 (ALR), have been developed to help solve the problem. ALR seeks efficient routes by using data about farms, trucks and processing plants in its street and road network database.
Truck-scheduling data were provided by the study cooperative for the period of October 3-9, 1999. The benchmark run was the actual milk collection and delivery routes used by the cooperative. The cooperative had a scheduled pick-up time for each farm and a delivery time schedule for each plant. A plus-or-minus 30-minute processing plant time window was included in the benchmark model, with no farm time window. In contrast, the alternative run was routed and scheduled by ALR with a plus-or-minus 30-minute processing plant time window and a plus-or-minus 2-hour farm time window.
The number of miles and the number of time window violations (a time window is violated if a truck visits a farm or processing plant before or after its time window) are two key results examined. The number of miles is directly related to the cooperative's cost of scheduling and routing miles from producers to processors. The number of time window violations (number and hours) implies the time-schedule performance of the cooperative's dispatchers.
Fewer farm time violations improve the satisfaction level of milk producers. The processing plant managers are more satisfied with lower processing plant time window violations. The benchmark and alternative runs were performed and compared in all of the cooperative's six service areas.
Plotting cost savings
The total mileage reduction between the benchmark and alternative runs range from a low 0.74 percent for the Service Area 1, to a high 14.01 percent for the Service Area 2. For all service areas, 3.36 percent (5,726 miles) of total mileage was eliminated by the alternative run when compared to the benchmark run. Based on $1.29 per mile, the cost savings corresponding to mileage reduction in all service areas was $7,387.26.
One reason for the different mileage reductions might be the nature of the service areas. More multiple-stop routes allow for more combinations in the route construction process, which has the potential for mileage reduction. For example, most farms in Service Area 1 (96.9 percent) provided a full load of milk for each truck. Thus, more than 95 percent of the trucks in this area made only one stop. The mileage reduction between the benchmark and the alternative run was 0.74 percent.
For Service Area 2, which had the highest mileage reduction (14.01 percent), only 53.2 percent of the total routes were one-stop routes. In other words, the more multiple-stop routes, the more potential mileage reductions.
However, these findings were mixed in Service Areas 5 and 6, where there is no direct milk-load delivery from dairy farms to processing plants. All trucks in the these two service areas returned to their terminals after completing the pickup process; there were no time window restrictions, unlike with the processing plants. Service Area 5 adhered to the correlation between the mileage reduction percentage and the percentage of multiple-stop routes; Service Area 6 did not.
Schedule violations impact efficiency
The hours of plant time window violations and the total number of time window violations are important components of overall dispatching efficiency for moving milk to milk-processing plants. The reduction in hours of plant time window violations between the benchmark and alternative runs was 83.71 percent for all service areas. The reduction in the number of plant time window violations between the benchmark and alternative runs was 55.20 percent for all service areas.
The reduction in hours of farm time window violations between the benchmark and alternative runs was 98 percent for all service areas. Meanwhile, the reduction in the number of farm window violations between the benchmark and alternative runs was 95.69 percent for all service areas.
Sensitivity analysis shows how the results change if some constraints are relaxed. Results from sensitivity runs were compared with those from the originally constrained alternative run.
In the first case, the sensitivity analysis results involving imported milk indicates little effect (less than a 0.4 percentage reduction in all categories) on the transportation system resulting from inclusion of the imported milk loads. In the second case, the reduction between the alternative and sensitivity runs was 3.25 percent for mileage and 69.49 percent in hours of plant-time window violations.
Mileage reductions were small, while plant time window reductions were large. This indicates that if the cooperative were allowed to pick up milk loads without farm time window restrictions, the processing plant time window violations would be reduced by almost 70 percent. Moreover, results from the third sensitivity run (no farm and plant time window constraints) showed little further improvement from the second sensitivity run (less than a 0.5 percentage reduction in mileage and total used time.).
Case study routes rated "efficient"
This case study concluded that the co-op's current truck routes are efficient. The route mileage could only be improved by 3.36 percent (5,726 miles) for a saving of $7,387.26. This is a low savings rate when the cooperative's actual mileage is compared to the actual route mileage (as determined by a software package that does not include all of the real-world uncertainties encountered by the cooperative drivers).
Even though all service areas taken together are efficient, Service Areas 2, 4, 5 and 6 have routes that may be reorganized to reduce route mileage and route time. They have the lowest percentage of one-stop routes and the largest reductions in route mileage between the benchmark and alternative runs. Routes in these four areas need to be re-evaluated and possibly reorganized. Processing-plant time window violations and farm time window violations for the alternative runs in all service areas imply a disparity between when milk is available from the farm and when the processing plants need the milk. Processing-plant time window violations (56) represent 6.46 percent of all loads. This means that 6.46 percent of the loads were not on time.
For farm time window violations, 1.38 percent of farm pick ups were not on time. These violations occurred with time windows that were 1 hour (the scheduled time of delivery plus-or-minus 30 minutes) and four hours (the scheduled time of pick up plus-or-minus 2 hours) in length at the processing plant and the farm, respectively.
There is no way to meet all the plant-time window requirements and farm-time window requirements with the current time windows. To increase the ability to pick up loads from farms and deliver milk loads to plants with the current delivery schedule would require an adjustment in farm and processing plant time windows.
Adding the 4-hour time window to the scheduled farm pick-up time in the alternative run significantly reduced the number of time window violations (470 to 19 for farms and 125 to 56 for processing plants) and the violation hours (427.40 to 7.71 hours for farms and 287.33 to 46.81 hours for processing plants) for farms and processing plants compared to the bench-mark run. An adjustment in time window length at the farm level not only reduced the time window violations at the farm level, but also at the processing-plant level.
Increasing the time window length also reduced the total route time for all service areas by 4.88 percent, or $4,095.88, and the total route miles by 3.39 percent (5,726.56 miles), or $7,387.26 for the seven-day period.
Schedule modification could help in some areas
The implications point to scheduling as the over-riding problem, and that modifications in the current schedule could improve efficiency in some smaller service areas. Current farm-to-plant scheduling does not allow direct farm-to-plant delivery without delays.
An efficiently routed and scheduled transportation system reduces mileage and route time. Adjusting the time windows and/or the scheduled pickup and delivery times reduces the total cost and time of moving milk from the farm to the processing plant. Areas with multiple-stop routes possess the potential for more route improvements than areas with mainly one-stop routes.
What is done at one level of the vertical market system has an impact on other levels of the system. To improve vertical coordination, processors, farmers and cooperatives must know how their actions influence each other.
