Congress mandated the National Food Service Management Institute (NFSMI) to conduct applied research activities that provide information to improve the quality and cost effectiveness of Child Nutrition Programs (CNPs). Among the research activities that have relevance to CNPs is the use of analytical techniques involving current technology to solve day-to-day operational problems. Engineering and other professions have used these methods for several decades.
This study was initiated to determine the feasibility of using computer simulation techniques to answer questions associated with customer service in CNPs. Results suggested that although the exacting procedures of computer simulation methodology present many challenges, simulation does have relevance in solving CNP operational problems.
RESEARCH
Computer simulation has been used in foodservice management since the 1960s, but it has not been applied to the CNP arena. Simulation is a process that uses a computer to design a model of a real system. We use computer models to learn something about a real life situation that cannot be altered directly, either because the system does not yet exist or it is too difficult to change just for experimental purposes. Simulation models can help explain, understand, or improve a system. The purpose of simulation modeling is to help the decision maker solve a problem.
Objectives
The purpose of the NFSMI's research focus on computer simulation was to explore applications in school foodservice operations. The specific objectives of this study were to:
Method Research Site
To build a computer simulation model, we needed to document times associated with service procedures in an actual school cafeteria. One district school foodservice director from a small school district agreed to allow data collection for the research study. Children in grades four through six attended this elementary school. The menu consisted of choices of entrees, vegetables, and fruit items each day. The foodservice staff placed at least one menu item on the compartment tray and the students picked up the other food items. We videotaped the lunch period and recorded the time required for students to receive a compartment tray; pick up food items, such as milk; and be processed by a cashier. Additional time data were recorded on the frequency and duration of food pan changes, the time between food pan changes, and the time to serve teachers.
Two days during the data collection period only one foodservice assistant served hot food to the students. Two employees were on the serving line the other seven days. All observations were videotaped for frequency of food pan changes, time between food pan changes, and time to serve teachers for the seven days when two foodservice assistants were on the serving line. The school manager provided estimates of the number of students released at one time into the cafeteria for use in the computer model.
Simulation Model
We worked with a consultant from Systems Modeling Corporation to build the simulation model using ARENA software (Figure 1). After the model was functionally complete, the model was verified and validated. Verification seeks to show that the computer program does as expected and intended, thus providing a correct logical representation of the cafeteria system being modeled. Validation establishes that model behavior validly represents that of the real-world system being simulated.
Animation of the Simulation Model
The simulation model can be run using graphics. Figure 2 illustrates the model 83 seconds into the simulation run. "Students in" was the count of students that have entered the cafeteria. "Hot Q" was the number of students waiting in line to receive a food tray. Here, one student was receiving a tray and four were waiting in the queue. "Cold Q" was the number of students waiting to pick up cold food items. The capacity of the cold food section was five; three students were moving through this station, thus, there were no students in the queue. "Cash Q" was the number of students waiting for the cashier. One student was being served by the cashier and one was waiting in line. "Students out" was the count of students processed through the serving line.
In the original data, two foodservice assistants served on the cafeteria line. When viewing the videotapes, we observed many delays due to stopping of the line to replace food pans. A research question was posed: If there could be no changes in the arrival time of students into the cafeteria and staffing of the serving line, what could be done to speed the movement of students through the serving line?
We proposed a simulation model which used one of the two foodservice
assistants scheduled to work on the serving line as a kitchen runner. The
responsibilities of this position would be to anticipate food pan exchanges,
retrieve food from the kitchen, and thus reduce service delays due to food
outages. Total time for students to go through the serving line and the
number of students standing in line to receive a tray were calculated for:
The average total time for a student to go through the serving line in the scenario with two foodservice assistants serving was four minutes and 31 seconds (271 seconds). This average represented the time beginning when the class arrived in the cafeteria, a student received a compartment tray, picked up cold items and milk, and was processed by the cashier. The time ranged from two minutes and 23 seconds (143 seconds) to over nine minutes (572 seconds) for a student to go through the line.
When the simulation model altered the staffing pattern to use one foodservice assistant on the line and a kitchen runner, the average time decreased to one minute and 58 seconds (118 seconds) for a student to move through the cafeteria line. This time ranged from one minute and 48 seconds (108 seconds) to two minutes and 17 seconds (137 seconds), and most children could obtain a lunch tray before the next class arrived in the cafeteria. Figure 3 depicts the number of children waiting to receive their lunch tray taken from one replication. Each peak represents a class entering the cafeteria. Clearly most of the children could obtain a lunch tray before the next class arrived in the cafeteria. In this example, there was only one instance where over 30 children were in line to receive a lunch tray. Computer simulation showed that the use of a runner to stock the cafeteria lines could save time and increase the efficiency of the serving line. Total student time through the serving line decreased an average of 153 seconds, or 56 percent.
Arrival Time Scenario
In the second scenario, we asked the following research question: What effect does the sequencing of class arrival times have on the movement of students through the serving line? The foodservice director had not indicated that scheduling was a problem in this school; however, the release of classes for mealtime is a point of frustration for many school foodservice directors. Too many students dumped into the cafeteria at once create long lines and a decrease in the students' ability to eat and enjoy their meals. Longer lunch periods, a solution to the first situation, creates havoc in academic scheduling for the school day. Ideal solutions are difficult to detect, but computer simulation represented an approach to use in analyzing alternatives. We wanted to test whether computer simulation could be used to analyze the flow of students into the cafeteria. Simulation allowed us to test scenarios without involving school administrators to alter meal schedules, and potentially disrupt lunch periods without knowing the end results.
We made several modifications in the arrival time of classes into the cafeteria. The simulation model was developed with one foodservice assistant serving on the cafeteria line and the second assistant as the kitchen runner. Total time for students to go through the serving line and the number of students in line to receive a tray were calculated.
In the original data for the elementary school, classes were being released into the cafeteria at various times. This time ranged from 1 minute to 7 minutes between arrival times of classes into the cafeteria. The number of students in each class ranged from 9 to 26. We ran the simulation model using several common arrival times: blocks of ten classes every 30 minutes; one class every 4 minutes; and two classes every 8 minutes.
As depicted in Table 1, the average time for a student to go through the cafeteria line ranged from 1 minute, 43 seconds to over 12 minutes. The arrival time of blocks of ten classes every 30 minutes scenario showed that students spent the longest time standing in line and moving through the serving line. Upon examination of the data from one replication, the minimum amount of time for a student to go through the line was 30 seconds. The maximum length of time in this replication was 27 minutes and 52 seconds; thus, some students were in line almost 28 minutes. If students had 30 minutes in which to eat, it is doubtful whether some students would have time to eat their meals.
Table 1. Average time for a student to move through the serving line
for various arrival rates into the cafeteria (a)
| Class Arrival Rate | Average Time |
|---|---|
| Original data (b) | 1 minute 58 seconds |
| Ten classes every 30 minutes (c) | 12 minutes 49 seconds |
| One class every four minutes (c) | 1 minute 43 seconds |
| Two classes every eight minutes (c) | 2 minutes 55 seconds |
(a) 95% Confidence level (b) N = 60 replications (c) N = 20 replications
The scenario with arrival times of one class every 4 minutes moved students through the serving line in the least amount of time. When one replication was examined, the data ranged from 15 seconds to 3 minutes 42 seconds. The longest length of time any student had to wait in line in this replication was less than 4 minutes. A disadvantage is that this scenario lengthened the lunch period to two hours.
The scenario of two classes arriving every 8 minutes also extended the lunch period to two hours. However, the difference between this scenario and the one previously discussed is that two classes are arriving at the same time. Data from one replication showed that the maximum amount of time for a student to move through the cafeteria line was 6 minutes 27 seconds. A student possibly could be in line for over 6 minutes compared with less than 4 minutes in the previous scenario.
Number of Students in Line
In the original data, most students could receive a lunch tray before the next class arrived in the cafeteria. Data from one replication showed the maximum number of students in line any time was 32.
When blocks of ten classes arrived in the cafeteria every 30 minutes, the number of students who had to stand in line to receive a lunch tray greatly increased. Data from one replication showed that as many as 202 students were in line to receive a lunch tray. All students did receive a tray before the next group of classes arrived; however, their time to eat was limited.
In the scenario of one class arriving every four minutes, all students in each class received lunch trays before the next class of students entered the cafeteria (Figure 4). When we examined the data from one replication, the maximum number of students waiting in line to receive a tray was 23. This closely corresponded to the number of students in each class.
All the students in the scenario of two classes arriving in the cafeteria every eight minutes also received their trays before the next block of classes reached the cafeteria (Figure 5). However, the number of students who had to stand in line to receive a lunch tray was greater. Data from one replication suggested that up to 43 students were standing in line to receive a tray.
We concluded that computer simulation is a viable alternative to generate strategies for evaluating meal schedules without physically moving students and disrupting the academic day. Only one of our scenarios slightly improved (15 seconds) the average time each student took to move through the serving line. The benefit of computer simulation was that it allowed us to see the ebb and flow of student movement, determine the the minimum and maximum amount of time for a student to move through the serving line, and contrast average time through the line with total waiting time. This would be virtually impossible to do by simple observation.
FOR MORE INFORMATION
Nettles, M.F. (1995). Use of computer simulation in school food service,
NFSMI-R21-95. University, MS: National Food Service Management Institute.
NFSMI. Champions of change: child nutrition personnel, TT102595, University, MS: author.
NFSMI. Go for the gold with customer service, ET9-94, University, MS: author.
NFSMI. Participation: The key to healthy food practices, TT30994, University, MS: author.
Order No. R105-96
Authors: Mary Frances Nettles, PhD, RD, and Martha T. Conklin, PhD, RD. Dr. Nettles is a Research Scientist and Dr. Conklin is the Director, Division of Applied Research, National Food Service Management Institute, The University of Southern Mississippi, Hattiesburg, MS. May 1996.