Labor Scheduling Research Articles

Microcomputer-based workforce scheduling for hospital porters.

This paper focuses on labour scheduling for hospital porters who are the major workforce providing routine cleansing of wards, transportation and messenger services. Generating an equitable monthly roster for porters while meeting the daily minimum demand is a tedious task scheduled manually by a supervisor. In considering a variety of constraints and goals, a manual schedule was usually produced in seven to ten days. To be in line with the strategic goal of scientific management of an acute care regional hospital in Hong Kong, a microcomputer-based algorithm was developed to schedule the monthly roster. The algorithm, coded in Digital Visual Fortran 5.0 Professional, could generate a monthly roster in seconds. Implementation has been carried out since September 1998 and the results proved to be useful to hospital administrators and porters. This paper discusses both the technical and human issues involved during the computerization process.

Labor scheduling, part 4

After forecasting demand, determining staffing levels to meet that demand, and specifying which employees will be needed, a manager must still monitor the day's actual demand to ensure that the workforce schedule is serving customers' needs as intended. A key element in that determination is to assess whether the real demand is matching the forecasted demand. Moreover, that determination should be made as early as possible so that the manager can make mid-course corrections to the schedule, if needed. One way to make an early call about the day's demand is to develop graphs of the operation's historical demand for each day of the week and compare the current day's demand to those patterns. That comparison usually will give the manager a reasonably solid indication of what level of demand the operation will experience for the day. That indication, in turn, allows the manager to assess whether the demand forecasts underlying the day's schedule were accurate and, if not, to take necessary actions to adjust staffing levels. The greater the manager's certainty about the day's eventual demand level, the greater the possibility that the manager can make timely, on-the-spot adjustments to the workforce schedule.

Labor Scheduling, Part 4

After forecasting demand, determining staffing levels to meet that demand, and specifying which employees will be needed, a manager must still monitor the day's actual demand to ensure that the workforce schedule is serving customers' needs as intended. A key element in that determination is to assess whether the real demand is matching the forecasted demand. Moreover, that determination should be made as early as possible so that the manager can make mid-course corrections to the schedule, if needed. One way to make an early call about the day's demand is to develop graphs of the operation's historical demand for each day of the week and compare the current day's demand to those patterns. That comparison usually will give the manager a reasonably solid indication of what level of demand the operation will experience for the day. That indication, in turn, allows the manager to assess whether the demand forecasts underlying the day's schedule were accurate and, if not, to take necessary actions to adjust staffing levels. The greater the manager's certainty about the day's eventual demand level, the greater the possibility that the manager can make timely, on-the-spot adjustments to the workforce schedule.

Labour scheduling at an airport refuelling installation

We have developed an automatic assignment procedure for labour scheduling at a continously operating airport refuelling installation. Different types of workers are subject to different conditions on the shifts they can do (morning, evening or night), shift lengths, days-off, holidays, terms of contract and working hours. The scheduling process is carried out on a weekly basis, but there are mid-term and long-term conditions and objectives which link each week with the preceding and following weeks. Our package uses firstly a tabu search algorithm to find the best schemes of shifts/days-off to be used to cover the requirements. Secondly, an assignment problem is solved to match the schemes with available workers. Finally, the starting and finishing times for each working day are determined for each worker. The system has been designed to run on a personal computer and gives good quality solutions in very short computing times.

Labor scheduling, part 3: Developing a workforce schedule

“Who will help me bake this bread?,” asked the little red hen. “Not I,” said the dog. “Not I,” said the cat. “Not I,” said the mouse. “Then I will,” said the little red hen.

Labor Scheduling, Part 3

Developing a proper workforce schedule is a key factor in a service business's ability to serve its customers well, but economically. Having forecast customer demand and determined the number of employees needed to meet that demand, the manager must next create the actual schedule. Classic schedule-development frameworks tend to match employee supply and customer demand in isolation for each planning period. The best schedule-planning mechanism accounts for employee-related factors as part of developing the schedule, rather than a two-phase approach of setting a schedule and then assigning employees to fill it. By considering employee work constraints in advance, the manager can account for complementary preferences as the schedule is created, with the result of a better match between employees and their shifts. Among the scheduling considerations is when controllable work will be performed. By taking all elements into account as the schedule is developed, a manager can meet both customer-service and economic targets, while at the same time meeting as many employee preferences as possible.

Labor Scheduling, Part 2

This paper, second in a series, focuses on using forecasts of customer demand for service to estimate the number of employees needed to provide that service. There are three basic approaches to translating demand forecasts into employee requirements: using productivity standards, using service standards, and using economic standards. As described in this paper, the staffing levels determined by using the economic standard provide the best economic performance. The aim of using an economic standard is to deliver the service most economically. Although productivity and service standards can sometimes match the performance of an economic standard, they do so only in narrow ranges of business volume.

Labor scheduling, Part2

You've figured out how many customers to expect, and now you need to decide how many employees to schedule. Alas, to do that well is more complicated than you might think.

Labor scheduling, part 1: forecasting demand

This paper focuses on forecasting demand for services, the first of a four-part series on developing effective workforce schedules. Serving customers well, at a reasonable cost, requires the proper number and mix of employees. Scheduling too many employees means high labor costs, while scheduling too few workers can mean poor service that drives away business. Forecasting customer demand involves eight steps: determine the nature of the work, identify those factors that generate the work (i.e., the labor drivers), determine whether the labor drivers are time variant or time invariant, determine the time interval for tracking the time-variant labor drivers, forecast the time-variant labor drivers, smooth those forecasts, track the error in those forecasts, and define the allowable window for controllable work. Balancing employees' skills and availability, plus governmental regulations, company policies, and contractual obligations regarding work schedules, can be a manager's nightmare.

Labor Scheduling, Part 1

This paper focuses on forecasting demand for services, the first of a four-part series on developing effective workforce schedules. Serving customers well, at a reasonable cost, requires the proper number and mix of employees. Scheduling too many employees means high labor costs, while scheduling too few workers can mean poor service that drives away business. Forecasting customer demand involves eight steps: determine the nature of the work, identify those factors that generate the work (i.e., the labor drivers), determine whether the labor drivers are time variant or time invariant, determine the time interval for tracking the time-variant labor drivers, forecast the time-variant labor drivers, smooth those forecasts, track the error in those forecasts, and define the allowable window for controllable work. Balancing employees' skills and availability, plus governmental regulations, company policies, and contractual obligations regarding work schedules, can be a manager's nightmare.

Labor staffing and scheduling models for controlling service levels

The problems of labor staffing and scheduling have received substantial attention in the literature. We introduce two new models of the labor staffing and scheduling problems that avoid the limitations of existing models. Collectively, the models have five important attributes. First, both models ensure the delivery of a minimally acceptable level of service in all periods. Second, one model can identify the least expensive way of delivering a specified aggregate level of customer service (the labor staffing problem and a form of labor scheduling problem). Third, the other model can identify the highest level of service attainable with a fixed amount of labor (the other form of the labor scheduling problem). Fourth, the models enable managers to identify the pareto relationship between labor costs and customer service. Fifth, the models allow a degree of control over service levels that is unattainable with existing models. Because of these attributes, which existing models largely do not possess, we expect these models to have broad applicability in a wide range of organizations operating in both competitive and noncompetitive environments. © 1997 John Wiley & Sons, Inc. Naval Research Logistics 44: 719–740, 1997

An integrated analysis of nurse staffing and scheduling policies

In this paper we present an integrated nurse staffing and scheduling system for analyzing nurse workforce management policies. A mixed-integer linear programming nurse staffing model is used to determine aggregate labor requirements for a 6-month planning horizon. Subsequently, a mixed-integer linear programming model is used to disaggregate the nurse staffing plan into 2-week labor schedules. The integrative system permits recursion between the staffing and scheduling models and thus enables nurse workforce management to rapidly evaluate the impact of both staffing and scheduling policies. We use the new system to study the effects of staffing and scheduling policies on labor costs. The results suggest that there are indeed important interactions between staffing and scheduling policies.

Microcomputer‐based working set generation methods for personnel scheduling

Presents a new approach to working set generation for personnel scheduling problems. In full‐time (FT) and mixed‐workforce (MW) experiments, generates the schedules in the working sets from the use of two‐phase heuristic labour scheduling solution procedures. The solution procedures were implemented on a 386 microcomputer and did not require the specification of the size of the working sets in advance. In the FT experiment, the general set‐covering formulations (GSCFs) associated with the produced working sets were solved with integer programming. The new working set procedure yielded optimal integer solutions for all 36 test problems in the FT experiment. Owing to the size and complexity of the problem data in the MW experiment, the GSCFs associated with the working sets were solved with linear programming, and heuristic rounding procedures were applied to obtain feasible integer solutions. The mean labour costs of these solutions averaged 0.69 per cent less than the mean cost of solutions obtained via the application of heuristic rounding procedures applied to the linear programme solutions for the GSCFs associated with the master sets. Compares solution costs for the new working set method with those associated with other working set generation/refinement procedures. Results indicate that the new method produces lower solution costs in less control processing unit time.

Foodservice Forecasting: A Comparative Study of Exponential Smoothing and Expert System Models

The forecast is one of the most critical decisions in the foodservice industry because it is reflected in the material costs, labor costs, labor schedule, and overhead costs. The purpose of this study was to compare the purchasing forecasts using an expert system (knowledge base system) and an exponential smoothing model to the actual usage data in order to determine which one is best for the operation of a university dining center. This university dining center utilized a food court concept. Three of the four service lines were forecast, these lines included mexican, italian, classic, and deli menu options. An expert system previously developed was used. The expert system contained more than 1000 rules that replicate the expertise of the manager of the foodservice operation in forecasting. Data from the fall semester 1993 were used to forecast spring semester 1994 using the exponential smoothing. Different alpha values were tested to find the forecast value with the minimum mean squared error (MSB). The optimum value was α = 0.45. Results showed that in forecasting the classic and mexican lines the expert system model outperformed the exponential smoothing model because it had the lowest mean absolute deviation and mean squared error. However, for the italian line, the exponential smoothing forecast outperformed the expert system.

Labor scheduling using NPV estimates of the marginal benefit of additional labor capacity

AbstractAn extensive literature exists on the problems of daily (shift) and weekly (tour) labor scheduling. In representing requirements for employees in these problems, researchers have used formulations based either on the model of Dantzig (1954) or on the model of Keith (1979). We show that both formulations have weakness in environments where management knows, or can attempt to identify, how different levels of customer servve affect profits. These weaknesses results in lower‐than‐necessary profits.This paper presents a New Formulation of the daily and weekly Labor Scheduling Problems (NFLSP) designed to overcome the limitations of earlier models. NFLSP incorporates information on how changing the number of employees working in each planning period affects profits. NFLP uses this information during the development of the schedule to identify the number of employees who, ideally, should be working in each period. In an extensive simulation of 1,152 service environments, NFLSP outperformed the formulations of Dantzig (1954) and Keith (1979) at a level of significance of 0.001. Assuming year‐round operations and an hourly wage, including benefits, of $6.00, NFLSP's schedules were $96,046 (2.2%) and $24,648 (0.6%) more profitable, on average, than schedules developed using the formulations of Danzig (1954) and Keith (1979), respectively. Although the average percentage gain over Keith's model was fairly small, it could be much larger in some real cases with different parameters. In 73 and 100 percent of the cases wa simulated NFLSP yielded a higheprofit than the models of Keith (1979) and Danzig (1954), respectively.

Gender and the Reorganization of Historic Zuni Craft Production: Implications for Archaeological Interpretation

Over the past century, there has been a dramatic reorganization of craft production at Zuni Pueblo, including a change from a gender-specific economic system to one in which gender is less of an organizational principle. Concurrently changes in learning contexts, resource acquisition, labor scheduling, production scale, product demand, and, ultimately, the finished products have occurred. The interrelationships among these variables are discussed, and the reorganization of production is placed within the context of broader changes in Zuni ecology and political economy. Archaeological implications drawn from the historic Zuni case study include: (1) the diversity of products and production scales that may co-occur within a gender-specific economic system; (2) the material correlates of changes in gendered systems of production; and (3) the interrelationships of changes in nonsubsistence and subsistence production.

A microcomputer cashier scheduling system for supermarket stores

Labour scheduling in supermarkets is a difficult task since most employees work various shift lengths instead of standard eight‐hour shifts. With front‐end cashiers, it is even more difficult, since the number of needed cashiers is based on customer volume, which is constantly fluctuating. To create an accurate schedule, a supermarket manager must spend a good deal of time trying to match cashier shifts to customer demand. Presents a shift selection integer linear programming model which has been implemented in stores of a supermarket chain in north‐east USA. The model runs on a personal computer within an electronic spreadsheet environment, making it user‐friendly for managers. The POS data, routinely collected by registers and scanners are used for forecasting cashiers′ demand. The goal in this scheduling system is to maintain an acceptable customer service at the checkout for customer satisfaction and retention.

The effect of demand characteristics on labour scheduling methods

Labour scheduling heuristic methods have been applied in service operating environments using both actual and synthetic demand patterns. Two important characteristics of these demand patterns are (1) demand smoothness and (2) mean demand. Investigates the effects of demand smoothness and mean demand on the solution quality associated with four prominent heuristic methods. Indicates that both characteristics can affect the performance of the heuristic methods. An especially important finding is that the two methods which use information from linear programming solutions are far more robust to changes in the degree of demand smoothness. Concludes that managers should consider linear programming methods as an alternative or supplement for making their scheduling decisions. Also recommends that labour scheduling researchers use multiple levels of mean demand and demand smoothness when evaluating new heuristic methods.

A microcomputer-based heuristic for tour scheduling of a mixed workforce

Labor tour scheduling research has recently addressed environments which allow both full- and part-time employees with varying cost and productivity levels. Mathematical programming representations of the associated scheduling problems have most frequently been based upon the general set-covering formulation originally suggested by Dantzig. A number of heuristic solution methods appropriate for these problems utilize the LP solution of the general set-covering formulation. However, these methods typically require substantial amounts of CPU time for large problems. Continued increases in problem size may result in mathematical programming formulations which are essentially intractable, regardless of the hardware and software used. In this research, a two-stage heuristic for mixed-workforce tour scheduling problems was implemented on a 386 microcomputer. The performances of the new heuristic were compared to those associated with single-phase linear programming-based procedures on mini computers in two test environments. Using considerably lower CPU times, our new method resulted in mean labor scheduling costs which were only 0.36 above and 0.19% below those obtained by the competing methods for these two environments, respectively. The latter result was especially impressive since the problems in the second test set were considerably more difficult than those in the first test set. These results are likely to be important to service operations managers who are often limited to the use of microcomputer hardware and software.

An interactive, optimization-based decision support system for scheduling part-time, computer lab attendants

The labor tour scheduling problem has attracted much recent research, focusing on the development and evaluation of optimal and heuristic methods to minimize labor costs while satisfying demand for labor. Researchers typically assume that a sufficient labor pool is available. However, service organizations such as fast-food restaurants, grocery stores, and video rental stores, as well as not-for-profit organizations using volunteer workers, typically use a large number of part-time employees with limited availabilities for work. This study presents an interactive decision support system that addresses the conflicting objectives of efficient labor scheduling and accommodating employee needs. The system uses a linear programming model to provide sets of optimal shifts from which employees can construct acceptable weekly schedules; the manager may override the schedule if necessary. The decision support system is used to schedule student computer lab attendants at a major university in an efficient and equitable manner.