In the face of high staffing costs, uncertain patient arrivals, and patients unsatisfied with long wait times, staffing of medical emergency departments (EDs) is a vexing problem. Using empirical data collected from three active EDs, we develop an analytic model to provide an effective staffing plan for EDs. Patient demand is aggregated into discrete time buckets and used to model the stochastic distribution of patient demand within these buckets, which considerably improves model tractability. This model is capable of scheduling providers with different skill profiles who work either individually or in teams, and with patients of varying acuity levels. We show how our model helps to balance staffing costs and patient service levels, and how it facilitates examination of important ED staffing policies.

The problem of no‐shows (patients who do not arrive for scheduled appointments) is particularly significant for health care clinics, with reported no‐show rates varying widely from 3% to 80%. No‐shows reduce revenues and provider productivity, increase costs, and limit patient access by reducing effective clinic capacity. In this article, we construct a flexible appointment scheduling model to mitigate the detrimental effects of patient no‐shows, and develop a fast and effective solution procedure that constructs near‐optimal overbooked appointment schedules that balance the benefits of serving additional patients with the potential costs of patient waiting and clinic overtime. Computational results demonstrate the efficacy of our model and solution procedure, and connect our work to prior research in health care appointment scheduling.

We examine the phenomenon of shifting production bottlenecks fr m an analytic per- spective. We quantify the propensity of a work center to be a bottleneck, efined as maxima! queue length, using a simple Jackson production network model. Compa 'son of the analytic model against an empirical simulation-based model shows that the two are in good agreement. A scalar measure of bottleneck shiftiness is proposed and used to investigate several policies for mitigating shiftiness. Simulation experiments show that several commonly observed managerial policies for coping with shifting bottlenecks actually increase shiftiness. but that shiftiness declines when the capacity of non bottleneck resources is increased. (CAPACITY; BOTTLENECKS; UTILIZATION; QUEUEING NETWORKS) We examine the shifting bottleneck phenomenon that often bedevils operating managers. Shifting bottlenecks occur when the location of the bottleneck work center in a production facility changes with time. At a given moment lone work center will be seriously backlogged, causing production delays, whereas 0 ly hours or days later another work center will be similarly afflicted. Shifting bottl necks create control problems for shop floor personnel, since reactive measures to a eliorate a bottleneck (such as expediting, extra labor, or overtime) require mana ement attention and cause disruption at other work centers. Frequently, when on bottleneck has been brought under control, another bottleneck suddenly appears i~ a totally unexpected location. We show that shifting bottlenecks are in fact an inevitable t eSult of variability or randomness in the production system. To measure and comp re different work cen- ters' contribution to the "shiftiness" of the facility, we first efine the bottleneck probability for each work center as the long-run proportion bf time a given work center has more jobs in its queue than any other. We obtain t~e exact form of these bottleneck probabilities for the well-known Jackson network ~odel and demonstrate

We document the thermal record of breakup of the conjugate Rio Muni (West Africa) and NE Brazil margins using apatite fission track analysis, vitrinite reflectance data and stratigraphic observations from both margins. These results permit determination of the timing of four cooling episodes, and the temperature of samples at the onset of each episode. All samples are interpreted to have experienced higher temperatures in the geological past due to i) elevated basal heatflow (palaeogeothermal gradient in Rio Muni-1 well decaying from 58 degrees C/km during the Mid Cretaceous to 21.5 degrees C/km in the Late Cenozoic) and ii) progressive exhumation from formerly greater burial depth. A well constrained history of changing palaeogeothermal gradient allows for much more precise quantification of the thickness of eroded section (exhumation) than if a constant heatflow is assumed. Cooling episodes identified from the palaeotemperature data at 110-95 Ma (both margins) and 85-70 Ma (Rio Muni only) coincide with major unconformities signifying, respectively, the cessation of rifting (breakup) and compressional shortening that affected the African continent following the establishment of post-rift sedimentation (drift). The interval between these separate unconformities is occupied by allochthonous rafts of shallow-water carbonates recording gravitational collapse of a marginal platform. The rift shoulder uplift that triggered this collapse was enhanced by local transpression associated with the obliquely divergent Ascension Fracture Zone, and thermal doming due to the coeval St Helena and Ascension Plumes. The data also reveal a c.45-35 Ma cooling episode, attributed to deep sea erosion at the onset of Eo-Oligocene ice growth, and a c.15-10 Ma episode interpreted as the record of Miocene exhumation of the West African continental margin related to continent-wide plume development. Integration of thermal history methods with traditional seismic- and stratigraphy-based observations yields a dynamic picture of kilometre-scale fluctuations in base level through the breakup and early drift phases of development of these margins. Major unconformities at ocean margins are likely to represent composite surfaces recording not only eustasy, but also regional plate margin-gene rated deformation, local 'intra-basinal' reorganization, and the amplifying effect of negative feedbacks between these processes. (c) 2008 Elsevier B.V. All rights reserved.

The challenge of balancing the interests of patients with those of healthcare providers is increased when patients fail to show up for scheduled appointments. Overbooking appointments mitigates the lost productivity caused by no-shows but increases patient wait time and provider overtime. In this paper, simulation analysis is used to develop and test the performance of scheduling rules that are designed specifically to accommodate excess overbooked appointments. Our analysis provides new insights into rules that perform well to increase provider productivity while balancing the increased waiting time and overtime costs of overbooked schedules.

Yield and revenue management have been extensively investigated for transportation and hospitality industries, but there has been relatively little study of these topics for appointment services where customers are scheduled to arrive at prearranged times. Such settings included health care clinics; law offices and clinics; government offices; retail services such as tax preparation, auto repair, and salons; counseling centers; and admissions offices, among many others. The problem of no-shows (customers who do not arrive for scheduled appointments) is significant for appointment services, with reported no-show rates varying widely from 3 to 80%. No-shows reduce revenues and provider productivity, increase costs, and limit the ability of the provider to service its customer population by reducing effective capacity. In this paper, we develop an analytic appointment scheduling model that balances the benefits of increased revenues and service with the expected costs of customer waiting and provider overtime. Our results demonstrate that effective appointment overbooking can significantly improve customer service and operations revenues while balancing the potential costs of customer waiting and server overtime.

The problem of patient no-shows (patients who do not arrive for scheduled appointments) is significant in many health care settings, where no-show rates can vary widely. No-shows reduce provider productivity and clinic efficiency, increase health care costs, and limit the ability of a clinic to serve its client population by reducing its effective capacity. In this article, we examine the problem of no-shows and propose appointment overbooking as one means of reducing the negative impact of no-shows. We find that patient access and provider productivity are significantly improved with overbooking, but that overbooking causes increases in both patient wait times and provider overtime. We develop a new clinic utility function to capture the trade-offs between these benefits and costs, and we show that the relative values that a clinic assigns to serving additional patients, minimizing patient waiting times, and minimizing clinic overtime will determine whether overbooking is warranted. From the results of a series of simulation experiments, we determine that overbooking provides greater utility when clinics serve larger numbers of patients, no-show rates are higher, and service variability is lower. Even with highly variable service times, many clinics will achieve positive net results with overbooking. Our analysis provides valuable guidance to clinic administrators about the use of appointment overbooking to improve patient access, provider productivity, and overall clinic performance.

One of the common and important problems in production scheduling is to quote a short and reliable due date to an arriving customer order. For many industries including aerospace, semiconductor, and food, accurate due date quotation (DDQ) is necessary to improve delivery speed, on-time delivery performance, to build a lean and high velocity supply chain, and to minimize tardiness costs. Because of their inherent simplicity, many traditional DDQ methods do not incorporate the complexities involved in the general DDQ problem such as nonlinear and interaction effects and customer preferences. In this dissertation, we examine the DDQ problem from artificial intelligence (AI), economic, and consumer behavior viewpoints. First, we study due date quotation at a tactical level and develop an analytical model to analyze the influence of variability in arrival and service processes on due date, price and order selection decisions of a make-to-order firm. We use a G/G/1 queue to model a firm's operations and show that under some combinations of market characteristics and firm capabilities, it is not profitable for the firm to enter a particular market. We then propose an enhanced DDQ algorithm that uses AI and meta-heuristic principles to improve the DDQ accuracy. Computational experiments reveal that the proposed model consistently outperforms conventional neural network-based DDQ. Finally, we develop a DDQ model of customer satisfaction using consumer behavior and decision making theories and show that DDQ and dispatching policies have the potential to influence firm's reputation and profitability in addition to reducing traditional scheduling costs.

We analyze appointment scheduling and daily patient flow through an outpatient mental health clinic. Using empirical data, we develop several new scheduling heuristics that work to increase patient throughput and clinic productivity, while decreasing patient wait-times. Simulation experiments demonstrate the effectiveness of our methods compared to various other heuristics reported in the practitioner literature. † Linda.LaGanga@colorado.edu ‡ Stephen.Lawrence@colorado.edu Clinical Appointment Scheduling: The Cost of No-Shows and the Value of Overbooking Introduction Cost-effective delivery of health care services requires service providers to attain high levels of productivity (Tonges, 1985) and utilization (Managed Care Weekly Digest, 2003). The failure of some patients to arrive for their scheduled appointments increases provider idle time and reduces the expected number of patients that are actually seen in a day (Shonick and Klein, 1977), which reduces the clinic’s revenues and denies some patients timely access to needed services. As the literature included in the next section demonstrates, some researchers in outpatient appointment scheduling recommend overbooking to solve the problem of no-shows, but have given little or no consideration to the costs. The contribution of this paper is to examine the performance impacts of patient no-shows and overbooking, including a new utility model that includes both the costs and benefits of overbooking and shows clinical providers and administrators how beneficial or costly overbooking would be for the specific operating conditions of their clinics. The rate of patient failures to arrive (“no-show rates,” Barron (1980)) can be significant. The problem may be particularly severe for community mental health centers, pediatric clinics, hospitals, and neighborhood medical and dental clinics (Bean and Talaga, 1995). For example, in an outpatient community mental health center that we observed, almost 30% of adult patients failed to show up for their appointments with psychiatrists. Thus for every 100 scheduled appointments, 30 of the time slots were unused, which reduced provider utilization and productivity. This also means that 30 patients who needed appointments were denied access until a later date, which may reduce customer satisfaction and quality of health care (Chesanow, 1996; Larkin, 1999; Murray and Berwick, 2003). Oct 19, 2005 LaGanga and Lawrence Page 1 of 36 Clinical Appointment Scheduling: The Cost of No-Shows and the Value of Overbooking As Barron (1980) recognized, the problems caused by no-shows may be solved by reducing the occurrence of no-shows or by scheduling to reduce their impact. On one hand, reducing the no-show rate reduces the source of the uncertainty in provider productivity, but may be uncontrollable by the clinic or costly to accomplish. On the other hand, overbooking compensates for uncertainty by boosting productivity but has the potential to harm customer service because of the day-to-day variability in the number of patients that actually show up. When the number of patients who show up exceeds normal system capacity, there are costly increases in patient wait time and the length of the clinical work day. The purpose of this paper is to support decision-making in responding to the problems caused by no-shows. We compare system performance at varying levels of no-show rate with and without overbooking. Using analytical and simulation models, we determine the value of overbooking in terms of expected utility obtained from serving patients, including the costs of patient wait time and overtime operation. We consider non-financial utility (Metters and Vargas, 1999) in our model because providers in not-for-profit health care systems often value serving patients in need more than they value revenue benefits. Therefore this study is relevant both to for-profit and not-for-profit health care providers. We extend previous outpatient scheduling literature by focusing specifically on no-shows and explicitly examining the additional costs incurred by overbooking. We add overtime operation to the cost function used in previous research that focused only on patient wait time and provider idle time. In addition, we analyze scheduling performance for the wide range of realistic no-show rates that have been cited in medical and health care literature to identify the conditions under which overbooking is helpful or harmful to scheduling performance, thus providing guidance in the important decision of whether or not to overbook. Oct 19, 2005 LaGanga and Lawrence Page 2 of 36 Clinical Appointment Scheduling: The Cost of No-Shows and the Value of Overbooking Overview of No-shows and Appointment Scheduling Many authors have contributed to the literature on no-shows from a variety of disciplines and perspectives including medical practice, health care administration, operations management, transportation planning (particularly airline revenue management), and marketing. Health care researchers and some practitioners have focused on finding causes of no-shows and eliminating or reducing them. They consider costs such as analysis of patients and their behavior and the implementation costs of programs or practices to boost patient attendance rates (Bean and Talaga, 1995; Campbell, et al., 2000; Garuda et al., 1998; Shonick and Klein, 1977). Reported reasons for no-shows include lack of transportation, scheduling problems, overslept or forgot, and lack of child care (Campbell, et al., 2000). The probability of patient noshows may relate to factors such as patient age, gender, number of previous appointments (Shonick and Klein, 1977), appointment lead time (Bean and Talaga, 1995) and Medicaid status (Rust et. al, 1995). McCarthy et al. (2000) and Sharp and Hamilton (2001) suggest that no-show rates might increase if wait times grow too long at the clinic. Approaches that have been successfully applied to reduce no-shows include sending patients reminder cards (Rust et al., 1995), calling patients to remind them of appointments, and providing information about public transportation (Bean and Talaga, 1995). Operations management and statistical perspectives are evident in studies of clinical appointment scheduling systems that measure performance as the weighted sum of patient wait time and provider idle time costs (Bailey, 1952; Bailey and Welch, 1953; Ho and Lau, 1992; Welch and Bailey, 1952). These studies identify the no-show rate as a significant factor in schedule performance and measure some of the effects, but do not focus on how to handle noshows or reduce their negative impacts in the scheduling system. Out of 36 articles categorized Oct 19, 2005 LaGanga and Lawrence Page 3 of 36 Clinical Appointment Scheduling: The Cost of No-Shows and the Value of Overbooking in a recent review of outpatient scheduling literature by Cayirli and Veral (2003), only 11 include the possibility of no-shows. Recommendations for handling the problem are even more limited. Only four of the articles reviewed (Blanco White and Pike, 1964; Fetter and Thompson, 1966; Vissers and Wijngaard 1979; Vissers, 1979) include scheduling adjustments or operational considerations such as overbooking to mitigate the effects of no-show behavior. Shonick and Klein (1977) show how to use the probabilities of patient no-shows to overbook enough patients so that the expected number of arrivals is equal to the target number to be seen, but do not consider overtime as a potential risk that could increase costs. Rohleder and Klassen (2002) consider overtime and overbooking as possible methods to deal with temporary or chronic high demand for appointments. They hold the no-show rate constant at 5% and do not include no-shows as an experimental factor in their simulation model of appointment scheduling rules. They also include ending time of the clinical day and server utilization as server-oriented measures of schedule performance, but do not integrate them into an overall performance measure. The medical practitioner literature recommends “wave scheduling,” using variable appointment intervals and patient batch-sizes to build small queues of patients while allowing the provider time to catch up at the end of each period in order to balance provider productivity with patient wait time (Barron, 1980; Baum, 2001; Chesanow, 1996; Chung, 2002; Cole, 2003; McCarthy, 2002; McCord, 1996; Schroer and Smith, 1977; Silver, 1975; Zeff, 1995). But this literature does not differentiate no-shows from varying service times in their impacts on schedule performance. There are large variations in no-show rates among medical specialties and geographic regions (Sharp and Hamilton, 2001) and patient populations and their reasons for no-show Oct 19, 2005 LaGanga and Lawrence Page 4 of 36 Clinical Appointment Scheduling: The Cost of No-Shows and the Value of Overbooking behavior (Garuda et al., 1998). Some studies, including a case study by Brahimi and Worthington (1991) and an official study of hospitals in England and Wales (Warden, 1995), have reported patient no-show rates of 10%. Sharp and Hamilton (2001) reported a 12% noshow rate at outpatient clinics in the UK. According to Barron (1980), eight studies at inner city, community health centers, and university medical centers indicate no-show rates of 10-30% while the estimated no-show rates for private practice are 2-15%. An even wider range of noshow rates, 3-80%, is reported in a study by Rust et al. (1995) of 200 public pediatric clinics. Our overbooking models are designed to handle a wide range of no-show rates, which, as these studies demonstrate, is necessary for the models to be useful in a variety of clinical practices. In contrast to the health care industry, in the airline industry the practice of overbooking to compensate for no-shows has been extensively studied as revenue management to predict and balance the costs and benefits of overbooking (Hilli

Dynamic job shop scheduling research has previously assumed that the probability of the customer placing an order is always one, implying that customers will place orders regardless of lead times quoted from the producer and customer lead time expectations. As firms increasingly compete on the basis of the delivery speed and reputation, the relative performance of quoted versus actual realized lead times will have a strong effect on whether the customer will place future orders or not. Previous research has not considered differing customer requirements or the impact on the customer’s subjective assessment of the overall purchase performance and subsequently on customer satisfaction and repurchases intentions. This paper formally integrates concepts from operations management, marketing, and consumer decision theory into a single due date scheduling framework in order to model the antecedents and consequences of customer satisfaction. We use the results on dynamic priority queues to propose the new way of quoting due dates by explicitly considering the customers lead time requirements. Simulation experiments reveal that job shop policies and machine capacities have significant impact on customer satisfaction and subsequently, on net profit. Finally, we show that dynamic due date quotation policy performs significantly better than several other policies previously tested in the literature.

The team formation problem consists of forming a work team from a large collection of candidates with disparate skills and attributes. The team formation problem is ubiquitous in practice, for example, product development teams formed from marketing, engineering, and finance skills; software development team formed from programmers and software engineers with needed programming and systems skills; and construction management teams formed from architects, civil engineers, designers, and construction engineers. The generally used methods to form team are random assignment, self-selection and facilitator assignment. We formulate the team formation problem as a mixed integer linear goal program. The use of goal programming is appropriate in this context because the constraints of the team formation problem are usually soft or fungible, and can be traded off against one another depending on their priority. Our formulation also allows for the inclusion of “hard” or categorical constraints. Since the team formation problem is known to be NP-Hard, we develop a heuristic solution methodology to rapidly find good solutions to the problem. We adapt the Greedy Randomized Search Procedure (GRASP) to the team formation problem and test on a variety of problems. In addition, we employed standard IP solution software (CPLEX) to solve our set of test problems to optimality, where possible. Preliminary results indicate excellent performance for the GRASP method in this context. For the problems tested, GRASP provided the optimal solution wherever an optimal solution could be found. In several cases for reasonably sized problems (50 team members) an optimal solution could not be identified even after 1 hour of computation. In contrast, the GRASP method found its solutions in less than 3 seconds, even for the largest problems. Future research will test a larger, more complex set of problems, and will verify the effectiveness of the GRASP method in solving the team formation problem across a large range of problem settings. † Fang.Liang@colorado.edu ‡ Stephen.Lawrence@colorado.edu

Decreased budgets, increased staff and faculty responsibilities, and user demand for access to more electronic materials makes determining the costs of providing materials essential for effective library management and decision-making. The literature does not include benchmarks for the functions and processes associated with identifying, selecting, acquiring, organizing, and maintaining paper books (p-books) or electronic books (e-books) for academic library users. This exploratory study begins to identify the resources needed for paper books (p-books) and electronic books (e-books) in the library of today and the library of the future. Eleven ARL librarians estimated labor, space, material, and equipment allocations for the selection, acquisition, cataloging, maintenance, circulation, warehousing and storage, and deselection of p-books and e-books. Nineteen academic librarians completed an abbreviated version of the survey one month after the first data were collected. The results of the second survey are similar to the responses for the estimated resource allocations for space and materials of the ARL librarians who participated in the original study. Although the results cannot be generalized, this study does provide a baseline that can be used for further study of resource allocations for paper and electronic materials in libraries.

An important issue for research librarians is the life cycle cost of acquiring and maintaining a collection. While purchase costs are easy to identify, associated acquisition, cataloging, circulation, and maintenance expenses are difficult to measure and attribute to specific collections. This paper develops a methodology to determine the life cycle costs of collections based on readily available statistical data collected annually by the Association of Research Libraries (ARL). ARL cost data (e.g., salaries and wages, materials expenditures, and operating expenses) for a specific library are allocated to collections (e.g., manuscripts, serials, and microforms) based on the size of the collection and its relative space requirements. By aggregating allocated costs, total life cycle costs for a collection can be estimated. Results of this research indicate that life cycle costs of collections are many multiples of their purchase costs. Results further suggest that the life cycle costs of monograph collections overwhelm the costs of other collections in research libraries-the cost structure of a research library is largely driven by its monograph collection. These results should prove useful in efforts to control costs and improve performance in research libraries.

In this paper we investigate the selection of process technologies under conditions of stochastic market preferences. We assume that the market evolves over time through m states or scenarios defined by the preferences of the market. In response to this evolving marketplace, a producer can respond by switching its facilities to one of t technological states defined by plant capabilities. We model the evolution of market preferences and policies for process selection as a Markov Decision Process and find optimal process adoption policies. In addition to optimal strategies, we define two alternative adoption strategies. “Perfect flexibility” is defined as the increase in profit that can be obtained by instantly matching process technologies to changes in market preferences, compared to a “robust” policy of selecting and employing only a single process technology. With an objective of profit maximization, we show that when the cost of switching production processes is very high, the optimal policy is to select a single robust process and to never switch from it. In contrast, when process-switching costs are zero we show that the optimal policy is perfect flexibility where production processes are immediately matched to market preferences. An optimal production policy provably exists between these two extreme policies. We define the expected value of perfect flexibility as the difference in expected profits between a perfectly flexible policy and a robust policy. The expected value of perfect flexibility provides an upper bound to the benefit of process switching and product flexibility when market preferences are uncertain. Several numerical examples illustrate our findings.

In this paper we investigate the value of perfect flexibility. Perfect flexibility is defined as the increase in profit that can be obtained by responding instantly to changes in market preferences, compared to a “robust” policy of offering only a single product mix over time. We assume that the market evolves over time throughm states or scenarios, defined by the preferences of the market. In response to this evolving marketplace, a producer can respond by switching its facilities to one of t technological states, defined by plant capabilities. We model the evolution of market preferences and policies for process selection as a Markov Decision Process. With an objective of profit maximization, we show that when the cost of switching production processes is very high, the optimal policy is to select a single “robust” process and to never switch from it. In contrast, when process switching costs are zero, we show that the optimal policy is “perfect flexibility” where production processes are immediately matched to market preferences. An optimal production policy will exist between these two extreme policies. We define the value of perfect flexibility as the difference in expected profits between 1Address: College of Business and University of Colorado, Boulder, CO 80309-0419, phone: (303) 492-4351 2Corresponding author. Address: Department of Business Administration, University of Illinois at Urbana-Champaign, 1206 South Sixth Street, Champaign, IL 61820, phone: (217) 333-8270 3Address: Babcock Graduate School of Business, Wake Forest University, Wake Forest, NC 27109-7659, phone: (910) 759-4423

In this paper we compare the static and dynamic application of heuristic and optimal solution methods to job‐shop scheduling problems when processing times are uncertain. Recently developed optimizing algorithms and several heuristics are used to evaluate 53 standard job‐shop scheduling problems with a makespan objective when job processing times are known with varying degrees of uncertainty. Results indicate that fixed optimal sequences derived from deterministic assumptions quickly deteriorate with the introduction of processing time uncertainty when compared with dynamically updated heuristic schedules. As processing time uncertainty grows, we demonstrate that simple dispatch heuristics provide performance comparable or superior to that of algorithmically more sophisticated scheduling policies.