Models and Approximations for Call Center Design

A call center is a facility for delivering telephone service, both incoming and outgoing. This paper addresses optimal staffing of call centers, modeled as M/G/n queues whose offered traffic consists of multiple customer streams, each with an individual priority, arrival rate, service distribution and grade of service (GoS) stated in terms of equilibrium tail waiting time probabilities or mean waiting times. The paper proposes a methodology for deriving the approximate minimal number of servers that suffices to guarantee the prescribed GoS of all customer streams. The methodology is based on an analytic approximation, called the Scaling-Erlang (SE) approximation, which maps the M/G/n queue to an approximating, suitably scaled M/G/1 queue, for which waiting time statistics are available via the Pollaczek-Khintchine formula in terms of Laplace transforms. The SE approximation is then generalized to M/G/n queues with multiple types of customers and non-preemptive priorities, yielding the Priority Scaling-Erlang (PSE) approximation. A simple goal-seeking search, utilizing SE/PSE approximations, is presented for the optimal staffing level, subject to GoS constraints. The efficacy of the methodology is demonstrated by comparing the number of servers estimated via the PSE approximation to their counterparts obtained by simulation. A number of case studies confirm that the SE/PSE approximations yield optimal staffing results in excellent agreement with simulation, but at a fraction of simulation time and space.     

Using mixture of Gamma distributions for Bayesian analysis in an M/G/1 queue with optional second service

The paper proposes Bayesian framework in an M/G/1 queuing system with optional second service. The semi-parametric model based on a finite mixture of Gamma distributions is considered to approximate both the general service and re-service times densities in this queuing system. A Bayesian procedure based on birth-death MCMC methodology is proposed to estimate system parameters, predictive densities and some performance measures related to this queuing system such as stationary system size and waiting time. The approach is illustrated with several numerical examples based on various simulation studies.     

A decision support model for establishing an air taxi service: a case study

Recent availability of relatively cheap small jet aircraft creates opportunities for a new air transport business: Air taxi, an on-demand service in which travellers call in one or a few days in advance to book transportation. In this paper, we present a methodology and simulation study supporting important strategic decisions, like for instance determining the required number of aircraft, for a company planning to establish an air taxi service in Norway. The methodology is based on a module simulating incoming bookings, built around a heuristic for solving the underlying dial-a-flight problem. The heuristic includes a separate method for solving the important subproblem of determining the best distribution of waiting time along a single aircraft schedule. The methodology has proved to provide reliable decision support to the company.     

Modelling variations in hospital service delivery based on real time locating information

Variations in service delivery have been identified as a major challenge to the success of process improvement studies in service departments of hospital such as radiology. Largely, these variations are due to inherent system level factors, i.e., system variations such as unavailability of resources (nurse, bed, doctors, and equipment). These system variations are largely unnecessary/unwarranted and mostly lead to longer waiting times, delays, and lowered productivity of the service units. There is limited research on identifying system variations and modelling them for service improvements within hospital. Therefore, this paper proposes a modelling methodology to model system variations in radiology based on real time locating system (RTLS) tracking data. The methodology employs concepts from graph theory to identify and represent system variations. In particular, edge coloured directed multi-graphs (ECDMs) are used to model system variations which are reflected in paths adopted by staff, i.e., sequence of rooms/areas traversed while delivering services. The main steps of the methodology are: (i) identifying the most standard path followed by staff for service delivery; (ii) filtering the redundant events in RTLS tracking database for analysis; (iii) identifying rooms/areas of hospital site involved in the service delivery; (iv) determining patterns of paths adopted by staff from filtered tracking database; and, (v) representation of patterns in graph based model called as edge coloured directed multigraphs (ECDMs) of a role. A case study of MR scanning process is utilized to illustrate the implementation of the proposed methodology for modelling system variations reflected in the paths adopted by staff.     

Bank-office Process Management in the Financial Services: a Simulation Approach Using a Model Generator

For the last ten years or so, developments in simulation software and methodology have been focused in two areas: (1) automating the model-building process; and (2) interfacing with other techniques and systems. That both academic and commercial research continues to be carried out in these areas is evidence of the continued popularity of the simulation technique. The most popular application areas where simulation modelling is used include manufacturing (engineering and process), warehousing and communications; not surprisingly, research and developments have therefore been concentrated in these areas. This paper describes the development of a simulation model generator for a different application area, that of clerical office processing of paper and documents for financial service companies. It outlines the reasons why there is a need for such a model generator and discusses the initial design requirements. The paper goes on to detail the developments and re-defined criteria required as the model generator brief changes as a result of practical applications.     

Developing railway timetables which guarantee a better service

In order to improve passenger service, a waiting cost function, weighting different types of waiting times and late arrivals, is designed and minimised. The approach is applied to a small part of the Belgian railway network. In the first phase of the approach, ideal buffer times are calculated to safeguard connections when the arriving train is late. These buffer times are based on the delay distributions of the arriving trains and on the weighting of different types of waiting times. In a second phase, standard linear programming is used to construct an improved timetable with well-scheduled connections and, whenever possible, with ideal buffer times. Simulation compares different timetables and optimises the LP timetable. For the case of the Belgian railway network, the final result is a timetable with well-scheduled connections and a waiting cost that is 40% lower than the current timetable. Since only LP modelling is applied, the proposed technique is very promising for developing better timetables—even for very extensive railway networks.     

A lot sizing model with queueing delays: The issue of safety time

The objective of this paper is to introduce the concept of safety time in a make-to-order production environment. The production facility is represented as a queueing model, explicitly including a non-zero setup time. A methodology is presented to quantify the safety time and to compute the associated service level based on the queueing delay. The main result is a convex relationship of the expected waiting time, the variance of the waiting time and the quoted lead time as a function of the lot size and a concave relationship of the service level as a function of the lot size. Most models in the literature assume batch arrivals. We relax that assumption so that an individual customer arrival process is allowed. We therefore have to derive a new closed form analytical expression for the expected waiting time. Both the deterministic and a stochastic case are studied.     

A single server queue with cyclically indexed arrival and service rates

In this paper we consider a queueing model that results from at least two apparently unrelated areas. One motivation to study a system of this type results from a test case of a computer simulation factor screening technique calledfrequency domain methodology. A second motivation comes from manufacturing, where due to cyclic scheduling of upstream machines, the arrival process to downstream machines is periodic. The model is a single server queue with FIFO service discipline and exponential interarrival and service times where the arrival and/or service rates are deterministic cyclic functions of the customer sequence number. We provide steady state results for the mean number in the system for the model with cyclic arrival and fixed service rates and for the model with fixed arrival and cyclic service rates. For the model with both cyclic arrival and service rates, upper and lower bounds are developed for the steady state mean waiting time in the system. Throughout the paper various implications and/or insights derived from the results of this study are discussed for frequency domain methodology.The authors acknowledge the financial support of the CBA/GSB Faculty Research Committee of the College of Business Administration, The University of Texas at Austin.     

Minimising average passenger waiting time in personal rapid transit systems

Personal Rapid Transit (PRT) is an emerging urban transport mode. A PRT system operates much like a conventional hackney taxi system, except that the vehicles are driven by computer (no human driver) between stations in a dedicated network of guideways. The world’s first two PRT systems began operating in 2010 and 2011. In both PRT and taxi systems, passengers request immediate service; they do not book ahead. Perfect information about future requests is therefore not available, but statistical information about future requests is available from historical data. If the system does not use this statistical information to position empty vehicles in anticipation of future requests, long passenger waiting times result, which makes the system less attractive to passengers, but using it gives rise to a difficult stochastic optimisation problem. This paper develops three lower bounds on achievable mean passenger waiting time, one based on queuing theory, one based on the static problem, in which it is assumed that perfect information is available, and one based on a Markov Decision Process model. An evaluation of these lower bounds, together with a practical heuristic developed previously, in simulation shows that these lower bounds can often be nearly attained, particularly when the fleet size is large. The results also show that low waiting times and high utilisation can be simultaneously obtained when the fleet size is large, which suggests important economies of scale.     

Simulation optimization for an emergency department healthcare unit in Kuwait

This paper integrates simulation with optimization to design a decision support tool for the operation of an emergency department unit at a governmental hospital in Kuwait. The hospital provides a set of services for different categories of patients. We present a methodology that uses system simulation combined with optimization to determine the optimal number of doctors, lab technicians and nurses required to maximize patient throughput and to reduce patient time in the system subject to budget restrictions. The major objective of this decision supporting tool is to evaluate the impact of various staffing levels on service efficiency. Experimental results show that by using current hospital resources, the optimization simulation model generates optimal staffing allocation that would allow 28% increase in patient throughput and an average of 40% reduction in patients’ waiting time.     

Developing a platform for comparison of hospital admission systems: An illustration

There is an increasing need to develop a platform for comparing hospital admission planning systems due to a shift in the service paradigm in the health sector. The current service concept of hospital admission planning aims at optimising the use of scarce hospital resources without paying much attention to the level of service offered to patients. As patients nowadays do not accept long waiting times for hospital admission, it becomes necessary to consider alternative admission service concepts. Waiting lists have also become a political issue, and alternative concepts have been advocated such as giving all patients an appointment for admission. A simulation model was built to examine the impacts of extreme admission service concepts in a simplified hospital setting. The alternative concepts considered are based on the ‘zero waiting time’ principle (immediate treatment), and the ‘booked admissions’ principle (using an appointment for admission). The results of these admission service concepts are compared with the results of the current concept, based on the ‘maximising resource use’ principle. The paper deals with the development of a framework and tool that allows evaluating different, somehow conflicting, hospital admission planning concepts and the usefulness of such framework and tool for more refined/real-life approaches to hospital admission planning.     

Equivalent single-queue–single-server model for a Pentium processor

In this paper, a Pentium processor is represented as a queuing network. The objective of this paper is to deduce an equivalent single-queue–single-server model for the original queuing network. Closed-form expressions for the equivalent service rate, equivalent queue lengths, equivalent response and waiting times of the equivalent single-queue–single-server model are derived and plotted. For large values of arrival rate, queue lengths increase faster than the response times and waiting times for both the cases. Performance measures like, queue lengths, response times and waiting times are higher for lower service rates and lower for higher service rates (which is expected) of the different servers in the original queuing network. Also, the reliability in estimating performance measures for homogeneous workloads is much better than that for heterogeneous workloads.     

Waiting times in a two-queue model with exhaustive and Bernoulli service

This paper deals with waiting times in a two-queue polling system in which one queue is served according to the Bernoulli service discipline and the other one attains exhaustive service. Exact results are derived for the LST's of the waiting time distributions via an iteration scheme. Based on those results the mean waiting times are expressed in the system parameters.     

Dynamic multi-priority,multi-class patient scheduling with stochastic service times

Efficient patient scheduling has significant operational, clinical and economical benefits on health care systems by not only increasing the timely access of patients to care but also reducing costs. However, patient scheduling is complex due to, among other aspects, the existence of multiple priority levels, the presence of multiple service requirements, and its stochastic nature. Patient appointment (allocation) scheduling refers to the assignment of specific appointment start times to a set of patients scheduled for a particular day while advance patient scheduling refers to the assignment of future appointment days to patients. These two problems have generally been addressed separately despite each being highly dependent on the form of the other. This paper develops a framework that incorporates stochastic service times into the advance scheduling problem as a first step towards bridging these two problems. In this way, we not only take into account the waiting time until the day of service but also the idle time/overtime of medical resources on the day of service. We first extend the current literature by providing theoretical and numerical results for the case with multi-class, multi-priority patients and deterministic service times. We then adapt the model to incorporate stochastic service times and perform a comprehensive numerical analysis on a number of scenarios, including a practical application. Results suggest that the advance scheduling policies based on deterministic service times cannot be easily improved upon by incorporating stochastic service times, a finding that has important implications for practice and future research on the combined problem.     

Out-patient Queues at the Ibn-Rochd Health Centre

An arbitrary policy of fixing the number of outpatient appointments specifying the dates (and not the exact times) of appointments created long queues and large waiting times in some departments of the Ibn-Rochd health centre, but considerable idle time for the consulting doctors in others. After narrating in detail these circumstances and examining a number of possible options, this paper describes the scientific approaches made to determine the number of appointments, the corresponding parameters of the queue and the system and the service per cent occupation.Finite source size, a random number of initial patients, group arrivals, non-exponential service time distribution, late start-up of the servicing unit and many other factors combined to render available theoretical results difficult to apply and results obtained by applying approximately equivalent theoretical models unreliable as compared with those observed in real life. It is shown that simulation could be more profitably applied in such situations.     

Optimal Sequencing of Servers in the Two-Station Series Queueing System

This paper deals with determining an optimal sequence of service stations in a series queueing system. Optimality is defined in terms of the total time spent waiting for service. Sequences are compared on the basis of the moments of their steady-state total waiting time. In addition, the rules of stochastic dominance are applied which allow comparison of sequences on the basis of their waiting time distributions. Analytical results in the sequencing of service stations in series queues have been limited to stations with constant or exponential service times. This study extends the investigation to service distributions with varying degrees of statistical regularity given by the family of Erlang distributions.Relationships are developed for predicting optimal sequences. Validation is accomplished by simulating a number of systems and comparing the waiting time distribution functions for each sequence. The relationships are shown to be good predictors and useful in the study and design of systems of servers in series.     

Pseudo-conservation laws in cyclic-service systems with a class of limited service policies

This paper considers a cyclic-service system with a class of limited service policies that consists of exhaustive limited, gated limited and general decrementing policies. Under these policies, the number of customers served consecutively during a server visit is limited by a vector of integers. The major results in this paper are derivations of expected amount of work left in the queues at the server departures for these three policies. Exact expressions of weighted sum of mean waiting times, known as pseudo-conservation laws, are subsequently obtained. The conservation laws for this class of policies contain unknown boundary probabilities. We estimate these probabilities using corresponding server vacation models. Numerical results presented for the exhaustive limited policy are noted to be very accurate compared with simulation results. Moreover, we have obtained analytical bounds for the weighted sums. Finally, we present a conservation law with mixed service policies, and mean waiting times for symmetric systems.This work was completed while the author was in the Ph.D. program at Rensselaer Polytechnic Institute.This work was partially supported by the Center for Advanced Technologies of the New York State.     

A simulation study of scheduling clinic appointments in surgical care: individual surgeon <Emphasis Type="Italic">versus</Emphasis> pooled lists

The purpose of this paper is to compare two methods of scheduling outpatient clinic appointments in the setting where the availability of surgeons for appointments depends on other clinical activities. We used discrete-event simulation to evaluate the likely impact of the scheduling methods on the number of patients waiting for appointments, and the times to appointment and to surgery. The progression of individual patients in a surgical service was modelled as a series of updates in patient records in reaction to events generated by care delivery processes in an asynchronous fashion. We used the Statecharts visual formalism to define states and transitions within each care delivery process, based on detailed functional and behavioural specifications. Our results suggest that pooling referrals, so that clinic appointments are scheduled with the first available surgeon, has a differential impact on different segments of patient flow and across surgical priority groups.     

Data-stories about (im)patient customers in tele-queues

Credible queueing models of human services acknowledge human characteristics. A prevalent one is the ability of humans to abandon their wait, for example while waiting to be answered by a telephone agent, waiting for a physician’s checkup at an emergency department, or waiting for the completion of an internet transaction. Abandonments can be very costly, to either the service provider (a forgone profit) or the customer (deteriorating health after leaving without being seen by a doctor), and often to both. Practically, models that ignore abandonment can lead to either over- or under-staffing; and in well-balanced systems (e.g., well-managed telephone call centers), the “fittest (needy) who survive” and reach service are rewarded with surprisingly short delays. Theoretically, the phenomenon of abandonment is interesting and challenging, in the context of Queueing Theory and Science as well as beyond (e.g., Psychology). Last, but not least, queueing models with abandonment are more robust and numerically stable, when compared against their abandonment-ignorant analogues. For our relatively narrow purpose here, abandonment of customers, while queueing for service, is the operational manifestation of customer patience, perhaps impatience, or (im)patience for short. This (im)patience is the focus of the present paper. It is characterized via the distribution of the time that a customer is willing to wait, and its dynamics are characterized by the hazard-rate of that distribution. We start with a framework for comprehending impatience, distinguishing the times that a customer expects to wait, is required to wait (offered wait), is willing to wait (patience time), actually waits and felt waiting. We describe statistical methods that are used to infer the (im)patience time and offered wait distributions. Then some useful queueing models, as well as their asymptotic approximations, are discussed. In the main part of the paper, we discuss several “data-based pictures” of impatience. Each “picture” is associated with an important phenomenon. Some theoretical and practical problems that arise from these phenomena, and existing models and methodologies that address these problems, are outlined. The problems discussed cover statistical estimation of impatience, behavior of overloaded systems, dependence between patience and service time, and validation of queueing models. We also illustrate how impatience changes across customers (e.g., VIP vs. regular customers), during waiting (e.g., in response to announcements) and through phases of service (e.g., after experiencing the answering machine over the phone). Our empirical analysis draws data from repositories at the Technion SEELab, and it utilizes SEEStat—its online Exploratory Data Analysis environment. SEEStat and most of our data are internet-accessible, which enables reproducibility of our research.     

Single-Server Queue with Markov-Dependent Inter-Arrival and Service Times

In this paper we study a single-server queue where the inter-arrival times and the service times depend on a common discrete time Markov chain. This model generalizes the well-known MAP/G/1 queue by allowing dependencies between inter-arrival and service times. The waiting time process is directly analyzed by solving Lindley's equation by transform methods. The Laplace–Stieltjes transforms (LST) of the steady-state waiting time and queue length distribution are both derived, and used to obtain recursive equations for the calculation of the moments. Numerical examples are included to demonstrate the effect of the autocorrelation of and the cross-correlation between the inter-arrival and service times. An erratum to this article is available at .