Content-Length: 172509 | pFad | https://www.academia.edu/127470552/Field_workforce_scheduling

(PDF) Field workforce scheduling
Academia.eduAcademia.edu

Field workforce scheduling

2003, Bt Technology Journal

BT employs thousands of field engineers across the UK to maintain networks, repair faults and provide service to customers. To allocate work more efficiently, BT launched Work Manager in the early 1990s — an information system automating work management and field communications, and now marketed as a.p.solve's TASKFORCE. In 1996, BT Exact's Intelligent Systems Lab enhanced Work Manager with a

Field workforce scheduling D Lesaint, C Voudouris, N Azarmi, I Alletson and B Laithwaite † BT employs thousands of field engineers across the UK to maintain networks, repair faults and provide service to customers. To allocate work more efficiently, BT launched Work Manager in the early 1990s — an information system automating work management and field communications, and now marketed as a.p.solve’s TASKFORCE. In 1996, BT Exact’s Intelligent Systems Lab enhanced Work Manager with a Dynamic Scheduler (DS) combining heuristic search and constraint-based reasoning. Since its national roll-out in 1997, DS has consistently reduced operational costs while preserving high quality of service. This paper gives an overview of BT’s workforce scheduling problem, the DS system, and its operational and commercial impact. 1. Introduction to the field workforce scheduling problem Field workforce scheduling is about sending the right engineer to the right customer at the right place at the right time with the right equipment — at any time and in any operational environment. For BT, which employs over 30 000 field engineers, workforce scheduling is critical and its ability to provide high-quality service while maximising productivity and minimising operational costs is vital to the company’s success and competitiveness. For practical and historical reasons, BT manages its field operations in a decentralised way that reflects the geographical and skill distribution of its workforce. Field operations are the responsibility of two separate divisions, each with distinct skill demands: BT Retail, and BT Wholesale. Each division is itself decomposed in a number of groups that operate within non-overlapping geographical domains. For instance, the 15 000-person workforce of BT Retail consists of 100 groups. Each group manages its domain alone and controls from a dozen to a few hundred engineers who perform from a few hundred to a few thousand tasks per day. All domains are subject to similar allocation constraints and corporate rules, for example: • each engineer operates within a predefined area, • each engineer has off-hours and predefined breaks during the day (e.g. lunch, personal appointments), • tasks must be performed within an agreed time window or must start or end by an agreed time, • engineers must be matched to tasks, † a.p.solve • • task duration depends on the engineer’s skills, • task execution may be broken according to rules that take into account task and break details. some tasks must be sequenced in time, others must be performed in parallel by different engineers, At the same time, scheduling decisions must satisfy corporate objectives and customer preferences such as: • maximising the productivity of the workforce, • improving service quality, measured according to the type of customer, the priority of the work (provision, installation, maintenance, repair), and any service level agreement, • making best utilisation of skills, • minimising the operational costs resulting from travel time, waiting time, and work in overtime, • satisfying work controllers’ and engineers’ preferences as to location and types of assignments. Constructing feasible and good-quality work schedules under these conditions is difficult. In fact, the problem may be viewed as a complex variant of the vehicle-routing problem featuring multiple vehicles, multiple depots, heterogeneous constraints, and conflicting objectives. The problem is further compounded by the inherent instability of the environment, since much of the schedule information is uncertain, imprecise or incomplete. For instance, BT or its customers may request, cancel, or amend jobs unpredictably; engineer availability is subject to last-minute changes and estimates of task duration and travel time cannot be totally accurate; work controllers may modify work assignments and review objectives at any time; the environment itself (weather, traffic conditions) is unpredictable. BT Technology Journal • Vol 21 No 4 • October 2003 23 Field workforce scheduling Therefore, any scheduling process must continuously incorporate new data to generate valid work assignments but it must also minimise the impact of this data on the current work schedule. Anyone interacting with the workallocation system must indeed be provided with schedule information that is as stable as possible over time. 2. dynamic scheduler scheduler for simple tasks on-line allocator Dynamic Scheduler In 1992, BT introduced Work Manager (WM) [1] — an information system automating work-management and field-communication processes (see Fig 1). WM was initially equipped with a real-time allocation algorithm (RTA) [2]. Each geographical domain ran an instance of the RTA every five minutes to identify and assign the next task to each engineer. Rolled out in 1993, the system made tangible improvements in productivity and quality of service. However, the RTA was clearly not optimal for situations that called for many scarce skills, for long tasks, or for tasks with interdependencies — all of which required greater look-ahead to schedule and allocate efficiently. BT needed a scheduler that would be as good as the RTA for very dynamic workloads with short response times but much better for more complex work and mixed workloads requiring greater look-ahead. Rule-based scheduling applications available in the market from companies like ClickSoftware, ServicePower and Telcordia simply could not address BT’s scalability and field service complexity requirements. In 1995, a cross-divisional team led by BT Exact’s Intelligent Systems Lab and involving the BT Managerial and Science Consultancy Unit (MSCU) and Work Management team (now a.p.solve [3] who currently market Work Manager as ‘TASKFORCE’) was set up to design and develop a new dynamic scheduling system called Dynamic Scheduler (DS) [4]. DS was based on two principles: • coupling an on-line allocator and a predictive scheduler to preserve responsiveness while benefiting from global optimisation (Fig 2), • using a generic and efficient predictive scheduling model based on constraint optimisation technology. scheduler for complex tasks schedule manager work management system Fig 2 Dynamic Scheduler architecture. The use of constraint-based reasoning (a field of artificial intelligence [5]) could ensure generating realistic schedules and encountering no limitations in scope if problem specifications were to evolve (as they did for the RTA). Yet, a pure constraint-based scheduler was inappropriate to tackle the optimisation dimension of the problem. The ideal approach would combine proven optimisation techniques with constraint-based reasoning. Heuristic search was chosen because of its track record in vehicle routing [6]. After running comparative tests on simulated annealing, tabu search and genetic algorithms, the first algorithm was selected based on efficiency, simplicity, and robustness criteria. The challenge in integrating simulated annealing and constraint-based reasoning was to preserve the former’s performance without losing the exactness and extensibility of the constraint-based schedule model. For this reason, a systematic algorithm was developed to preschedule the most-constrained tasks — the long-duration tasks (typically, over eight hours) and the interdependent tasks (tasks that must be synchronised to satisfy a complex work request). This algorithm performs a tree-search and is work allocation job management field dispatch fixed BT Wholesale customer service system inter-system communications network BT Retail other applications Fig 1 24 visualiser BT Technology Journal • Vol 21 No 4 • October 2003 Overview of Work Manager. user-interface communications network hand-held terminal laptop








ApplySandwichStrip

pFad - (p)hone/(F)rame/(a)nonymizer/(d)eclutterfier!      Saves Data!


--- a PPN by Garber Painting Akron. With Image Size Reduction included!

Fetched URL: https://www.academia.edu/127470552/Field_workforce_scheduling

Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy