Example Application

To illustrate the features of BAL and its GIS system, a simple example of an M&R program is presented. The program resembles a school board having a M&R program involving 9 schools. First, the nine schools were selected from a spreadsheet having all school locations at the Greater Toronto Area in Canada (Figure 2). Each school is defined by its address (recognizable by the GIS software). For each site in the list, various pre-planning information was stored, including traffic volume and local productivity factors (monthly values that depend on weather, site access, and other conditions). These productivity factors will be used in the scheduling process to adjust activities’ durations, thus, providing realistic execution conditions. As shown in Figure 2, the GIS automatically reads the sites’ information and shows a map of their locations and other related information. Various types of data representations are available, such as pie chart, column chart (shown) or sized circles. These data representations provide simple and legible options to compare the values (e. g., productivity factors) at the various sites.

In the present M&R example, 12 activities are involved in each typical site, as shown in Figure 3. The work quantities and three estimates that represent normal, overtime, or weekend work options are shown in the same figure. While the cheaper estimate is the default in generating an initial plan, the other options (faster but more expensive) may be necessary in case durations need to be

Vary the No. of Crews: 1*

Vary the Order of Site Execution: Г*

tQft « SftC *2.927

Consider Finish Constraints: I*

lortiacted to Невад Ігк Irom Feb 15 2003 Mar 11.2003

Fig. 6. le optimization.

shortened to meet a strict deadline. In this case, finding the proper combinations of work options for the activities becomes part of the optimization feature.

Before producing a detailed schedule, the user needs to enter some general data such as start date (February 11, 2003, as shown in Figure 4) and the execution constraints such as deadline (March 12, 2003), incentive ($10,000/day), liquidated damages ($100,000), and any resource limits. As such, this example M&R program has only 22 working days (excluding weekends) to complete the 9 sites.

With the default option being to use in-house crews, the next step was to enter information regarding any special sites that were decided to be outsourced to contractors. In this example, two of the nine sites were pre-specified as outsourced. Once all site-specific information was specified, a detailed execution schedule was generated, as shown in Figure 5.

The initial schedule shows the various color-coded crews. The two contracted sites are shown at the top part of the schedule. It is noted that the initial site order is the one entered by the user in Figure 2. Based on that site order, project duration and cost (considering crew-moving time and cost) were automatically calculated, taking into account the site productivity factors. This resulted in a schedule of 26 days, as opposed to the 22-day deadline. Accordingly, the cost becomes $283,882, plus liquidated damages of $400,000, for a total of $683,882, as shown in Figure 5.

To improve the schedule, it is possible to manually change the crews, work options, and site order to try to meet the deadline and reduce cost (user modifiable options shown in Figure 5). Rather than

North America Canada Ontario Toronto

Stlecroft PS 50 Slilecroft Dr. _*

Start: Feb. 19,2003

EUaydon PS 25 Blaydon Ave. *_

Start: Feb. 14,2003

Crew 1_________________

Crew 1_________________ I

Cru

Щ

bsnctt Don

York Maintenance Shop 40… JL

Start: Feb. 11,2003

Crew 1_________________

Fig. 7. isignment map.

this manual adjustment process, it is possible to activate the optimization feature, and accordingly, a form for specifying optimization options appears (Figure 6) and allows the user full control over the optimization variables, to suit the project objectives. For this example, various optimization experiments were conducted. The optimum schedule is shown in Figure 6. Project duration of 21 days meets the deadline at a total cost of $334,299.

Once a satisfactory schedule was obtained, site-specific schedule data were automatically ex­ported to the GIS system to provide legible maps of detailed movement of the various crews. An example map generated by the GIS system for the present example is shown in Figure 7. The figure shows two color-coded paths for two crews involved in an activity. The start time for each site is also shown on each site. As such, the map is simple to read and shows when each crew needs to start in which site. An alternative chart that shows detailed travel directions, which can also be generated from the GIS system.

Conclusions

This paper introduced a scheduling model and implementation software, BAL, for optimizing re­source allocation in infrastructure projects with multiple-distributed locations. Two unique aspects of the program are discussed in this paper: (1) the powerful scheduling engine that optimizes the execution plan; and (2) the underlying Geographic Information System (GIS).

The GIS system stores and represents various levels of information about the scattered sites involved in a construction/maintenance program. Pre-planning information includes location, pro­ductivity factors, land survey data, and traffic volume, etc. Using this information, the GIS system automatically calculates the distances from one site to any other, considering the shortest travel routes. Based on the distances, the GIS system calculates the travel time from each site to any other, considering the speed limits specified for the highways or local roads along a route. These are then directly used to determine the time and cost to transport resources from one site to the other.

During the planning stage, BAL’s scheduling engine stores the user-input data of available re­sources, construction methods, and the time/cost/other constraints. The scheduling engine then runs a Genetic Algorithm to optimize the number of crews to use, the site order, and the set of construction methods. Based on the optimized schedule, another layer of GIS information is generated; containing activities’ start and finish dates at the various sites along with the assigned crews. This layer of information is then used to represent the work assignment in a legible manner to the various project participants. One of the outputs is a legible crew assignment map.

An example application was used to illustrate the benefits of using GIS to support schedule computation and better visualization of multiple-site multiple-crew execution plans. The proposed computer program is potentially usable by municipalities and owner/contractor organizations ad­ministering a large number of infrastructure assets, such as buildings, highways, and bridges, etc.

Reference

Hegazy, T., Elhakeem, A., and Elbeltagi, E. (2004) Distributed Scheduling Model for Infrastructure Networks. Journal of Construction Engineering and Management, ASCE, 130(2), 160-167.