Optimizing Preventive Maintenance

GIS Gives Elevator Company a Lift

Using an elevator or escalator as part of our daily lives, we don't stop to think about the driving forces that guarantee safe and continuous operation of these conveniences. Every elevator and escalator needs preventive maintenance to operate reliably and economically. Highly specialized technicians, directed by a superintendent who is responsible for service contracts in his/her area, perform this service work. A worldwide giant in the elevator/escalator Business is Schindler Elevator Corporation. Schindler and its division, Millar Elevator Service Company, are the United States leaders in this competitive service Business. The Swiss-based company tasked Esri to build a planning and scheduling system to optimize their preventive maintenance. After only seven months of development, the Planning Assistant for Superintendent Scheduling (PASS) system went into successful operation in August 1999.

Background

Schindler and Millar operate about 140 offices in Canada and the United States. The offices cover a wide range of service areas, from highly dense areas such as Manhattan to territories covering multiple states.

Service work for each technician is scheduled in daily routes, with emphasis on geographically tight areas of operation to allow quick response in case of emergency calls (if you're stuck in an elevator, you're hoping the technician won't have to drive three hours to get there).

Technicians usually service many buildings. Every piece of equipment has a preassigned periodic maintenance schedule for greasing, cleaning, etc. Schedules include 10 or more visit periods ranging from daily to quarterly. Additionally, each building has a visit schedule influenced by customers' preferences and including hard constraints (it is generally not well received if an elevator is taken out of service during the busiest time of the day). All U.S. and Canada service contract information is stored in a centralized SAP R/3 Enterprise Resource Planning system in Morristown, New Jersey.

The Challenge

Schindler's first priority was user friendliness and ease of use. These requirements led naturally to a GIS-based system with a street map to display all locations and routes. Schindler needed a system capable of not only handling scheduling but also planning workloads between technicians and superintendents. The requirement was to unassign and reassign jobs (locations) from one technician to another while using a weighted function that considers overtime, contract dollars, work hours, or travel between locations. The system had to be able to restructure a complete office with hundreds of locations and tens of thousands of tasks. A bar chart graph was required to simplify comparison of workloads between routes. Data had to be downloaded from the SAP R/3 system and uploaded after modification. The response time had to be fast. The system had to be installed on about 500 office PCs using Windows 95, Windows 98, and Windows NT.

The most challenging part was the development of a periodic routing algorithm that would consider technician skill sets, technician working schedule, number of available technicians, customer requirements (skills, time windows, visiting day), visit frequencies, and visit duration.

The Solution

Esri decided on a client/server architecture using a MapObjects client with local street map storage to prevent overloading the dial-up network connections between offices and Morristown.

PASS offers three main views: a list view with tabular information about locations, routes, and visit schedules; a bar chart view indicating route data for profitability, workload, travel time, and contract value; and a map view including various routes in predetermined colors and symbols. The user can drag and drop locations to new routes and query jobs via a mouse-click.

The PASS server is connected to SAP R/3. Downloaded data from SAP is stored on the PASS server in an Oracle database. Superintendents can modify their data using the PASS client interface to match their needs and then upload (save) it back into the SAP R/3 system. The brain of the PASS system is an optimization algorithm (weighted function) that will produce an optimized result using parameters that can be changed by the user.

The parameters include travel time, balance of contract value or contract hours, overtime, free time between jobs (for emergency calls), plus overtime and time window violation (customer visit constraints). PASS uses NetEngine for travel time calculation between locations.

The Result

Schindler ordered PASS to increase efficiency in route planning. Using the ability to view customer locations and job assignments spatially allowed them to restructure offices by reassigning customers between routes and superintendents.

The new PASS work schedules are now downloaded to remote field units to inform every technician about daily schedules and tasks.

According to an article published by Fortune (Small Business Edition 374[C], November 1999), Schindler estimates they will save several million dollars annually by using the new GIS-based PASS system.

For more information, contactWolfgang Hall, project manager (e-mail: whall@esri.com).

ArcNews home page