PSR Group Ltd

ROUTE
OPTIMIZATION

Simulations show ways to boost plow
productivity up to 90%

Can off-the-shelf software help develop a more cost-effective roadway snow plowing route? In simulations in Ottawa, operators plowed software-optimized routes from 60 to 90 % faster.

W. Maxwell Miner, P.Eng., is a project engineer with the City of Ottawa (Canada) Department of Engineering and Works.

This article first appeared in the April 1997 edition of  BETTER ROADS magazine

During the winter months, the City of Ottawa's Operation Branch employs 220 wage field staff, 30 to 50 staff from a casual employment pool, and operates and supervises over 300 pieces of equipment. During typical storm clean-up operations, the branch  hires up to an additional 75 hired graders over and above the current 10 city-leased graders, 120 snow haulage trucks, and 10 to 15 sidewalk plows in addition to the current city-owned/leased 40 sidewalk plows.  All this labour and equipment clears snow from 456 mi. of city streets, 136 mi. of regional roads, 25 mi. of NCC streets, and 600 mi. of sidewalks and pathways.

The city receives, on average, 86.5 in. of snow per year.   The city performs an average of 10 full roadway plowing events on more than 60 predefined roadway plowing beats, 100 full salting events over 22 predefined salting beats city-wide, 40 sidewalk plowing events on more than 70 predefined sidewalk plow beats, and four to eight full snow removal operations on over 340 mi. of streets.

The winter maintenance budget for the corporation is approximately $10 million.  The corporation is also responsible for the winter maintenance of the regional road network inside the city limits, on which an additional $6 million is allocated.  If a small percentage reduction in mileage could be accomplished, or if operation efficiencies could be introduced, immediate savings could be realized when applied to the hired equipment fleet (that is, equipment hired by the hour).

The current roadway plowing route, or beat system, consists of streets geographically sectioned off within the city.  For example, a beat could be all residential streets between Woodroffe, Maitland, Carling, and the Queensway.  The streets within a plow beat are listed roughly in order or priority, as well as in order in which an inexperienced or unfamiliar operator could plow them.   As an operator becomes more familiar with a beat, he or she will modify the plowing sequence to maximize right-hand turns and minimize left-hand turns.  Sidewalk plowing beats have been developed over time in a similar manner.

Salt beats fall victim to different operating parameters (that is, salt application with the underbody in the upward position, salt application with the underbody in the downward position, hopper capacity, salt dome location, and so on).

Route optimization

In response to information on arc-routing software (as opposed to point-to-point routing) collected through a number of sources, our staff initiated a search into the various off-the-shelf software products that could perform this type of routing for winter maintenance activities. 

After identifying five different vendors that claimed to provide arc-routing optimization, and after measuring the functionality of each software package against a set of predetermined requirements laid out by operational staff, the GeoRoute software package emerged as the only package capable of performing route optimization for winter maintenance activities with a minimum of custom modification to the software or to the graphical map data currently available.

The purchase price for the software and the graphical map is approximately $60,000.  Also, before analyzing and implementing optimized beats, in-house staff would be required to populate the GeoRoute database and be trained to operate GeoRoute software.

   At this point in its evolution, this approach to route optimization was deemed to be too expensive an undertaking for an as yet unproven savings, especially with the expense coming from operational accounts.

As a result, a local consulting firm, the PSR Group, experienced in the application of GeoRoute for winter maintenance activities and licensed with  GeoRoute software, was contracted to conduct a pilot project.

The study's objective was to investigate the costs of designing computer-optimized roadway plow beats.  An associated field trial was included to verify any operational savings or identify any operational concerns which the implementation of optimized beats might introduce.

Study components

The winter maintenance activity chosen as an area for potential savings and operational improvements was roadway plowing.  The rationale behind selecting this winter maintenance activity was based upon the fact that a significant portion of the costs are incurred by hired equipment and not by full-time city staff.  For each roadway plow run, the city hires 75 graders and their associated operators by the hour to complement the 10 leased graders operated by city staff.  On average, a typical plow run takes approximately 12 to 18 hrs., and there are, on average, 10 plow storms per winter season.  Using these figures, the hired grader costs per winter would be $675,000 to $1,000,000.  A 10% reduction in the time it takes to plow a beat could represent $100,000 in current year savings.

The GeoRoute software requires three types of data: graphical street network data, attribute data for each street section, and vehicle performance data.

The graphical street network data used for the project was the computerized network available through StatsCanada.  Included with this data set was street address ranges by side of the street by block, as well as street name, from street, and to street; all indexed by a unique section number.

The attribute data was supplied by the city.  It included street width, street length (although, the network editor can also calculate the object length of each street segment), number of lanes, whether the street is one-way or two-way, and so on.

The vehicle data was supplied by the city, and detailed the performance capabilities of the roadway plowing fleet, which, in this case, were all graders.  Although the city does not have different vehicle types performing the same winter maintenance activity, the software can accommodate different vehicle types, vehicle speeds, and so on.  Typical data required by the software would include travelling speed (while plowing), deadhead speed (blade up), as well as some operational assumptions, such as U-turn duration (that is, extra time included to tuck snow into snow banks around a corner), and duration to cross a street.

Study results

The consultants were asked to simulate three separate operations:

1. Simulate the current roadway plow beat in Woodward, and maintain the plowing sequence documented on the beat sheets.

2. Simulate Woodward's current roadway plow beats such that the physical streets remain the same for any particular beat; however, only re-sequence the order in which an operator would drive the beat.

3. Simulate the beats as the GeoRoute software would re-sequence and optimize the beats.

In a city routes re-sequenced scenario, GeoRoute illustrated that productivity gain can be achieved by simply re-sequencing the streets in an existing plow beat.  Although 18 graders are still required to plow the re-sequenced tandem beats, based on the GeoRoute software, the longest plow beat now takes only 5.5 hrs. (versus 6.5 hrs.) to complete, and the shortest takes 2.0 hrs. (versus 2.1 hrs.) to complete, with the average duration of all 9 beats being 3.5 hrs. (versus 4.2 hrs.).  This approach would provide a more consistent level of service to the city's arterial and collector streets.

Similarly, although 24 graders are still required to plow the re-sequenced single beats, based on the GeoRoute software, the longest plow beat now only takes 8.0 hrs. (versus 11.9 hrs.) to complete, and the shortest takes 2.5 hrs. (versus 3.7 hrs.) to complete, with the average duration of all 24 beats being 5.0 hrs. (versus 6.9 hrs.).  Again, it appears that re-sequencing the order in which an operator should drive a beat would provide a more consistent service to the residential streets.

Optimized routes

When routes were optimized, plowing time was cut even more.  A first attempt at optimization fixed the time for completion of arterials at 3 hrs. and collectors at 4 hrs., as per the quality standards.  This iteration did reduce the time required to plow the beats, but it did not reduce the number of beats required.  When the duration to plow all arterials and collectors was set at 4 hrs., a reduction in the number of actual beats was experienced.  Therefore, as the difference between 3 and 4 hrs. was an improvement, relative to the variation in service that some of the streets were already experiencing, it was decided to fix the time required for completion of both the arterials and collectors at 4 hrs.

Once this adjustment was made, optimization reduced the number of tandem beats from nine to eight.  Based on the GeoRoute software, the longest tandem plow beat now only takes 4.5 hrs. (versus 6.5 or 5.5 hrs.) to complete, and the shortest takes 3.0 hrs. (versus 2.1 or 2.0 hrs.) to complete, with the average duration of all 9 beats being 4.1 hrs. (versus 4.2 or 3.5 hrs.).  This approach provides a more consistent service to arterial and collector streets than re-sequencing the existing beats.

Similarly, optimization reduced the number of single residential Class III beats from 24 to 19, and also reduced the hours to complete.

When translating the number of single and tandem plow beats and their respective expected duration's into total grader hours per storm event, the optimized beats require 261 grader hrs., while the original beat system required 316 grader hrs.  The savings on hired graders per plow storm is estimated at $4,125 for the Woodward Maintenance District only.  The estimated savings per plow storm city-wide is calculated to be approximately $15,000, or $150,000 per year for 10 plow storms.

Trials possible

Optimizing residential roadway plow beats improved productivity in terms of miles-plowed-per-hour by 60% to 90%.  Nevertheless, extra pre-season effort must be made to inform operators (city staff and hired operators, alike) and familiarize them with the routes or beats for which they will hopefully be responsible throughout the winter.

Based on the trend towards increasing the use of an on-off hired equipment fleet for winter maintenance activities, in lieu of replacing retiring year-round, full-time staff and carrying city owned/leased equipment, route optimization is recommended as a consistent, defensible method for preparing efficient roadway plow beats.

In fact, due to erosion in the number of field staff available to supervise this hired equipment (the hired equipment fleet does not come equipped with two-way radios), route optimization is recommended as a basis for ensuring that winter maintenance work to be tendered out to the private industry will result in a consistent delivery of services.  It is recommended that, in tendering this type of work, consideration be given to tendering it by the metre or mile, as even the efficient routes will take hired equipment longer to finish than needed, if they are unsupervised.