Click for next page ( 259


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting 5.3 Advanced Runcutting In the previous sections we have developed runcut solutions for a scenario con- LEVEL taining a relatively basic schedule, simple work rules, and minimal constraints. 3 The Line example is of course simple in that it involves one route, one garage, and one relief location (using one type of relief). In those examples, the work rules used were limited to a basic set of hard rules, with no implied limits based on preferences or unwritten practices. The reality of scheduling is that many situations presented to the scheduler are more complex than this, with multiples in many of these categories--mul- tiple routes, garages, relief locations, and types, for example. However the Line schedule presents many constraints that often do not exist in many run- cuts. The scale or size of the runcut is a significant factor in the difficulty of creating a solution. Cutting just a single run often results in a less optimal solution than cutting an entire system. However, contrary to many expectations, this is often an inversely proportional relationship, i.e., the larger the runcut the easier it is to find a solution! Looking back at any of the block diagrams for the Line runcuts we see limited options to cut pieces. For example there were only two shorter peak blocks in each peak, and those blocks did not allow us to form runs within the -hour spread limitation. In another situation, with or vehicles we might have had a range of shorter blocks to choose appropriate starts and finishes to match pieces into legal split runs. For this reason there is less need to work through further examples of larger problems. Just simply apply the same approaches, philosophies, and tools to dealing with larger runcuts. In this section we will look less at the mechanics of the runcuts (although there will be numerous examples), and consider more of the discussion around runcut approaches, and how more dif- ficult issues can be resolved. The later parts of this section revolve around a discussion of some key policy issues relating to runcuts. Larger or more complex problems can add the following elements to the examples we have worked through to this point: Volume of trips, blocks, and routes If runs are cut at a route level (i.e., each run only operates on one route) the approach to cut- ting a route with more service or longer service span is exactly the same as in the previous examples. There will of course be increased options, more runs, additional pieces, etc., but the approach remains the same. 5-54

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Additional Garages Creating runs for multiple-garage solutions adds complexity to the scheduler's task, both at a blocking and runcutting level. Where there are multiple garages, another level of consideration is required: At which garage should the piece start/endt? Can the block be adjusted to change garages? What are the travel times between the relief locations and each garage? Can meals occur at a different garage than the "home" garage for a run? Are compatible bus types assigned to each garage? While these issues do add complexity to the runcut process, it tends to be more an issue of in- creased options rather than placing constraints on the scheduler. However in considering those options the impact of optimizing start/end locations is critical to minimizing overall deadhead- ing or travelling costs. Two key considerations are: Vehicles to be balanced at the start and end of the day, i.e., the same number of ve- hicles must pull out of a garage that pull in at the end of the day Generally operators must start and end their work day at the same garage These two factors must be taken into account, along with the considerations noted above, when undertaking multi-garage runcutting. Additional Relief Types and Locations Our Line example had one relief location and one method of traveling to that location. In many cases routes will have multiple relief locations and possibly other methods of travel. Additional relief opportunities actually only help to make the problem more readily solvable-- through presenting the scheduler with more options for cutting pieces. The extension of concepts for relief types is discussed in more depth later in this section. More Complex Work Rules Work rules present the most difficult and constraining factor in runcutting and also represent the most variability from one operator to another. Earlier sections of this manual have already discussed concepts of hard rules, soft rules, constraints, and practices. At times these can con- tradict each other, or be difficult to understand or explain. 5-55

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Take the work rules relating to meal breaks for one transit system: At least one of the following constraints must be met in straight runs over hours: x minute break and x minute break; or x minute breaks; or minutes total layover ( % of runs maximum) The agreement states that not only must one of these be scheduled in each run, but also that % must be achieved in actual operation. In actuality, virtually all scheduled break require- ments are expected to be achieved in operations. Runs are further limited by the requirement that no more than % of runs can involve a ve- hicle change. This limits options to create multipiece runs to deal with the breaks. Just reading and understanding these interrelated work rules is a complex task. And these con- straints apply to only a small part of the overall work rules at the agency (i.e., the meal breaks). The limitations applied to scheduling processes by more constrained work rules or practices present the biggest challenge to the scheduler. This manual cannot even begin to attempt to explain myriad work rules and practices adopted or applied by transit systems. We have noted some of the "typical" kinds of rules applied and discussed how to create runs with these rules in mind. We would simply recommend that strong consideration is given to work rules when schedul- ing, and that the rules be appropriately applied. By "appropriate" here, we mean as intended by the agency to fit within the overall agency's objectives and policies. Using a computerized system actually provides some good checks and balances in this respect, as the scheduler must actually write down the rules, constraints, limits, and preferences. These must be succinctly defined and expressly stated for the system to be able to create legal runs. When using other scheduling techniques, this approach should be mimicked--write down the specific rules, not in the words of the labor agreement, but as input into runcutting. Summarize these rules or practices and circulate them to gain wider approval. On the other hand, the computer is not as good at making nuanced judgments as a human scheduler can be. Only an expert scheduler knows which rules can be "bent" and how often-- the computer will "do as you say" but will not make the complex trade-offs and judgments that an experienced scheduler can make. This is why, even at properties that utilize up-to-date computerized scheduling packages, an experienced scheduler should always review the end result and see if further optimization could take place with a few reasonable adjustments. 5-56

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Runcutting Multiple Routes The number of variables increases dramatically when cutting multiple routes. The classic ex- ample of multiple-route scheduling involves hooking together peak blocks to form split runs. In our original Line example, we had four standalone peak blocks that had a spread too long to fit within spread limits, or at least to avoid incurring high spread premiums. If we add a second route into the runcutting mix we can see some potential for immediate ben- efits, as per the diagram below: Line 97 Blocks 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 Run 1a 2 3 4 Run 2a 5 6 Run 2b 7 Run 1b Line 98 Blocks 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 2 Run 3a 3 Run 4a 4 Run 4b 5 Run 3b On Line we have a much shorter peak, with : and : starts, and two : finishes. If Line was blocked standalone we would have two split runs (the white Runs and ) with : and : report times, and : sign off times. At the same time we have split runs on Line that are either illegal (spread too long) or have long spreads with high premium paid. If we matched the late starts on Line with the late finishes on Line , and the early starts on Line with the early finishes on Line , we get four legal split runs with a reduction in total spread premium. The following diagram represents the newly created multiroute split runs. 5-57

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Line 97 Blocks 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 Run 1a 2 3 4 Run 2a 5 6 Run 4b 7 Run 3b Line 98 Blocks 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 2 Run 3a 3 Run 4a 4 Run 1b 5 Run 2b This is a simple example where the pieces matched easily. In practice such outcomes are not always the case. However the concept and approach remains the same--that the introduction of multiple routes allows the potential for increased options and better runcuts. As with larger problems, the issue is one of size, and the more options present more potential solutions. Trips, Blocks, and Runs In our earlier examples, we learned that it is sometimes possible to make small changes to blocks to improve the runcut solution. But in those examples, the changes made were only minor, affecting only two or three blocks. Let's now go back to a statement made in a preceding section--that the runs are, in effect, a group of trips that form an operator's work day. Often schedulers limit the ability to produce effective runcuts by assuming the blocks are fixed, or by limiting changes Note here that to some extent we are playing with fire. The primary reason to begin runcutting with a set of blocks is that blocking is designed to minimize the number of vehicles required to operate a given schedule. By essentially throwing out the blocks and starting with a list of trips, we are optimizing for driver hours, but not necessarily for peak vehicles. Peak vehicle availability is often fixed at a transit operator and may be an even more rigid limitation than the number of driver hours. 5-58

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Under no circumstance Runcutting Multipiece Runs Tip should the runcut result in Many transit systems avoid multipiece runs and many labor agreements forbid or constrain increased peak vehicle requirements them. The reason appears to be primarily one of simplicity and control. A set of runs comprised over the blocking solution. of long blocks cut into straights, and peak blocks matched to splits (or assigned to part-time runs) is a simple and coherent runcut option. However as we have discussed and seen in our examples, the limitations of relief opportunities make it difficult to cut straight runs, and/or match pieces into split runs, in an optimized man- ner. We can go back and reconsider our definition of "optimized" here. Our Line example was possibly "optimal" with the constraints placed on it. But was it the "optimal" solution in terms of minimized paid hours for the given set of trips? The answer is almost certainly not. multipiece runs A primary aim of creating more and smaller pieces is to be able to combine and match them Runs made up of pieces from into runs that better improve the runcut solution. This can potentially result in reduced overall multiple blocks. In most cases, split cost and a lower pay-to-platform ratio. runs are inherently multipiece runs. The pieces used in multipiece runs can be cut at the same relief location, or perhaps at different But either half of a split run could locations with a travel in between. The aim is to create pieces of work that can be best matched itself could have multiple pieces, cut together across the whole runcut. In the previous section we did this to improve our split runs from multiple blocks. and then rehooked the blocks to avoid the "multipiece run" name. But in effect that is what happened--we created runs with multiple pieces to improve the runcut solution. We are treading here on some prevailing scheduling theory in that multipiece runs are often frowned upon as "bad" or "wasteful." This can be true for one or two runs, particularly if there is a paid break or long travel between the pieces. However in the context of the total runcut the creation of some smaller pieces can provide significantly enhanced overall efficiencies. So how do multipiece runs work? The concept is simple--cut smaller pieces and match them together to form runs, in much the same way that we matched pieces together for split runs in earlier examples. But in this case we are trying to match pieces with smaller breaks (so as not to incur paid break time), while meeting other runcut objectives. The advent of computerized scheduling makes it possible to test different types of runcutting solutions. These systems allow run types such as multipiece runs (of any sort--straight, split, part-time) to be tested and reviewed, within an overall optimized runcut. One reason multi- piece runs are not always embraced is the sheer complexity they cause for schedulers. If mul- tiple piece runs are allowed, the potential number of pieces and cuts increases dramatically. 5-59

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Pull Reliefs and Multipiece Runs We can go back to our example in the intermediate section ("Rehook the Trips") to see multi- piece runs with pull reliefs which were then rehooked to provide updated blocks. A key feature of this example was that the blocks were rehooked outside the peak periods. This is a very important point. Allowing runcutting processes to rehook trips during the peaks (when peak vehicle requirements are defined) can result in increased vehicle requirements. Outside of the peak, the use of pull reliefs can increase vehicle needs with little consequence (an exception being if you have a very low peak-to-base ratio, in which case off-peak vehicle increases can affect other operational requirements such as availability of vehicles for maintenance). When using computerized systems, this requirement must be carefully monitored, as the runcut modules at times fail to consider the cost of peak vehicles in their cost optimization process. The potential costing impacts of pull reliefs need to be carefully considered by the scheduler before any widespread application. Greater use of pull reliefs is almost guaranteed to increase vehicle mileage, often at the expense of cheaper car mileage. As noted, the evaluation of use of different relief types can be undertaken quickly and effectively through use of computerized scheduling systems. If multipiece runs have been created, using pull reliefs, it usually is a simple matter of recreat- ing the blocks after the runs have been created. Again, be careful not to overlook some consid- erations taken into account by the scheduler when blocking such as: Congestion/capacity at terminals; Vehicle type limitations; Reliability on certain routes; Interlining limitations; and Maintaining adequate recovery time. Runcutting for Meals So far, our examples have included only three types of run--straight runs, split runs, and part- time runs. Our straight runs have been one-piece runs, and the splits have had a break back at the garage. Many agencies now have labor rules mandating some form of operator breaks. In many parts of the United States new rules mandating meal and rest break requirements are being devel- oped and implemented. Rules relating to meal breaks can differ significantly in complexity and in how they affect efficiencies. These may change from one transit agency to another. 5-60

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Many agencies struggle The types of breaks required may include: Tip with providing meal Minimum layover requirements. Sometimes expressed as a percentage of layover breaks, due to the increased time during a work day, or at least one layover of x minutes. complexity and potential constraints. Stipulated meal breaks. Meal breaks come in many forms with varying requirements. However, meal breaks, if the right The breaks may be paid or unpaid, taken back at the garage or at relief locations (or labor rules exist, need not add significantly to the cost of operating in-vehicle), may have minimum/maximum lengths (above the maximum length the run service. This holds true in particular typically becomes a split) , and may be required to be spaced at certain times during the where meal breaks are unpaid and work day. can be taken at the terminal rather Other rest breaks. These may be required in addition to scheduled meal breaks (such than returning to the garage. as one -minute layover every two hours). Many agencies view meal break requirements as immediately resulting in cost increases. However, as we will see as we explore some of the detail, this is not necessarily the case. In many parts of the world transit systems operate with meal break requirements while creating extremely efficient run cuts. Depending upon the combination of rule requirements, the scheduling of meal breaks can make the runcutting task significantly more complex. If a piece of work can only be five hours long, for example, it means that all straight runs must have at least two pieces. This doubles the number of reliefs that must be considered, exponentially increasing the mathematical complexity of the problem to be solved. In the case of minimum layover requirements, it may be possible to build the blocks with these already built in. In our Line example we built the schedule with minutes of layover at Point A. When we hooked our trips into blocks, we considered that breaks might be required. If we go back and review our runs in detail we will see that the straight runs (those that are most likely to require meal or rest breaks) have a -minute break every minutes. In this case our thinking ahead allowed the requirement to be met. But what happens when there is a more complex or restrictive rule--a typical meal break requirement such as "the operator must have a meal break of between and minutes, be- tween the second and fifth hour"? The break may be paid or unpaid; in this case we will assume it is unpaid. If we assume everything else remains unchanged we can go back and rebuild our runs. At this stage we can consider two potential approaches to runcutting to meet the meal break requirements. The first involves cutting longer pieces, across multiple vehicles, and assigning operators two pieces of work in a run. The second involves operator drop-backs at terminals. 5-61

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Approach 1: Recutting the Runs This approach to cutting the runs changes the mechanics of the process significantly, since we now must have run "portions" (the work either side of a meal break) no longer than five hours. And those pieces must be less than five hours, given that we will possibly need to add travel time and report time to the actual platform time of the piece. In this case we'll go back and use our original Line blocks and assume all street reliefs at Point A once again. We will assume that meal break must be taken back at the garage, requir- ing travels to and from the depot. We will also assume the meal break will also be unpaid. As we will discuss later, these are two key factors in the cost impacts of meal breaks. In this exam- ple we have taken a less restrictive view of one factor (paid meals) and a more restrictive view as to where meal breaks can be taken (in this case at the garage and not at trip terminals). Where do we start? As always the scheduler should make some macro observations about the number of runs (by type), the nature of the blocks, and potential pieces. We know from the previous examples that we had eight full-time runs. However with hours of platform time we could be looking to create as few as seven runs. And the runcuts produced had almost three hours of guarantee due to the runs being generally shorter than we would have preferred. We also need to think about how many pieces we will cut--we will assume all runs are now two pieces for a straight or split, and one piece for a part-time run. If we are looking for eight-hour runs, and a piece cannot be more than five hours, that means we'd ideally like all pieces to be between three and five hours. Before starting to create pieces or runs, we should review the blocks and consider how often we will need to cut. This is shown simply below. Block Length Pieces 1 3:33 1 2 12:37 3 3 12:29 3 4 3:33 1 5 13:33 3 6 4:48 1 7 3:59 1 Total 54:31 13 Average 4:11 5-62

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting As a starting point, look The table gives us a very rough idea of the number of pieces we are aiming for and the average Tip to the blocks with the length of each piece. The actual number of work pieces and run length will be determined in earliest starts as opportunities for large part by the opportunities we have for reliefs. The number of pieces will also be affected straight runs, and blocks with later by the need to combine pieces into straight runs. The odd number of pieces suggests we may starts and early finishes as possibili- end up with either a part-time run or a longer straight run with more than two pieces (and pos- ties for split runs. sibly two meals). But for now it is enough to know how many pieces and runs we are aiming for. So the -piece indication is only an approximation at this point and could change. A general rule of thumb is to begin with the early start times as candidates for straight runs, leaving the splits to be the later starts and earlier finishes. This minimizes spread premium costs and is a generally accepted preference for operators. If we review our blocks again this means that we are looking for the starts of blocks , , and to be the first piece of our straight runs. In the PM we want our early finishes to be the splits and later finishes to be PM Straights. In effect we are ensuring our shorter splits, with starts at : (Block ) and : (Block ), and finishes at : (Block ) and : (Block ). This is the same matching process we undertook when rehooking blocks to arrive at our preferred splits, with minimized spread premium, in the earlier example. In effect we have already created the split pieces. We just need to know when to finish the two AM pieces and start the two PM pieces. The finishing time of Block is mandated by the fact that it is a smaller peak-only block. So we assume Block is the first half of one of our split runs. To minimize our spreads we want to match this with the piece on Block ending at : . Assuming we are looking for around an -hour run, this dictates a cut at before : . A process of cutting and matching is then undertaken to create and match pieces to form legal and efficient runs. Again we turn to a graphical representation of the runs to provide an over- 5-63

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting view of how the blocks have been cut into pieces. Pieces with the same run number (e.g., run is pieces of Blocks and ) form a Straight Run (Run --see Run guide below). 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 Run 1 2 Run 3a Run 2b Run 3b Run 5b 3 Run 2a Run 6a Run 4b 4 Run 4a 5 Run 5a Run 7a Run 8 6 Run 6b 7 Run 7b As the diagram depicts, the additional pieces required by creating runs with meal breaks adds significant complexity. The process requires consideration of numerous combinations of poten- tial pieces. However some basic rules can be applied: Look at the start and end pieces initially to provide an idea of the pieces and runs. Work- ing forward from the block starts (to a maximum of hours) and backwards from block ends will provide an indication of how many of each type of run will be required. Avoid cutting for meals during the peaks where possible, since this in effect means two operators are covering the one peak. Cut the AM straight pieces as the early starts. PM straight pieces should generally be as long as possible to ensure operators can cover both the PM peak and late night services. This of course depends on how late the ser- vice runs. In this case the night service is minimal. Split pieces should be cut with spread limitations in mind. The number of straight runs will be a function of the level of early, midday, and night service levels. With only three midday blocks in our example, and no night service, the opportunities to cut straight runs are reduced. Now let's move on to the solution. The run guide is again provided, this time with some addi- tional information to explain the straight runs. There are eight total runs. Of these, we have four straights, two splits, and two part-time. Note the total paid time has increased from : in our previous solution to : . However this is probably a good outcome given the additional travel time required to have all straight runs return to the depot for meals (in effect adding an additional : of time into the runs). 5-64

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting If we simply take those four pieces of work assigned to split runs and convert them to part-time runs the cost impacts can be easily reviewed, and are provided below. Split Part Time Total Runs 2 4 Platform Hours 15:01 15:01 Travel Hours 0:30 0:30 Spread Premium 2:46 Overtime 0:35 Guarantee 0:00 Total Paid Hours 18:52 15:31 5-84

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting In this example the savings of using four part-time runs to replace the two split runs is over three hours and equates to almost %. If the hourly rate for the part-time operator is lower than for full time the savings become even more significant. It may also be possible in other cases for part-time operators to work two small peak pieces (assuming total work time is within labor rule limits), thereby replacing some split runs on a one-to-one basis. Appreciable gains in efficiency can be achieved by this strategy. This is a basic example, and the conditions for splits and spreads made the split runs relatively inefficient to begin with. However, it illustrates the potential savings that creation of part-time runs can provide. Part-time runs may also be useful where there is an imbalance between the peaks and the peak blocks cannot all be assigned to split runs. Chapter : Rostering notes an- other potential benefit of the use of part-time operators: to fill open run imbalances on certain days of the week. Use of computerized scheduling packages can allow more sophisticated modeling of the impacts of varying levels of part-time runs, where total paid hours can be compared and multi- plied by actual wage rates. More Complex Runcutting The example presented described a basic situation with relatively simple rules and few con- straints or limitations. There are many more situations where part-time runs can be effectively used to produce an efficient runcut. The complexity of work rules around all run types can result in pieces of work that are difficult to assign to full-time runs. For example, in the meal breaks section we created a runcut that required two smaller pieces of work, which were then assigned as part-time runs. A set of more highly constrained work rules is more likely to result in a need for shorter run types ideally suited to part-time operators. As the complexity of the rules and number of con- straints increase, it becomes more difficult to cover all of the blocks with full-time runs. Typically there will be a limitation on the number of part-time runs that can be created within a runcut or assigned in a roster. In many cases the "optimal" number of part-time runs, from a runcut perspective, will be close to or above this limitation. 5-85

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Operational and External Impacts The above discussion indicates the well-known potential efficiency benefits in application of part-time runs and operators. However there are a number of other issues relating to part-time operator deployment that can affect the overall benefits. The relative competence, training requirements, and retention rates of part-time operators all affect the overall benefits of their use. These factors can translate into measurable cost impacts such as higher accident rates or increased unscheduled overtime (through either increased absenteeism or unfilled operator positions). The specific impacts of such factors tend to be local in nature, and our research indicated different viewpoints as to the benefits or costs relating to part-time operators across different transit agencies. Ideally historical data at an agency can be obtained, and any related factors can have costs or values applied to allow a reasonable estimation of the total impacts (i.e., measure the runcut or roster savings against any operational or external cost impacts). C. 10-Hour Runs Ten-hour runs and rostering of four -hour days for a driver work week offers another tool to schedulers. As with the discussion of balancing overtime and part-time operators, introducing -hour runs can be considered as purely a runcut issue, or as a runcut and rostering issue. Returning to some of the examples presented above, creating longer runs at the runcut level is likely to result in reduced travel, report, and guarantee. Taking another simple example we will look at hours of platform time and assume the runs can be cut as either all eight-hour or all -hour. We assume that there are minutes of report and travel time for each run, and that we are looking for an average of : and : , respectively. 8-Hour 10-Hour Total Runs 13.3 10.5 Platform Hours 100.0 100.0 Report Hours 6.7 5.3 Travel Hours 2.2 1.8 Total Paid Hours 122.2 117.5 -4.7 -3.8% 5-86

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting The idea here is to demonstrate that, as fewer runs are created, the total cost of travel time and report time is likely to decrease because these cost factors are related to the number of runs. Rules Affecting 10-Hour Runs There are of course many different implementations and rules governing the use of -hour runs, differing by agency. However some general types of rules exist and may specify: The minimum and/or maximum number of -hour runs allowed, usually expressed as a percentage of total runs Overtime and guarantee will tend to apply at the -hour mark and not the usual - hour mark Spread penalties will apply, but probably only to split runs Other rules will be applied to -hour runs as they are to eight-hour runs Workforce Factors While the example above would suggest that -hour runs are a tool for saving money, the re- ality is not that clear. The assumption with -hour runs is that they are tied to four-day work- weeks. If a -hour solution requires more operators to be hired, it may actually be a higher cost solution since every operator comes at a fixed cost for benefits. At this stage we need to take a step back and consider the workforce impacts of our two options. We take our runcut numbers from above and use them to estimate the number of operators that are required. 8-Hour 10-Hour Total Weekly Runs 66.7 52.6 Days Worked Per Week 5 4 Operators Required 13.3 13.2 And the answer is the same! Taking a step back like this makes sense, as with either - hour or - hour workweeks the total number of worked hours is . The table does show a mathe- matical reduction of . operators required (less than %), but this does not necessarily trans- late into any real workforce reductions. The implication here is that - workweeks do not result in reduced labor requirements. The basic tenet is that the total number of operators required is a function of how much work is undertaken in a week, not in a given runcut. The reduced numbers in a runcut therefore need to be considered against the total days worked. 5-87

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Runcut Impacts Further consideration is required into how -hour (or any longer) runs affect the runcut. As with many runcut issues, this is a function of many inputs. Let's return to our original Line blocks and consider the potential to create -hour runs. 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Block 1 2 3 4 5 6 7 Here there are three -hour straight runs highlighted. But notice the effect on the rest of the runcut. We are left with seven shorter pieces to make into split or part-time runs. In this case, spread limits constrain the number of split runs to at most two, leaving us with three part-time runs. The impact of the -hour runs in this case is to increase the number of shorter part-time runs to three. This may or may not be a desirable outcome. Our total number of runs remains at eight (three straight, two split, and three part-time), yet when it comes to rostering/bidding there will actually be an additional operator required (as the -hour runs will be covered by operators working only four days). In this case the answer is probably that -hour runs, combined with four-day workweeks, would result in an efficiency loss. However there are many cases where -hour runs and -day workweeks may be appropriate. Again the key driver of efficiencies through this approach is the reduction in nonproductive time in the runcut (primarily travel and sign-on/clear time) through fewer total runs. Issues with 10-Hour Runs The discussion above has covered some of the impacts of -hour runs, whether combined with -day workweeks or not. Below are some further issues to consider when creating -hour runs. A longer span of service, with a higher level of off-peak service, tends to lessen the runcut impact of -hour runs. For example if all blocks are only -hours long, cutting -hour pieces limits the number of straight runs that can be created (as in the above example). However longer blocks reduce this impact. Eighteen-hour blocks allow one -hour run and one eight-hour straight run to be created, for example. 5-88

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting A larger runcut generally provides more options for "the rest of the runcut" if - hour runs are applied. Simply put, higher volume in the runcut allows more options for the remaining pieces. Typically there will be limitations on the number of -hour runs that can be cre- ated. As a rule of thumb it would be rare for more than % of runs to be of the -hour type. -hour runs should incorporate the same philosophy as eight-hour runs, where possible. That is, there should be a balance between runs over hours and under hours. However in some circumstances driving time limits (in some cases set at hours) limit this capability. In this case guarantee time should be avoided and runs cut as close to hours as possible. Split runs are difficult to increase in length at their extremities (i.e., before the AM Peak and after the PM peak) due to spread constraints, meaning that -hour splits will tend to creep into the off-peak periods between AM and AM and PM and PM. This becomes less feasible if the off-peak service levels are low and can also reduce the number of straight runs. However, it is generally difficult to build longer split runs, and use of -hour splits often leads to an increase in make-up time (assuming a guarantee of hours). Work rules at many systems allow for the creation of longer (up to -hour) runs as part of a normal five-day workweek. Whether such runs are efficient depends upon the overtime approach adopted (see the earlier discussion on optimized overtime). The decision to apply -hour runs to four-day workweeks is as much a rostering and work- force utilization issue as it is a runcut issue. Longer runs can have operational and absence coverage impacts. For example an operator who works a AM to PM straight run may be available to cover a school trip- per or short PM block as overtime, possibly to cover an unscheduled absence. However if the work day is longer there may be less opportunity for such coverage, or safety is- sues through longer driving times may arise. There are potential safety impacts of operators working longer runs. Your agency should have some data on any link between longer daily driving hours and accidents, for example. D. Workforce Utilization The three preceding sections have each covered aspects of workforce utilization. The runcut can affect workforce utilization through the mix of run types applied, by the length of runs cre- ated, and by creating a runcut that allows for rosters to be built effectively. While each of the 5-89

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting three discussions was presented in isolation, the tradeoffs between overtime, part-time opera- tors, and -hour runs are made interactively by schedulers when creating runcuts. Workforce utilization from a scheduler's perspective relates to how the runs and rosters are built to apply the available labor resources most efficiently, within the constraints provided by contracts and work rules. In this sense it relates not only to the scheduled cost of the runcut but also how this translates to extraboard utilization, absenteeism management, and unscheduled overtime. In the preceding discussions we have noted a strong link between the makeup of the runcut/ rosters and the management of labor in terms of levels of absenteeism and unscheduled over- time. There are certainly no prescriptive answers here. Below we note some of the factors that influence how effectively the planned runcut translates into efficient operations: The current labor market Agency-specific wage rates How the roster is constructed, e.g., is weekend work rostered or left as uncovered for overtime? Are trippers created? If so, are they left uncovered for overtime, or assigned to the extraboard? What is the size of the extraboard? E. Cost, Complexity, and Quality A good scheduler understands that at times there are tradeoffs between the overall cost of a runcut and the complexity of the runs created. We have touched on this topic repeatedly throughout the manual, as we covered issues including: Interlining. There can be a tradeoff between the amount of interlining and the cost of the solution. Garage assignments. Keeping garage assignments simple may result in additional deadhead mileage and hours. Run Types. In particular the use of part-time and -hour runs can affect efficiency lev- els and operational complexity. Reliefs. We have covered in detail the potential to use different relief types, at differing levels of complexity, to achieve a range of outcomes. 5-90

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Multipiece runs, which allow potential for improved efficiencies but represent a more challenging operational outcome. Meals. Having meal breaks in runs, however handled, makes the runcut and the opera- tion more complex. Meals can be created in a manner that reduces this complexity. Events and exceptions, which can be scheduled in a manner conducive to consistency of operation from one period to the next. In all aspects of scheduling the scheduler is attempting to create operationally sound and efficient runcuts. The important consideration here is that these are not mutually exclusive outcomes. The section below notes the ability of computerized systems to produce improved quality out- comes for the same efficiency levels. What does this mean? It means that the scheduler should not only model cost constraints and work rules but also quality and operational preferences. These can then be considered as part of the runcutting solution as well. Consider a case where we have two split runs with spreads of : and : , and maximum of : allowed in the labor agreement. By swapping the PM pieces we could end up with two -hour spreads. The cost of the spread premium is the same but now we have a potentially better outcome. The computerized system can be told this--that we prefer equalized spreads or fewer spreads close to the maximum limit. We can then run an option with this limit and without this limit, and see if there is any discernable change in efficiency levels. F. Computerized Scheduling Revisited During the course of previous sections we have noted the capacity of computerized scheduling systems to assist in and improve the scheduling process. Our overall observation is that the widespread introduction of computerized systems has resulted in efficiency gains, particularly in larger systems where the size of the runcutting prob- lem means the sheer computational power of the computer provides better solutions. When undertaking runcutting, there are several approaches that need to be applied in order to best harness the computational power of the system and work within the needs of operational outcomes. Some of these are noted below. Understand the solution that the system should be trying to achieve. Consider some of the approaches we have discussed here and get to know what the solution should be, before the computer generates one. 5-91

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting Test the rules by manually creating, within the system, various types of runs. Make sure these are legal and are costed correctly. The system must be able to accurately create and cost runs according to the rules provided, and this calibration step is critical. The rules should be tested and refined with a "one-at-a-time" approach to ensure each aspect of the labor agreement is accurately modeled. The first solution is not the best solution--ever. "Push button" scheduling is never an option for producing quality schedules, runcuts, or rosters. In every scheduling solution there is potential for improvement, whether created manually or by the computer. Cre- ate solutions, tweak the rules or parameters, fix some preferred runs--do any of these and recreate the solution, always looking for improvement. Running or versions of a solution is not uncommon, nor is it unreasonable. Use the interactive tools provided. Some systems provide excellent tools for a sched- uler to create a runcut interactively, without submitting an automatic solution. The abil- ity to cut pieces, form runs, and see results as the runcut develops is extremely powerful in the hands of an experienced scheduler. Reconsider the rules/parameters/constraints for each new runcut. Schedulers can fall into the habit of a "set and forget" approach to rules as defined in a computerized sys- tem. For every new sign-up, there are likely to be changes to the trips, blocks, or even work rules. Consider how these have changed and adjust the system accordingly. Recall objectives that could not be achieved last time and attempt to achieve them this time. Start with simple rules. Do not overconstrain a system in the first instance. Start by putting in only basic "hard" rules--durations, counts, limits. Do not decide things for the system if you are not sure of the outcome. Then start to add in constraints and changes to penalties or weightings, preferably one at a time. Build up a more complex set of rules as you go. Carefully consider every rule, parameter, and constraint. Each rule is effectively a line of code that will have a value assigned during an automated runcut. Each rule is there- fore like writing a line of computer code. Make sure the rule has the desired impact by changing only that rule and recreating the solution. Look for solutions that allow improved quality within the same resource require- ments or total costs. Once you are satisfied with the costing levels (number of runs, paid hours, etc.) look to improve the quality of the runcut. This can be both the opera- tional quality (simplicity is generally better) and the quality of life for operators. These outcomes are not mutually exclusive, and a good scheduler will be aware of how far a system can be used to provide these multiple outcomes. 5-92

OCR for page 258
Level . Advanced Runcutting Chapter 5. Runcutting Note that many of these suggestions hold true for both computerized and noncomputerized scheduling processes. Even the most sophisticated computer system still requires review and hands-on involvement of the scheduler, who can make adjustments throughout the automated process to enhance both the quality and the efficiency of the outcome. 5-93

OCR for page 258
Chapter 5. Runcutting Level . Advanced Runcutting LEVEL End of Advanced Runcutting. 3 Rostering continues on the next page. 5-94