Category Archives: 4-Define your constraints

Breaking the Rules

· · Leave a comment

No matter all the constraints you put in, an event will happen that will force the planner to break the rules. The most common is to ask employees to stay for another shift (double shift) which breaks the minimum rest rules. Areas like transportation do not allow this practice by law which would expose them to severe fines, but manufacturing and healthcare industries often ask their employees to work double shifts.

 

Another common broken rule is the maximum consecutive days of work. Once the schedule is published and the actual planned day is about to be worked, the planner will ask for overtime and employees may end up working more days in a row because of an additional overtime shift on a day that was planned to be a day off.

 

While you can be very detailed about your constraints, you also need to identify which ones can and cannot be broken. Otherwise, the planners will decide themselves which ones to break causing legal issues later down the road.

 

Some industries identify a set of constraints that apply when the schedule is built and published and another set of constraints for the maintenance and the management of the schedule. This way, the planners do their best to create a schedule that covers the workload knowing full well that things will change anyways. Therefore, some flexibility is given to planners so that they can make modifications to the schedule when reacting to changing circumstances.

 

The important thing to remember is that some constraints should be put in place as guidelines and others should be enforced as law. Once all of them are listed and documented, it is up to the business to determine the value of each constraint and decide to apply it or not. The more constraints planners have, the more difficult schedules become.

 

The Agreement

· · Leave a comment

As a business grows, there comes a point where all the constraints need to be clearly stated in an employee agreement. Sometimes, the employees themselves take the lead and form a union to force a clear agreement between the business and the employees. These agreements all have their use and importance as they are meant to clear out all the details on many aspects including scheduling.

 

That’s the theory. In practice, when these agreements are negotiated and discussed, the planners are not invited. These things usually happen between employee representatives and Human Resources which represent higher management. The constraints that end up in an agreement may (and will) be interpreted to the letter by both parties to their advantage. What is seen sometimes as an advantage at the time of negotiation all of the sudden becomes a burden because of an unforeseen side effect

 

For example, during negotiations, employees ask to move the minimum rest between shifts from 8 hours to 11 hours. This was due to the fact that the employer can force employees back to work by reverse seniority if staffing is too low to run the plant. Employer says no problem because it won‘t cost him a thing. All agreed and employees are happy not be forced back to work too quickly when staffing is too low.

 

The side effect that no one saw coming is that because the minimum rest was now higher, employees had difficulty arranging shift trades with their buddies. Shift trading was always blocked because the minimum of 11 hours was not respected. Shift trades got to an all time low, and so was employee satisfaction. This turned into higher absenteeism which in turn caused more overtime and more costs.

 

This is just one example of a constraint that was determined without necessarily testing it through from A to Z. A planner could have simulated the changes of the previous month and just wave a red flag on it explaining that all the shift trades would have been refused last month if this rule was in place. Folks would have probably looked for a different solution to the problem.

 

Although some constraints seem insignificant at a first glance, the domino effect can be devastating if it is not looked at the tactical and practical level. Strategic issues will always run the negotiations, but any strategy will eventually turn into a tactical practice. That practice has to be clear before writing the agreement.

 

Training

· · Leave a comment

Depending on your sector of business, you may have extensive training to keep your employees at an acceptable skill level. Recurring training is sometimes mandatory when public safety and hazardous activities come into play. This adds to the planner’s burden of pre-scheduling training and making sure all employees are kept up to date with the recurring training or license before scheduling them. Training will also remove production capacity from the pool of employees.

 

Although it is seen as a constraint, training is a fixed cost to make sure the business stays productive. If training does not occur, then more production capacity will be lost in the near future when training will become no option because the employees will have lost their licenses to perform the appropriate activities. Planners sometimes cut corners short by canceling training to solve an immediate production coverage problem, but will in the end only cause a greater problem down the road. Scheduling training sometimes involves many groups and resources and therefore are difficult to schedule by themselves. An employee will have to sign up for the next class which may occur only months from now. Therefore, removing an employee from a planned training needs to be a calculated move.

 

In some industries, you also need to train new employees on the job and associate them with a skilled trainer. The planner therefore has to plan a new employee and a trainer to work at the same times and also count the employee’s effectiveness into account. Although there are 2 employees, they only produce as one employee. This association needs to be updated and clearly stated all the way to the point of working to schedule. The employee needs to know who the trainer will be and vice-versa.

 

Some industries will even schedule on the job testing where many instructors will observe an employee before granting them a license. These businesses have the additional challenge to keep their employees up to date using a trainer on a one-on-one basis, but also keep the trainer’s skill at that same level plus the skill of training in itself.

 

Although essential, training does remove a lot of capacity from a productivity perspective. All this training adds constraints to the planner’s ability to schedule efficiently.

 

Self Scheduling

· · Leave a comment

When self-scheduling is involved, a planner needs to set guidelines as to what self-scheduling means and what is expected of the employees. Self-scheduling can take many forms, but the goal of it is to enhance employee satisfaction by giving them control on their own lives.

 

The first step to do is to clearly count and display the workload that needs to be covered. The workload is the prime aspect and the breakdown into shifts is the thing where everything starts. Coverage measures must therefore be the first constraint employees must follow.

 

The usual next constraints pertain to night and weekend shifts (or any shift that employees typically don’t want). As a planner, you will determine the constraints that apply to each employee in regards to how many good shifts versus bad shifts each get and put together either ratios or direct counts based on the employees full time equivalent value.

 

What About Seniority?

· · Leave a comment

In some unionized environments, employee seniority is used as a sorting list for offering the best schedule or the any shift that has additional pay. Seniority can be calculated in many different ways. The usual way is to use the employee hire date. In some European countries, seniority is determined based on your age and not on the day when you were hired.

 

Any business that sorts on seniority is using the criteria as an unbiased accepted form of tie breaking and distribution of employee satisfaction. For example, manufacturing industries with day, evening and night shifts will offer the shifts based on seniority. The most senior employee will pick an available shift or express a top preference that will be followed by the planner. All employees may want the day shift, but only the most seniors will actually get the day shifts.

 

Seniority is used in many areas, such as shifts, activities to do (some tasks are harder than others and the most senior employees avoid them), overtime offers, weekend premium shifts, etc.

 

Reverse seniority is also used to force employees to come in to work. For example, if all employees refuse the overtime but employees are still missing, then employees are forced back to work using a reverse seniority order (the most junior employee being the first called back to work).

Fairness and Rotations

· · Leave a comment

In order to increase employee satisfaction around their schedule, some industries have created constraints to ensure fairness and rotations of difficult shifts (typically night and weekend shifts).

 

For example, a constraint to distribute night shifts fairly to all employees would need to be defined on both the how and the span (see horizon constraints). Lets say you count shifts that are from 11pm to 7am over a 3-week period. Is that enough to be fair? Employees who will be on vacation will work less night shifts during that time span and therefore will work less night shifts.

 

The fairness aspect of any counter relies more on its perception than on the real value. I have seen planners count on an endless horizon and employees had the number of hours worked in their entire employment time with some employees with well over 15,000 hours in that counter. A new employee comes in at 0…

 

Fairness is not only about what employees don’t want, but also about what employees want. Instead of counting night shifts, you could count day shifts since that is what everyone wants. You could count both and make sure everyone get their fair share.

 

So what is a fair share? First, you need to make sure that the share is counted properly between employees. A part-time employee should not work the same amount of night shifts as a full-time employee, but should work the same relative number of night shifts than a full-time employee. A fair share can be counted in percentage (night shift count divided by all shift count). If you exclude absences from the total shift count, then you have a value you can share with all.

 

Now some will get to be picky, and say they worked 25% night shifts this period and buddy here just worked 22%, so next period that employee should be lower. What you can establish as a fairness of distribution is to plan against a rolling average of percentages using the last 10 periods for example. This way, an employee who worked more night shifts than others would have to work less the following periods in order to reduce the rolling average and keep everyone at the same level.

 

If you need to equalize what employees are trying to have, one methodology is to keep a rolling counter. Usually, employees fight for things that give more money like overtime. If you need to equalize overtime, you need to offer the same fair chance of overtime to everyone. Since overtime is not always planned for, a planner needs to keep counters on who should be called next. One of the simplest ways to keep track of who is next on overtime is to keep a rotating list of calls. You establish the list of employees in a certain order and keep moving on to the next employee to offer overtime shifts. Notice that I used the verb ’offer’ instead of ‘assign’. This is because employees will refuse overtime for all sorts of reasons and although they refuse, they would always be at the top of the list. This causes the planner to call that employee all the time even though the planner knows for sure to get a refusal. By counting what’s offered (overtime accepted or not), the number of calls for the planner is reduced while keeping the staff with a higher sense of justice.

 

This rotating list can be counted in occurrences, but in some industries the overtime offered will never be of the same duration. In these cases, a planner should count the hours that were offered. The next employee to call is therefore the one with the least amount of overtime hours offered. Just like any other counter, this one also needs to be on a certain horizon. It can either be reset at 0 at the beginning of each year or be on a rolling horizon of the past 10 months for example.

 

Whatever you decide on what your fairness counters should be, you also need to define the rules that dictate how to count and what to do in rare cases. For example, where do you insert a new employee in the overtime list with what value? That new employee has 0 hours and would always be called until enough hours are offered for that person to catch up to the others. Now that would not be fair. Most businesses pre-assign this new employee with the highest value on the list (so they would be last as of their first day of work) or assign the average value of all employees.

 

Either case, the point is that some guidelines are required around the interpretation of fairness counters.

Sequence Constraints

· · Leave a comment

The constraints on sequences of shifts are used to control the different types of shifts that employees would work consecutively. For example, if employees share the night shifts and all have to work them, you may need to control the maximum number of consecutive night shifts.

 

Sequence constraints can also apply to mandatory sequences. For example, it is mandatory to have 2 days off after 3 night shifts. You may also have forbidden sequences where employees can’t work a day shift after an evening shift.

 

In short, sequences control the different combinations of shift types a planner can use to build a schedule. These controls are usually in place for both fairness and fatigue. When an employee works on all shifts, any schedule they get will be difficult and full of mistakes from the employee’s perspective.

 

Planners need to protect themselves by stating clearly what sequences are controlled, for fairness and/or fatigue.

 

Horizon Constraints

· · Leave a comment

As you can see, we started with a shift, then to a day, moved on to a week, and we could then move on to bi-weekly, and so on. These constraints can be combined into one single concept.

 

If you think of time as one long endless axis, a week is simply a delimitation of that axis, or a chunk of that axis of time. The day that we just discussed earlier is exactly that: a chunk of 24 hours on the axis of time. So we could apply exactly the same logic to any time span we know of like the week, the month, the year, etc.

 

Therefore, this section is called horizon constraints. The horizon represents that chunk or span of time that the constraints will be measured against.

 

A horizon is determined by three important aspects:

  1. The span: the length of the horizon (a week, a month, a specific number of days, etc)
  2. The offset: when does that span start on that axis of time? Is it Monday, Sunday? If it’s longer than a week, which date do you start counting? You need to determine a date or a day that says: this is where we start counting.
  3. The step: this one is a little tricky. The step is the value to add to the offset to get to the next time span to validate. In most cases, you don’t need to think about this one because once you’ve defined your span to be a week for example, you simply count on that week and never overlap the next week (the step has the same value as the span). But in some industries like transportation, safety comes first. Therefore, there is a maximum number of worked hours that is measured against a ten-day period, but that is for any ten-day period. That means that if a planner looks at work from February 1st to February 10, the schedule would show an employee working say 50 hours. The planner also needs to look at the amount of work from February 2 to February 11 and make sure the sum of work still does not exceed the maximum. That would therefore be a constraint that has a horizon of 10 days and a step of 1 day.

 

It’s important not to confuse the definition of a constraint, or how you count the constraint, to the time span it applies on. For example, if you have a minimum number of weekends you need to count, the time span is not the weekend. The time span is how long you measure the weekends for. The weekend is in the conditions that make the constraint count or not.

 

Examples of Horizon constraints:

Maximum of 80 hours of work on a bi-weekly basis: In this case, the span is 14 days, the offset is a date in the past that is a Sunday or Monday (depending on your week start), and the step is 14 days.

Maximum of 140 hours of work per month: Span is monthly, offset is a date in the past that is usually in the first days of the month. You don’t want to count a monthly thing starting on the 29th of a month because it gets confusing for a planner when February comes along. Best practice is to pick the first of the month when you count monthly constraints.

Minimum of 5 days of rest any 25 days of work: Span is 25 days and the offset could actually be any day in the past since the constraint says ‘any 25 days’. The planner will have to measure all spans of 25 days across the schedule both when creating it as well as maintaining it.

 

Week Constraints

· · Leave a comment

A week needs a frontier just like the day has one. Usually, the time you have selected for the day will be the same for the week; you therefore only need to pick the day that starts the week (which in most cases is either Sunday or Monday).

 

Ideally, your week definition will be the same as the week on which you pay on. That is just to keep things simple. The most important thing for employees after their schedule is their pay check so if it’s easy for them to follow both the schedule and the pay check together, all the better for everyone. But the pay week and the schedule week are actually two totally separate concepts. The schedule week is the one actually driving things like overtime pay and bonus pay. The thresholds on overtime are and should be counted based on the scheduled time span. This means that if you schedule from Sunday to Saturday, nothing is stopping you from counting everything based on that week, and then pay based on a Thursday to Wednesday week.

 

You may wonder who in their right mind set would do that. Well again, it depends on the business. Some scheduling constraints are based on a 10-day pattern and overtime is calculated based on that 10-day time span. As you can imagine, the employees are not paid every 10 days. This happens in areas where 24/7 coverage is required and the schedule is split in standard cycles (see chapter 5) that are not dividable by 7 (a standard week).

 

You’ll also need to establish the definition of regular days on and regular days off. Just like we explained on how to associate a shift to a day in order to count hours, you also need to define what is a day on and what is a day off so that you can eventually count the constraints that are about days.

 

Most of the time, a day on is a day with an associated shift and a day off is one that has no associated shift. If you have selected to divide the hours of shifts that overlap your day frontier, then you may decide that a day on is a day that contains regular worked hours.

 

Some even elect to differentiate depending on when the shift was scheduled. For example, someone schedule to work Monday to Thursday on a day shift will have 4 days on and 3 days off in the week at the time of posting the schedule. If extra work comes along and you add a shift to that person on Friday, this shift is scheduled on a regular day off and would be subject to premiums including overtime even though the employee still has not reached 40 hours. It is simply because that employee was not planned to work. So the Friday, even though has a shift, remains a day off.

 

Although the example sounds weird, it exists. And the planner has to keep different definitions of days on and days off and be very clear on which one applies to which constraint. In these cases, the planner would be very precise in the definition of each constraint to explain how it is counted against both the hours, the occurrence, and the publication time.

 

The weekly constraints are therefore about sums of daily constraints and also about sequences of events that occur in the week:

  1. Minimum hours in a week: As mentioned in chapter 3, employees may be hired with a minimum hours of paid hours in a week. Note that this minimum is not of hours worked, but hours paid and therefore include vacation or other time-off.
  2. Maximum hours in a week: This constraint can be split in two. There is the maximum number of regular hours and the maximum number of hours overall. A 40-hour full time employee will have a maximum number of hours of 40 hours. That’s easy. But when you need to schedule that employee in overtime, what becomes the real maximum where an employee starts being burned out? 55 hours? 72 hours? There is a second maximum that is the sum of all hours worked regardless of how they are paid. That second maximum usually applies after the schedule is posted when employees are getting called for overtime shifts (even though sometimes, overtime can be scheduled).
  3. Number of shifts per week: The number of shifts worked per week is usually constrained by a maximum. Some industries offer a guaranteed minimum amount of shifts instead of a minimum number of hours, but most are counting shift occurrences to control a maximum.
  4. Maximum number of consecutive days on: As you would imagine, someone can’t work every day on and on with no day off. This maximum can be interpreted in different ways. It can be the number of days on scheduled at the time of posting and excluding overtime. It can also be the number of days on overall, meaning that the week is not a boundary. If the employee works Friday and Saturday this week, and also Sunday and Monday next week, that is a total of 4 consecutive days that span over two weeks. Some industries keep the maximum to one week and do not look at previous or next weeks. In that case, the example above would only show 2 consecutive days.
  5. Minimum number of consecutive days off: All the constraints we have seen so far are about things happening. Days off occurrences are about things not happening. When nothing happens, you count the day off based on the definitions you established earlier. This constraint of giving a minimum number of consecutive days off (most likely 2 days off) is mostly for employee satisfaction. We usually enjoy a weekend, but for people who work shifts, there is no fixed weekend necessarily. Therefore, trying to give them consecutive days off is one way to increase employee satisfaction.

 

Day Constraints

· · Leave a comment

Before listing the constraints of a day, we must define what a day is. A day, for most of us, starts at midnight and ends at midnight. Then the date changes and we are on another day. That’s the easy definition.

 

Some businesses that need to stay open 24 hours a day 7 days a week will have a different definition of the day. For example, a casino has its peak of activity at midnight which makes things difficult to break down if a planner needs to split the hours when most employees are scheduled. In these cases, the split of the day happens at the lowest point of business in a day. Again in a casino, this would be at 5am or 6am. Hospitals and manufacturing also measure and count schedule constraints against a day that fits standard shifts. This would typically be at the end or the beginning of the night shift. For example, if the night shift is from 11pm to 7am, you either count that shift to be part of the day it starts (making the day be from 7am to 7am) or to the day it ends (making the day be 11pm to 11pm and making one hour of the previous day belong to the day).

 

Note: try not to make that frontier 2am or 3am. These hours are when time changes occur twice a year in most areas of the world. Having a frontier at 2am will make things confusing when that 2am becomes 3am when daylight savings time kicks in. You’ll feel like an hour has disappeared into oblivion.

 

Once you’ve established what the day is, you also need to determine how the shift is tied to that day. If your constraint says that employees can work a maximum of 12 hours in a day, how do you handle the shift that overlaps that day frontier? Let’s say that day frontier is midnight and your employee is scheduled for an 8-hour shift starting at 11pm. There is one hour on the day the shift starts and seven hours on the day the shift ends. Therefore, you need to establish the rule on how hours are counted for the employee. In most cases, all hours will be counted towards the day the shift starts, but I’ve even seen businesses determine the day of a shift based on where does the middle of the shift falls. For example, a shift from 7pm to 3am would belong to the day it starts (the middle being 11pm), but a shift from 9pm to 5am would belong to the day it ends (the middle being 1am).

 

Now that you have established how to count, we can take a look at what to count:

  1. Minimum hours in a day: Although it may sound like the same constraint as the minimum shift duration, it is quite different. First, it depends on how you count hours in the day. Second, you may have split shifts. A split shift is when an employee works a few hours in the morning and a few hours in the afternoon. Typical employees working split shifts are bus drivers where they work during rush hours. So although you have a minimum duration of 3 hours on a shift, you may have a minimum duration of 6 hours on a day for example. So if you schedule a 3-hour shift, you need to schedule a second one on the same day to respect that daily constraint.
  2. Maximum hours in a day: For the same reasons, and again depending on how you count your hours, you may have to define a maximum hours per day constraint. This is also different than the maximum shift duration constraint since split shifts may come into play (see above).
  3. Minimum rest between shifts: This constraint is the most common. How much rest time do I need to give an employee before scheduling them for another shift? This minimum rest is the duration between two shifts where you expect them to go home (this is not the separation time between split shifts). You need to establish a proper value based on the average shift duration you normally schedule and the expected travel/rest time anyone would require before being back and productive. The longer the shift duration is, the longer the minimal rest should be. You may also have different minimum rest values depending on the shift that is worked. For example, after a night shift, you may force a 12- hour rest period instead of the standard 8 hours. The same goes to have a different rest period apply before a shift. If the day shift is particularly busy, you may have a 10-hour rest period before the shift is started to make sure the employees come in rested and productive.
  4. Maximum number of shifts per day: This constraint will be required depending on how you associate shifts to a day. If you schedule an 8-hour shift starting at 7am and another one for the same day starting at 11pm, you have two shifts starting on the same day with an 8-hour period in between. You have respected the minimum rest rule but the employee would end up being awake for 24 hours in a row basically (not much time to go home eat, sleep and come back). Also for split shifts purposes, you may define how many splits may occur in a day by having a maximum number of shifts per day.
  5. Shift separation: Shift separation is for split shifts. As explained above, a split shift is sometimes required by the business simply because of the workload profile that has multiple peaks in the same day. Although these are good for the business, they are usually not very popular with employees. With this type of shift, you need a reasonable minimum and maximum duration between the two parts of the shift. The minimum value is driven by the business. If the minimum is too low, then the separation becomes one long break that the employer would have to pay. The maximum value is driven by employees. If the maximum is too high, then employees are away from home for a very long duration. Thus, a reasonable balance is required. Some employers pay up to 25% the duration of the shift separation in order to be able to schedule them and not end up with outright refusals from employees. This all depends on the business requirements and demographics that are involved.