Functionality¶

What is a split activity?¶

In general it is expected that activities that require scheduling will be completed by a single resource in a single visit. The resource will arrive at the location for the activity, carry out any work that is required and once the visit is complete the activity should also have been completed.

However there may be some activities for which it is expected that a significant amount of time will be required and it is therefore expected that the resource will need to visit the activity multiple times to complete the work. Any such activity, which is expected to be split into multiple visits, is defined as a 'splittable' activity. This can be applied to any activity of class 'CALL'.

The exact number of visits required will depend on a number of factors, for example:

• The expected working time required to complete the activity.
• The length of the resource's shifts.
• The travel time to and from the activity.
• The time of day that the resource starts working on the activity.

For this reason the DSE will calculate itself how many visits are required based on the input data supplied and the current working schedule. Of course, even if an activity is splittable, the DSE may be able to schedule it in a single visit. Where more than one visit is scheduled the activity is defined as a 'split' activity.

Reasons for splitting.¶

There are two basic methods of splitting an activity:

1. If the activity cannot be completed before the end of a shift it is split into the following shift.
2. The activity can be interrupted to allow the resource to carry out another activity, for example:
   • A break activity, such as a lunch break.
   • An appointment that must be attended within a short timeframe.
   • An urgent activity that must be attended as soon as possible.

The behaviour that determines when an activity can be split is controlled by a variety of conditions defined via the input data. Some of these are hard constraints, and others are preference options to encourage the DSE to schedule in a certain way.

All normal constraints that apply to non-splittable activities can also be applied to splittable activities. This includes skills, regions, availabilities, SLAs and pre- and co-requisite constraints. The status of a visit to a split activity can also be updated to reflect the current dynamic state of the visit.

A call may have a duration overhead linked to it via its location (e.g. for meet and greet, site entrance delays, preparation etc). If the call is split, for each split visit that has a start location that is different to the previous location of the resource, a duration overhead is added to the split visit duration. This means that the expected duration of all the split visits will be longer than the initial expected duration if one or more split visits have longer durations due to duration overhead.

Simple Example¶

As a simple example we can consider an activity with an expected duration of 6 hours. If the activity is not splittable the DSE chooses to schedule it on the second day, since there is not enough time remaining on the current day to complete it.

If the activity is instead allowed to be split across shifts, the DSE is able to schedule it earlier, starting immediately after the break on the current day. However it is not scheduled to start before the break because at present the break is not allowed to interrupt the activity.

If the break is also allowed to interrupt the activity the DSE then schedules the resource to start travelling to the activity immediately, and the activity is split into a total of 3 visits.

Controlling behaviour¶

There are a rich variety of options to control how the DSE will schedule a splittable activity. These constraints can all be specified by individual activity in the schedule input data.

Basic Constraints¶

The DSE will always schedule every visit to a splittable activity such that the total duration of all visits is at least the expected total duration required to complete the activity. By default, these visits will all be scheduled to the same resource. In addition there are three basic constraints:

1. In order for the DSE to consider splitting an activity it must be marked as splittable, so by default the DSE will only ever schedule a single visit for an activity.
2. In order for the DSE to consider splitting an activity into a later shift, the shift must be marked as splittable. Therefore if a resource has Monday-Friday shifts, and the Monday shift is not splittable, the DSE would not be able to split an activity across the weekend.
3. In order for the DSE to interrupt a splittable activity with another activity, the second activity must be marked as interrupting.

When splitting across shifts, the DSE will only do this if the schedule would otherwise be invalid. If the remainder of the activity can be completed within the remaining shift time (including any overtime), and the travel home can also be completed within the allowed time, then the DSE will not attempt to split the activity into the next shift.

When interrupting a split activity, the DSE will do this if either

1. The interrupting activity cannot be validly scheduled after the splittable activity, or
2. The loss in the value of the interrupting activity by scheduling it after the split activity exceeds the 'cost of split' of the splittable activity. This allows urgent activities to interrupt a split activity even if they could validly be scheduled after the splittable activity.

Minimum Visit Duration¶

Each visit to a split activity must be of at least a minimum length, which is again configurable by activity. This applies to every visit to the split activity including the last visit, which may mean that the total scheduled duration of the activity is greater than the expected duration.

For example, suppose an activity is expected to take 8 hours with a minimum visit duration of 2 hours. If the DSE can only schedule 7 hours in the first visit, it will schedule a 2 hour visit the next day, making a total visit duration of 9 hours.

Applying a Split Cost¶

Each time the DSE splits an activity it will apply an additional cost to the total value of the schedule. This is used to encourage the DSE to use fewer visits where possible. If there is no preference the cost of split can of course be set to zero so that no penalty is applied. Furthermore, any costs applied will be balanced against the other values and costs which apply in the schedule, as the DSE tries to maximise the total value of the schedule.

For example suppose an activity is expected to take 7 hours to complete, so can be completed within a single day provided it is started first thing in the morning. The DSE can validly schedule the activity either in the afternoon of the current day (so split over two days) or as a single visit starting first thing the morning after. The DSE will simply choose whichever option results in the highest value schedule, which will depend upon:

• The cost of splitting the activity into two visits.
• The cost of travelling to and from the activity.
• The value loss in not scheduling the activity until the morning of the second day.

Therefore if the SLA deadline for the activity was on the first day, the DSE may well choose to split the activity into two visits in order to meet the deadline. However if the deadline was not until the second day it may choose to schedule it as a single visit and so avoid the cost of split and extra travel costs that would be incurred.

Split Priority¶

There may be occasions where finer control is required over when an activity may be split. For example a schedule may include some activities which must be scheduled within a single week, and others that are allowed to span multiple weeks. This is achieved by using split priority. Any splittable activity can be assigned a minimum split priority. It will then only be split into a later shift if the split priority of the shift is at least the minimum priority.

In our example the Monday shifts can all be assigned a split priority of 1, while the other weekday shifts have a split priority of 2. Activities that must be completed within a single week are then given a minimum split priority of 2, and so cannot be split over the weekend and into the Monday shift. Meanwhile the other splittable activities have a split priority of 1, so can be split over the weekend.

Similarly activities which are allowed to interrupt can be given an interrupt priority, and will only be allowed to interrupt a split activity if the interrupt priority at least matches the minimum required split priority.

Split Multiplier¶

It is also possible to exercise finer control on the cost that is applied when an activity is interrupted by another activity. This is achieved using a multiplier on the cost of split which depends on the interrupting activity. The value defaults to 1, but can be set to any non-negative value.

For example, suppose we apply a cost of 100 when an activity is split over shifts, however we do not wish to apply any penalty if the activity is interrupted by a break activity. We can set the split multiplier on the break activity to 0 to achieve this.

Suppose we also have an appointment activity which is allowed to interrupt, but we only want to do this if there is no other alternative. We can therefore set the split multiplier on the appointment activity to perhaps 2.5 and a cost of 250 will be applied if the appointment activity is scheduled to interrupt.

Splitting Across Multiple Resources¶

There may be cases where an activity should be allowed to be split across multiple resources. For example, supervision work may require another resource to take over when the initial worker goes off shift.

When a follow on visit is required and this option is enabled (by setting the maximum number of resources for the activity), the DSE will attempt to schedule it to the best resource from the ones that are available to be split to.

The above image shows a single activity spread to three resources.

By default, the DSE will only consider resources with its next shift being earliest, so as to immediately resume the work if possible. For example, if there are two resources with the first having a shift at 9:00am, and the second at 9:30am, the second resource will not be considered for a future visit. Changing the resource leeway time will allow the DSE to account for small differences like this. In this example, setting that value to 30 minutes would allow both resources to be potential candidates for future visits.

If it is preferable to keep the work with the same resource, it is possible to set a penalty to the scoring to disincentivise the DSE from scheduling to multiple resources where possible.

An activity can have a maximum delay set on it, such that the split visits may only happen if the time between the first visit and second visit is less than this delay. Specifying this amount as 0 will ensure that the activity must be uninterrupted and that the DSE will only schedule the split across multiple resources if there is availability immediately with another resource.

In addition, a delay cost per hour can be set, resulting in the DSE applying a penalty for the time where an activity is waiting for a visit. This will encourage the DSE to schedule other visits as soon as possible, without enforcing a maximum delay period.

Handover time can be specified to allow an overlap period where one resource can hand over the activity to the second. This time will be in addition to the required time for the visit. For example, if an activity has 3 hours remaining, and a handover time of an hour, the visit for the second resource will be 4 hours in length, with the first hour of the visit overlapping with the visit of the first resource.

Dynamic Updates¶

The DSE is by nature a dynamic scheduling engine, and when running schedules dynamically it is expected that the latest status of each activity will be fed into the system and the DSE will adjust its current schedule accordingly. Dynamic updates are also fully supported for splittable activities, where the status of each visit to the activity can be updated individually.

The diagram below shows all valid statuses for an activity. Two of these relate specifically to splittable activities. These are the status of 'Follow On' (status 20) and the status of 'Visit Complete' (status 68).

Follow On Status¶

The status of 'Follow On' is used by the DSE once at least one visit to a splittable activity has been committed (i.e. is at a status of at least Committed). The DSE will determine the start time for the committed visit following the normal rules for committed activities. For example the start time of a visit at status travelling will be calculated based on the travel start time and the estimated journey time.

However, for a splittable activity it is not necessary to complete the entire activity in this one visit, and the DSE will continue to apply the constraints and preference options described in the previous section in determining when the committed visit should end. This includes both splitting the activity into another shift and interrupting the activity with another activity (which may or may not be committed).

Any future visits scheduled for the activity are not themselves committed, but there are restrictions on how the DSE is able to schedule these visits as a result of the committed visit. The DSE must use the same resource as is used for the committed visit and can only split or interrupt visits where the constraints allow. For this reason all future visits are returned with a status of 'Follow On'.

Since the DSE is still able to interrupt committed visits to splittable activities, the duration of the committed visit may change as a result of updates to the schedule. For example if a new urgent activity is entered, the DSE may choose to interrupt the committed visit with this urgent activity, and so shorten the length of the committed visit. The minimum visit length will still apply, so in the example below the urgent activity is scheduled to start once the resource has completed one hour's work on the splittable activity.

There are now two later visits to the split activity, and both are given a status of 'Follow On'.

Visit Complete Status¶

The status of 'Visit Complete' is used to indicate that a visit to an activity has been completed, but the activity itself is not yet complete. In this case a follow on visit will always be scheduled with a duration of at least the minimum visit duration. This applies even if the total time spent on the activity already exceeds the expected duration.

Once the activity is itself completed the status of the final visit should be updated to 'Complete', and then no further visits will be scheduled. Any future visits that had been scheduled will be removed. For example, an activity may originally be split into two visits since it is expected to take two days to complete.

If the resource is able to complete the activity by the end of the first day, the status of this visit is updated to 'Complete', and the visit on the second day is then removed by the DSE.

As an alternative example, suppose an activity is originally scheduled to be completed in a single visit. The schedule below shows that the resource is already running over slightly on the activity and is approaching the end of his shift.

If he is unable to complete the activity in this visit, the visit status is updated to 'Visit Complete'. The DSE then schedules another visit to the activity on the second day, with length the minimum visit duration of one hour.

Updating Multiple Visits¶

The status of each visit can be updated individually and combined with status updates to other interrupting activities. In the example below the first visit is completed and the resource has now started a break. The second visit is downloaded to his PDA and there are third and fourth follow on visits scheduled for the second day.

Invalid Fixed or Committed Split Visits¶

Once an activity has one or more committed or fixed time and resource visits, it is possible that the follow on visits for this activity cannot be scheduled validly. If this situation arises, the DSE will attempt to schedule as many visits as can validly be scheduled. If it cannot validly place another visit, it will place the remaining time at the end of the last valid visit. Note that this will result in an invalid schedule, and the schedule exceptions list will be populated accordingly.

As a simple example of this, consider one activity with a duration of 20 hours, with two shifts of 8 hours each available. We then fix the start of this activity to be the start of the first shift. The DSE will first split this activity across the two shifts, and will be unable to schedule the final 4 hours. Rather than cancel the split completely, it will instead merge these 4 hours with the final valid visit. Overall there will be two visits of this activity, one for 8 hours in the first shift, and one for 12 hours in the second. This is demonstrated in the image below.