Periods

Activities in a Network model occur within time periods (Activities and Periods). The Periods page allows you to define these periods.

The period duration is calculated from the start and end dates of each period. If the period spans a daylight-savings time change (based on the current culture of your operating system) the duration may not be a whole number, as it includes the loss or gain of time associated with the change in daylight savings.

You can deactivate periods by clearing check boxes in the Active column. When a period is not active, any supplies or demands in that period (and their constraints) are not applied. Similarly, period-specific constraints are not applied.

You can assign periods to an Optimization Group. Periods in the same optimization group are optimized together as a single classical multi-period optimization, while periods in different groups are optimized sequentially according to the period start and end date sequence.

Periods which do not have a group assigned are allocated to a default classical optimization group (Classical versus Sequential Multi-Period Optimization).

Tip: Dates can be copied and pasted from the clipboard. It may be easier to setup the required dates in another application such as Microsoft Excel, and then copy and paste them across to the Periods table.

Add and remove periods

To add a period, do either of the following:

• Click Add Period in the Periods group of the Home ribbon tab.

• Right-click anywhere in the grid and choose Add Period from the context menu.

A new row is added to the grid. You can then edit the row cells to customize the new period.

To remove one or more periods, follow these steps:

1. Select the rows corresponding to the periods you want to remove.

2. Do either of the following:

   • Click Delete Period in the Periods group of the Home ribbon tab.

   • Right-click anywhere in the grid and choose Delete Period from the context menu.

3. Click Yes in the confirmation dialog box that appears. The periods are removed from the grid.

Enable variable period lengths

By default, the periods you create have fixed lengths. You can however enable variable period length optimization in the Periods page. To see the relevant controls, select All from the View list in the Grid group of the Home ribbon tab.

When you select the All view, new Minimum and Maximum column bands appear, allowing you to define a range within which each period length can vary. To set a period to have variable length follow these steps:

4. Clear the Active check box in the Fixed column band.

5. Enter at least a Minimum constraint value, and optionally a Maximum constraint value as well.

6. Enter the correct Model End date and time in the field above the table.

When you have fixed period lengths, the Model End field is read only. This is because its value is calculated automatically from the Model Start value and the period lengths. With variable period lengths, you must ensure that the overall duration of the model simulation is compatible with the constraints on period lengths. For example, if you define an overall simulation period of 20 days, and three periods with 5 days maximum length each, the case cannot be optimized and an error appears in the Diagnostics window.

Note: When you enable variable period length, you may no longer use any quantities that are defined by date, but only those defined by period. The reason is that a quantity defined by date cannot be assigned to the right periods, since the period boundaries themselves are subject to change during optimization.
For example, supplies and demands have start and end dates, and therefore cannot be used with variable period length. On the contrary, you can use purchases and sales included from embedded plant cases, since the plant cases themselves are defined with start and end periods on the Plants page.

Lock solutions

Solution values can be locked during subsequent runs. This means that a solution value will not change in any subsequent run, even if other input to the run is changed.

Example: The flow of naphtha from a crude could be locked, and in subsequent runs it would not be able to change. Therefore, if a fixed swing cut point was changed and the model re-run, the crude blend would have to change in order to match the fixed flow rate.

Locking a solution value acts in a similar way to fixing the value. The advantage over fixing a value is that the lock can be clearly seen in the user interface, so it is simpler to distinguish true fixed values from those which are temporarily locked.

To lock one or more solution values, select the corresponding rows in the table and do either of the following:

• Click Lock Solution in the Behaviour group of the Home ribbon tab.

• Right-click on the table and choose Locking > Lock from the context menu.

To unlock one or more solution values, select them and either click Unlock Solution in the ribbon or choose Locking > Unlock from the context menu.

Customize the grid

Customize the grid from the Grid group of the Home ribbon tab:

• Click Find to show or hide a text field at the top of the grid to filter the grid contents. Type in the text field to show only the rows with text matching what you have typed. The matching text is highlighted in the grid. See Filters for more details.

• Click Columns to open a small dialog windows where you can drag column headers to remove columns from the grid. You can then drag the headers back to the grid to restore the columns. See Add and Remove Columns in Manage Columns for more details.

• Choose a View from the list, or click the cogwheel icon to remove a view or save any changes. See Views for more details.

Optimization across and within periods

A network activity can be defined to occur within a specific period or across two or more periods.

If a network activity occurs within a specific period, the amount of activity is distributed evenly across that period. For example, a period may be defined with 7 days duration, and a demand of 700 tonnes of gasoline entered. This demand is divided across the period as a relative demand of 100 tonnes/day.

It is not possible to define an activity to occur within a subset of the period. For example, it is not possible to force 350 tonnes/day of gasoline production within the last two days of the above period. Network activities are always averaged across their allowable timespan. To force gasoline production in the last two days, split the seven day period into two periods of five and two days, and apply the demand to the two-day period.

If a network activity spans multiple periods, by default the activity amount is pro-rated across the different periods relative to the period durations. For example there may be two periods, one of 7 days and one of 14 days. A gasoline demand of 2100 tonnes is applied across these two periods. The amount is divided between the two periods based on their length, so both periods have an applied demand of 100 tonnes/day.

Certain grouped items can be set so that they are not pro-rated. In this case, the amount is a sum across all periods. If a grouped supply had a fixed amount of 2100 tonnes where the supply is pro-rated, the supply in two 7- and 14-day periods would be 100 tonne/day. Where the supply is not pro-rated the only constraint would be that the total production would be 2100 tonnes. The amount could be 2000/100 tonnes per period, or 1500/600 tonnes per period, and so on.

The ability to set items as pro-rated or non-pro-rated applies to:

• Supply groups

• Demand groups

• Exchanges

• Terminal inventory groups

• Transport link groups

When Pro Rata is active, the constraint is applied equally across all periods, but weighted for period duration, so the same amount per day is used:

When Pro Rata is inactive, the constraint is applied to the aggregate total but can be utilised unequally in different periods:

Warning: If you deselect the Pro Rata option, you cannot use hybrid multi-period optimization.

Hybrid multi-period optimization

By default, optimization in a multi-period model is classical. The periods are all connected within a single optimization, and initial periods can see later periods and optimize their variables accordingly.

You can also define periods to be run as part of a sequential optimization, in which each period is optimized independently and any inventory is passed to subsequent periods. When using sequential roll-forward optimization, you are responsible for ensuring that appropriate inventory is available in subsequent periods to achieve feasibility. Within a large period sequence it is possible to run a mixture of sequential and classical optimizations, known as hybrid optimization.

To enable sequential multi-period optimization, each period must be associated with a different period group. Periods that share the same period group are optimized as a single classical multi-period optimization. By default, all periods are allocated to the same period group. To allocate periods to different sequential period groups, enter a unique period group name in the period list.

Periods with the same period group identifier are optimized as a single classical multi-period optimization. Final inventory from this classical optimization is passed to any subsequent period groups. Therefore, if following a classical multi-period optimization with another optimization, it may be necessary to set final inventory limits in the last classical period.

Periods with no group identifier are optimized as part of a default classical optimization group.

Warning: Periods in the same period group must be neighbors in the period sequence. For example, it is possible to assign January, February and March to the same period group, but it is not possible to assign January and March to group A and February to group B. if February is in group B, March must either be in group B or be in a new group C. Period sequence is determined by the start and end dates of the periods. The group identifier is not used to determine period sequence.

The following tables indicate how periods are assigned to optimization groups based on their group identifiers.