Setting up a model requires basic data about the product and process characteristics. A unique characteristic of these simulators
is their capability to use spreadsheets as an input or output mechanism, so that changes in parameters can be made in a familiar
format. Figure 1 shows a high-level model construction fed by the process times and rules specified in the spreadsheets.
There were multiple product types planned for the facility. Liquid products in either vials or syringes do not get processed
through the lyos; they are formulated, filled, and then capped. An assumption was made that if lyophilized syringes were filled,
there would be a separate loading system to the lyos. Lyophilized products in either vials or syringes go through all processes.
Some of the parameters to describe the process included:
- Formulation times and rules: SIP is done before processing starts and formulation tanks are cleaned by the CIP system after
completion. Formulation of the next batch may begin while the cleaning and changeover in filling is in progress. Maximum
hold times were also a constraint in scheduling when to start formulation. Liquids batches had to be completed filling and
capping within 24 hours of formulation completion. Lyophilized batches had to be in the lyos within 32 hours of formulation
- Filling and capping: Times were specified for VHP decontamination before a batch, and manual cleaning after and changeover
times for different products. VHP decontamination took approximately eight hours, and had to be done within 24 hours prior
to starting. Filling of a liquids batch could also not start until both the filling and capping lines were completed cleaning
and changeover. Filling startup times included line purging, weight checks, bringing up to temperature, and also had some
estimated yield loss. As shown in Figure 1, the number of both filling and capping lines were variable, from 1 to 3 each.
Other key variables were the lines speeds (vials or syringes per minute) that the lines were run at.
- Lyo systems: The number of systems was variable (4–6), as was the freeze dry time (3–5 days), size of shelves per lyo (33–58
square meters), and number of shelves. For example, capacity of a 40 square meter lyo was 58,000 vials, 16 shelves with 3,000
vials per shelf, or 1,450 vials per square meter. Similar decontamination rules before and a one-day cleaning process after
applied to the lyos.
- Carts and conveyor systems: For the robotic cart system, the number of carts, number of CIP systems, the track configuration,
and the locations of the filling and capping lines were all parameters to be changed and tested. Liquids products were moved
from filling to capping, bypassing the lyos. The track or conveyor layout dictated which lyos and sequence possible could
be practical for loading and unloading. The conveyors were fixed and carts could not cross each other on a single track. Thus
the flexibility to designate different sets of filling and capping lines dedicated to specific lyos was included. The filling,
loading and unloading, and capping times varied by configuration also, since the distance varied. Therefore, scenarios included
defined routings allowed, which defined how many lyos could be loaded or unloaded simultaneously; e.g., with six lyos, 3 &
3 indicates two simultaneously, one from any of the first three and one from any of the second three.
The carts and conveyor had similar cleaning and VHP decontamination rules. Carts used CIP systems, of which one or two were
considered, and required decontamination before the next use. The fixed conveyor system had multiple zones possible (up to
five considered), and the cleaning and de-contamination had to be done sequentially.
All processing rules and cleaning are dependent on having skilled people available at the times required. Early runs of the
model showed a high dependency on crewing because of the asynchronous nature of events occurring. The flexibility to change
crew sizes and scheduled days per week, shifts per day, and hours per shift was included.
The value add of using simulation becomes apparent when the overall process rules and variability that exists in the actual
operation are added to the problem definition. The asynchronous nature of a complex system like this is impossible to "get
your head around" all of the parameters without it.