We all know that forecasts are always wrong.  So why do we make them, then?

Here’s a story about some decisions made in 1992 still relevant 13 years later, based on forecasts made then.

The question was “What freeze drying facilities should be provided to cater for lyophilization demand over the next decade?”

It’s always best to be flexible about plans for the future. Business changes, new products come and old ones go. If you are lucky, you can cope with most capacity requirements by hiring more people, working more shifts, renting some offices or using simple easily reproduced manufacturing processes. However, the chances are that you have got some processes that require large capital expenditure, take a long time to put into place and can’t easily be changed once you’ve built them.

Lyophilization is one such process. Great for ensuring long term stability and activity of reagents of biological origin, but requiring special equipment, elaborate validation and special facilities. So, when you are planning lyophilization facilities for product lines lasting more than a decade, you face some severe forecast challenges that boil down to “how many and what size should the lyophilizers be?”  Errors in judgment can easily result in running out of capacity too soon, requiring additional capital expense, not to mention a probable space problem (where do we put it?), along with having to rush an unrushable validation program.

Alternatively, you will have to justify surplus capacity “just in case” under the spotlight of the annual budgeting round.

A forecast is needed of the likely products and batch sizes, of course. What is needed more, though, is a recommendation that is least sensitive to this forecast being wrong.
Here’s an example. The author was asked to answer this question in 1992. In a new facility, how many and what size lyophilizers are needed? For the needs of the next 13 years!

There are a range of variables: 

• A new range of products under development
• Unknown new components to be lyophilized
• Volume growth or decline in each
• Batch sizes dependent on shelf life and demand
• Unknown marketing initiatives in the future
• Changes in lyophilization cycle time
• Regulatory rules as to what can and can’t be lyophilised together
• Existing components still in design that might or might not need lyophilizing

Forecasts produced a chart for total lyophilization volume as follows:

Fig 1

With the minimum forecast volume 20% of the maximum, there was at least a conviction that zero was not an option!   The business had faith in its long term future.

So, what were the options?

Build capacity for the highest forecast figure?

Build capacity for the most likely figure and supplement with some other capacity elsewhere in the organisation (maybe on another continent), should it be needed?

Find a subcontractor to completely outsource the problem?

Persuade the R&D department to design out the need?

Persuade the marketing department to discourage customers from buying lyophilised components?

The latter options were tested first: avoiding capital expenditure is always welcomed at board approval level. However, it was impossible for R&D or marketing groups to imagine a future where nothing was lyophilised, even if as a contingency. Experience with the biological components used in this industry had to rule here.  Subcontractors were sought out.  However, even those that could possibly have helped in the short to medium term could not be relied upon to provide long-term security of supply. Here was a strategic process that needed to be kept in-house in order to maintain that security. The subcontractor option was ruled out, at least for the “most probable” forecast.

Now the original question had to be answered: How many and what size should the lyophilisers be?

The forecast had produced an aggregate volume from a listing of current and projected components.  What it did not do was to clearly identify what batch size each would be made at.  This is crucial. If you want to lyophilise 20,000 vials a year, you could do it with different sizes of lyophiliser as follows:

• 20,000 vial capacity and do the task once a year, using approx 1/360 of its capacity with a drying cycle around 1 day.
• 5,000 vial capacity and make 4 batches in a year, using 1/90 of its capacity
• 60,000 vial capacity and lyophilise 3 different concentrations in the same cycle [if the product was supplied that way]

This provides many options.  What if product performance, at some time in the future, requires that a component must be made 4 times a year [perhaps its shelf-life has to be reduced, for example]? If you had specified the 5,000 vial capacity machine, then this would make no difference.  If you had specified the 20,000 vial machine, then you would still have to use it 4 times a year, but each time only filling it ¼ full.  Its capacity utilisation would be the same as the smaller one.  In this situation, why buy the larger one?

Multiply these contingencies by the full product projected product range, and you can see that building large to provide vial capacity does not necessarily provide capacity that can be used. At the other extreme, if you build too small, you will be able to use the capacity better, but at a price.  Sometimes you will have to make batches smaller than you really want in order to fit the lyophiliser space.  Making more batches to do this will incur the batch related expenses of doing so, e.g. more formulations, more QC testing.

So, how best to resolve this?

Using a spreadsheet to calculate the required number of batches across all the predicted products, the effect of lyophiliser size on the number required can be seen (Fig 2).

 
Fig 2

The exercise can be repeated with maximum and minimum estimates to give a spread. The number and size can be chosen to cover the worst case, without providing an excess.

In this case study, this was sufficient to make the choice and demonstrate its worth.
If necessary, the analysis can be extended to include balancing the capital cost of larger lyophilisers against the extra revenue expense of making and testing the additional batches that the smaller one will demand.

What Actually Happened?

After 13 years, the actual volume was within the forecast band.  During that period, volume had dropped off dramatically, whilst at the end it was on an upward trend.  Other completely unpredicted things happened too. New components were introduced (as had been expected), but these were only needed in very small quantities. The effect was to demand many more lyophilisation cycles, but each one only containing a small quantity.  With hindsight, these components could easily have been dried using a much smaller lyophiliser instead.  However, the machines were built and installed and once there could be used for anything. Such has been the case.  Capacity to still sufficient to meet current demand, and is certainly not too much.  The forecast has worked out right!

Now What?

At time of writing, the existing facility will be needed for several more years, but then may be replaced.  The 1993 question has been asked again.  It has been answered in the same way.

by James La Trobe-Bateman, reMODEL Consultants International Ltd