The Expansion Was Not Limited
By the Line They Were Buying
A craft brewery growing out of its brewpub wanted to add kegging and bottling — new markets, contract brewing, real capital. The model said the new lines were not the problem. Install either one and it would over-run the fermenting and serving tanks, which nobody was planning to buy.
The plan
Craft brewers typically start with on-premise sales for quick return, then add kegging and bottling to reach off-premise markets and take on contract brewing for other labels. Growing that way means committing to an increasingly flexible operation: more flavours, more packaging formats, faster changeovers, flow routing that can be reconfigured on demand.
Each of those adds complexity to a production network that was previously simple. The question in front of management was whether the expansion, which required significant capital, would return what they expected.
Where the constraint actually was
Brewing has a lopsided time structure that is easy to underestimate. The active brewing steps — mash and lauter tun, kettle, whirlpool — are fast. Fermentation and aging are not. They are, in the paper’s words, so much larger that they dominate how much variety the operation can hold at once.
The utilisation trace showed that installing either a kegging line or a bottling line would over-run the fermenting and serving tanks unless capacity was added there first. The bottleneck of the expansion sat upstream of the expansion.
This is the same shape of finding as an olive plant discovering its constraint was downstream in packaging rather than at the pitting machines: the equipment under discussion was not the equipment that decided throughput. It is a hard result to reach by inspection, because the tanks were already installed and already appeared adequate — they were adequate, for the product mix the brewery had before.
What it was worth
Because cost and revenue were tied to the process changes in the model, the brewery could see a dynamic view of financial performance across scenarios rather than a single static ROI. Management did not avoid the capital; they went ahead with more confidence, knowing which capital actually had to be spent and in what order.
The method
This is the class of problem that first motivated discrete-rate simulation. Conventional piece-and-part simulation models entities flowing through a queue, and beer is not entities. As the paper puts it, attempts to aggregate bulk product into entities produce models that either run too slowly or are too inaccurate to use, while combined discrete/continuous languages push you into writing differential equations.
The brewery model used a rate-based architecture built for bulk and fluid flow — the lineage that leads directly to the engine in ReliaSim. The work was published as Modeling a Microbrewing Facility Using Discrete Event Simulation (PDF) by Andrew J. Siprelle, so the method and the reasoning can be read in full.
Related: The Hidden Bottleneck · Starved vs Blocked · All case studies