Every process, whether it's compiling code, assembling a car, or making a latte, has a speed limit. This limit is its capacity. Understanding capacity isn't just about knowing how fast you can go; it's about figuring out what's slowing you down. To do this, we need a few key metrics.

Let's imagine a simple coffee shop. The process of making a coffee can be broken down into a series of steps. By visualizing these steps, we can start to analyze the system's performance.

This is a process flow diagram. It maps the journey of a single unit—in this case, a customer—through the system. Each step has its own capacity. The capacity of a step is the inverse of its processing time. For example, the 'Take Order' step takes 1 minute per customer, so its capacity is 1 customer per minute, or 60 customers per hour.

In any sequential process, the overall throughput is limited by its slowest step. This slowest step is called the bottlenecks. It's the constraint that dictates the maximum output of the entire system.

Let's calculate the capacity for each step in our coffee shop:

• Take Order: 1 customer/min = 60 customers/hour
• Make Coffee: 1 customer/3 min = 20 customers/hour
• Take Payment: 1 customer/2 min = 30 customers/hour

The 'Make Coffee' step is the bottleneck. Even if the cashier can take orders faster, the system can't produce more than 20 coffees per hour. The throughput of the entire coffee shop is 20 customers per hour.

A good way to quantify this is by looking at utilization. It tells you how busy each part of the process is. If the system is running at its full capacity (20 customers/hour), we can see how it affects each step.

At a throughput of 20 customers/hour:

• Take Order Utilization: $20 / 60 = 33.3\%$
• Make Coffee Utilization: $20 / 20 = 100\%$
• Take Payment Utilization: $20 / 30 = 66.7\%$

The bottleneck step is always the one with the highest utilization. When the system is running at its maximum possible rate, the bottleneck's utilization will be 100%.

So, we know how many customers we can serve. But how long does each customer spend in the shop? And how long are they waiting in line? For this, we turn to a simple but powerful formula from queuing theory: Little's Laws.

Imagine on average, there are 5 customers in our coffee shop at any given time ($L = 5$). We already know the throughput ($\λ$) is 20 customers per hour.

We can rearrange Little's Law to find the average time a customer spends in the shop ($W$): $W = L / \λ = 5 \text{ customers} / 20 \text{ customers/hour} = 0.25 \text{ hours}$, or 15 minutes.

Our calculations so far assumed everything is perfectly consistent. But in the real world, things are variable. Some customers order complex drinks, and others just want a black coffee. A barista might be a little faster or slower depending on the time of day. This variability is what causes queues to form, even when the average utilization is less than 100%.

When arrival rates and service times are variable, the simple formulas get more complex. Two key factors are the variability of arrivals (how randomly customers show up) and the variability of service times (how consistent the workers are). The more variability, the longer the wait times for the same level of utilization.

Consider two lines at a grocery store. One has a cashier who is very consistent, taking almost exactly 2 minutes per customer. The other has a cashier who averages 2 minutes, but is highly variable—sometimes 30 seconds, sometimes 5 minutes. Even with the same average service time, the line for the second cashier will be longer and more unpredictable on average. Managing variability is just as important as increasing raw capacity.

Understanding these core concepts of capacity, bottlenecks, and variability is the first step in optimizing any system. They provide the quantitative foundation needed to model and improve complex business operations.