The call came at 4:47 PM on a Thursday in March 2024. I was packing up, mentally already home, when a client's quality director said: "We have 72 hours before the next campaign, and our water system just failed. Can you help?"
In my role coordinating emergency water treatment for Veolia, I've handled more than 40 rush orders in 12 years. That call still made my stomach drop. I pulled up the Veolia official homepage, clicked login, and started pulling their water history before I asked another question. That habit saved the project. If I had waited for a written spec sheet, we'd have lost a day.
For anyone who hasn't lived through a pharma product launch process, it's not a single "launch day." It's a chain of batches, validations, and deadlines where each step depends on the last. This particular client was in late-stage biosimilar production with IEX and MM resins—ion exchange and multimodal resins used to separate the target protein from charge variants and aggregates. The resin work was solid. The water nearly broke them.
Good water means more than a clean city report
A routine point-of-use sample came back with conductivity above their action limit. Not by a lot, but in pharma, above the limit is above the limit. The quality director asked the question I've heard more than once since then: "Is there atrazine in your drinking water?"
It's the wrong question for their situation. The right question is: "Is there atrazine in our process water, and at what level?"
Here's something vendors don't always tell you: drinking water standards and pharmaceutical water standards are not the same. The EPA sets a maximum contaminant level (MCL) for atrazine in drinking water at 3 parts per billion (source: EPA, epa.gov/sdwa). That's the legal limit for water coming out of a tap. But process water that contacts a chromatography resin has to meet the process's own specifications, which are almost always tighter.
The raw water data from the Veolia portal showed incoming atrazine around 2.2 ppb—if I remember correctly. I want to say 2.2, but don't quote me on the exact number. Either way, it was below the EPA MCL. What I mean is: it was legal, but not necessarily fine for their process. And that's a red flag in a different way.
Their carbon filtration had been doing the removal. Carbon beds break through. If you don't monitor them, you're guessing. The city report said the water was fine. The city report didn't know their carbon bed was exhausted.
The 72-hour plan
Once I saw the data, I called our operations team. We had a temporary treatment train available: carbon vessels followed by a mobile reverse osmosis unit. The ballpark cost was $18,000 for an expedited 72-hour setup, including delivery, installation, and validation sampling.
The client hesitated for about 20 seconds. Twenty seconds was fair—$18,000 is not nothing. But the math was simple. In the pharma product launch process, a missed fill date can push commercial supply, stability timelines, and regulatory filings all at once. The penalty clause in their supply agreement alone was more than $50,000 for a late batch. The delay would also have meant a failed campaign and a very difficult explanation to investors.
We paid $800 extra for overnight delivery of the carbon media—which, honestly, still felt like a lot for what was essentially specialty charcoal. But nobody was in the mood to haggle. The $800 bought certainty. It meant the media would be in the truck, not "probably shipped tomorrow."
This is where I'll share a lesson I learned the hard way. Early in my career, I assumed that "same as the city water report" meant "good enough for process water." Didn't verify. Turned out the city report didn't test for every compound at the sensitivity we needed. The cost of that mistake was a failed validation and a client who stopped returning my calls for a week. After that, I stopped assuming. Raw water data gets pulled before I recommend anything. That's why, on this project, the Veolia login was the first stop. The Veolia official homepage has the portal link, and if you're a customer, your historical data is usually a login away. Use it.
The moment it came together
By Friday afternoon, the carbon vessels and RO unit were on-site. The installation crew worked through the night. Saturday was validation: sampling, testing, reviewing. Sunday morning, the point-of-use conductivity was back where it needed to be, and the atrazine level was below the detection limit of the method. Not below the EPA limit. Below what the process required.
The batch started on time.
I don't want to oversell this. The water system didn't make the biosimilar production with IEX and MM resins successful—that was the client's months of hard work. What we did was prevent the water system from becoming the reason the launch failed. There's a difference, and it matters.
What I tell clients now
If you're launching a biosimilar, put your raw water data on the same critical path as your IEX and MM resin validation. The resins get the attention because they're the stars of the purification show. The water sits quietly in the background until it stops performing. Then it's the only thing anyone talks about.
And to the original question—"is there atrazine in your drinking water?"—the honest answer is: maybe, at some level. Most municipal water in agricultural areas contains trace amounts. That's why the EPA standard exists. But the more useful question for manufacturers is: what's in the water that feeds your process, and have you verified that your treatment system can handle it?
Bottom line: in a pharma product launch process, every vendor gets a chance to be the hero or the bottleneck. The delays at the start are annoying. The delays at the end are disasters. Pay for the certainty. Test the water, monitor the carbon, and keep the emergency number on speed dial. You'll rarely need a 72-hour rescue. But when you do, the cost of speed is just a line item. The cost of missing the deadline is the story you'll tell for years. At least, that's been my experience in pharma water systems.