Unit Operations

Putting WFI to Work in Juncos

Amgen process engineers share Water-for-Injection expertise from the ever-expanding facilities in Juncos.

By Paul Thomas, Managing Editor

Amgen is bullish on Puerto Rico. It has added bulk manufacturing to its facilities in Juncos — the company’s only commercial fill-and-finish site — three times since 2000 to expand operations of its most successful products, including Epogen, Neupogen and Enbrel.

Expanding facilities require expanding water capabilities. The most recent $400 million, 500,000-square-foot expansion included the design and implementation of a new Water-for-Injection (WFI) system, with USP-grade water processed by a high-capacity multiple-effect still. The company is in the process of validating the new lines and ramping up production.

Amgen lead process engineer Pete Jarvis, and Tony Cope, a process engineer and utility systems expert with Fluor Corp., headed up the WFI project. Jarvis has worked with Amgen in Puerto Rico for four years, and for six years previously in Boulder, Colo. The two worked with equipment provider Paul Mueller Co. (Springfield, Mo.) to specify still capacities and options, then designed the distribution system and assisted in start-up. Design included pumping systems including back-pressure and pump controls, heat exchangers, subloops and extensive troubleshooting.

In this article, Jarvis and Cope share lessons learned in Juncos and WFI best practices based upon their years of expertise.

WFI for processing

Amgen in Juncos uses WFI for most of its process or product-contact water, Jarvis and Cope say, including cell culture media preparation, “sink drops,” glass washer final rinses, and even the final rinse of equipment cleaning. This requires a WFI output of thousands of liters per hour. Lower-grade RO (Reverse Osmosis) water is used for the early stages of Clean-in-Place.

“I have worked for other plants, such as Amgen in Colorado or the Neupogen facility in Juncos, where we didn’t use WFI for all of the process water,” Jarvis says. “For the upstream operations, we just used RO water — to make media for a fermenter, for example.” Amgen decided to take the conservative approach and use WFI for all cell culture processes in Juncos to ensure that water purity would not be an issue and hamper start-up of expanded operations.

A new wrinkle tested at Juncos was to collect the WFI of the final CIP rinse and reuse it as the first prerinse of the subsequent CIP cycle for different equipment. This is a key water conservation step, Jarvis says, and is not altogether uncommon in the industry. “If it’s not already, this is starting to become standard practice in the industry,” Jarvis says. As long as the step is validated, “It makes sense if you can recover a rather large volume of water and reuse it,” adds Cope.

Because of the conservative approach, WFI validation proceeded smoothly, though there was one hitch, Cope recalls. “We started validation while we were at the tail end of construction, so it made it difficult to get a passing sample,” Cope says. “We would get debris in the samples, or TOC [total organic carbon] presumably from paint fumes or cleaning fluids, and it delayed validation a bit. In one area of the plant, we started gowning and cleaning sooner than planned so that we wouldn’t contaminate our WFI samples.”

“I’ve been on a lot of projects where we’ve had to sample WFI during construction, and we’ve had to make sure we take the right measures,” Jarvis adds. “You have to think it through carefully. You have to figure out how you’re going to get a good sample — how to protect them and how to schedule them.”

Still life

Amgen chose the Paul Mueller multiple-effect still after a competitive bidding process. Multiple-effect and vapor recompression are Amgen’s two approved methods for water purification. Amgen’s new standard is vapor recompression, Jarvis says, since the systems can process greater amounts of water for a lower operating cost. Vapor recompression requires much less plant steam than multiple-effect and no cooling water if it is stored hot, though it also uses more electricity and can lose its cost-saving advantage if local electricity rates are high. For larger capacity requirements, vapor recompression units also require less floor space than WFI stills.

As biopharmaceutical processes and WFI needs expand, vapor recompression becomes more of the industry standard as well. But Amgen evaluates both technologies for each project. In Puerto Rico, multiple-effect WFI made sense. Amgen’s team discovered that the island’s water has large amounts of silica, which would require extensive pretreatment for a vapor recompression system. “Recompression becomes a lot more economical if you can forego some of the pretreatment and use a lower-quality water,” Cope explains, “but the silica level here is so high that it will ‘scale’ the unit, so you have to send it through an RO anyway.”

A six-effect still similar to that used by Amgen in Juncos. Photo: Paul Mueller Co.



Multiple-effect stills have other advantages. They have few moving parts and thus good mechanical reliability. It is also possible to get back-up clean steam drawn from the still’s first effect, but the Junocs facilities have not taken advantage of this option. Vapor recompression systems, on the other hand, have high-horsepower motors, moving parts, plenty of bearings and seals, and of course noise.

“Another thing that comes into play is the preference of the maintenance people on site,” notes Jarvis. “We came to a site here in Puerto Rico that only had multiple-effect. It made sense to align a system with the existing ones on site.” As Juncos expands, so must WFI capacity, and Jarvis believes that cost, not user preference, will become the driving factor in selecting future WFI systems.

Automation and control

Amgen’s WFI unit is completely automated. “There’s no routine operator interaction,” says Cope. “We send a signal to turn on and start producing WFI, and tell it to turn off.” Mueller MES run on Allen-Bradley programmable-logic controllers (PLCs), and include an operator-interface terminal (OIT) for monitoring and configuration of set points, delay times, and PID (for “Proportional-Integral-Derivative”) loop parameters.

The WFI control is integrated with the process control system — Rockwell’s RS Batch — which is connected to a plant information system for data archiving. Integrating the WFI control with the batch system was fairly standard, Cope notes. “The biggest problem I’ve noticed between our system and any vendor PLC system was the ControlNet communication between the two,” says Cope. “We had some problems getting it set up and configured between the various skids. It was a little rocky at first, but once we figured it out, it went smoothly.

Design concerns

Both Jarvis and Cope were instrumental in designing WFI distribution at the site. The greatest challenge, Cope says, was accounting for the size of the facility and the numerous WFI applications throughout. “We have so many users spread over such a wide area that we ended up with some pretty long loops,” he says. “This presents a problem in maintaining a constant pressure.”

With long loops, it is best to keep flow rates low in order to keep loop pressure low, but this can lead to control issues. “Overall flow rates and back-pressures tend to fluctuate widely and make it difficult to control constant pressure,” says Cope. The solution, in one situation, was to make the loop with a larger diameter so that more fluid could be pumped to keep pressures constant.

In another instance, a control valve that could handle both high and low flow rates was placed on the end of the loop. This has been a “mixed success,” Cope notes, because the valve controls can not be tuned to easily accommodate the differing needs of the many WFI users. “The control valve doesn’t necessarily like to be tuned to handle many different scenarios because they all require it to respond differently,” he adds. “Essentially, you can tune the valve for one or two scenarios and see if it can respond reasonably to the dozens of other situations that arise.”

The results are “tolerable” since pressure requirements are not strict, Cope says, but the company will search for better solutions as it continues to expand. “I’ve only seen loops handled in a few different ways — with a back-pressure control valve, a spray-ball, or with a VFD [variable-frequency drive],” says Cope. “If you have no control valve at the end of the loop, you can use your VFD to maintain a constant pressure, but it increases the flow rate more than with just using a control valve.”

Science and sampling

Some would say that WFI is overused in most facilities, and that in many instances high-quality purified water would suffice. Jarvis agrees. “I’ve worked in successful plants where we’ve used high-quality purified water for upstream process steps,” he says. “It wasn’t until we got into purification, where we had unit operations that were designed to remove endotoxin, that we wanted to make sure we had low-endotoxin water downstream. Even then, the highly-purified water that we used upstream met WFI requirements for endotoxins.”

So why do it, then? Better safe than sorry, Jarvis says. “We have to be conservative.” As the WFI implementation in Juncos becomes more established, Jarvis and Cope plan to put more science behind their water choices.

The Juncos team samples its WFI water daily for conductivity, TOC, colony forming units and endotoxins. Most of this is done by analyzing samples in the laboratory. It also measures conductivity and temperature on-line using standard set points. In some Amgen plants, Jarvis notes, on-line TOC measurements are taken as well to back up lab data.

The entire science of WFI distribution needs upgrading, Jarvis and Cope believe. The subject of rouge discoloration on internal surfaces is a case in point. WFI systems should be re-passivated and checked for rouge every few years, and Juncos has a regular passivation regimen. “But it’s really not science-based,” Jarvis says. “We need to get better knowledge of whether or not rouge is a bad thing. We tend to think, ‘We’ve got rouge, we’ve got to get rid of it.’ But it’s possible that you could be doing more harm than good by doing this. You could be tearing your system up, so we’ve got to get some more science behind it.”

Free Subscriptions

Pharma Manufacturing Digital Edition

Access the entire print issue on-line and be notified each month via e-mail when your new issue is ready for you. Subscribe Today.

pharmamanufacturing.com E-Newsletters

A mix of feature articles and current new stories that are critical to staying up-to-date on the industry, delivered to your inbox. Choose from an assortment of different topics and frequencies. Subscribe Today.