Case Study: Retrofitting Two New High Purity Water Systems

Published on: 
, ,
BioPharm International, BioPharm International-05-01-2017, Volume 30, Issue 5
Pages: 42–46,49

a capital investment project was required to replace the existing, obsolete high purity water (HPW) generation system and water-for-injection (WFI) generation system with new, reliable technology.

GlaxoSmithKline (GSK)’s R&D Biopharmaceutical Pilot Plant in Upper Merion, PA is a R&D clinical trial material (CTM) manufacturing facility with an aggressive processing schedule that requires minimal shut-down interruptions. Utility reliability is paramount to achieve production demands and regulatory quality requirements. To meet the utility demands for increased CTM output, a capital investment project was required to replace the existing, obsolete high purity water (HPW) generation system and water-for-injection (WFI) generation system with new, reliable technology.

The project drivers were:

  • Existing generation systems (HPW and WFI) had insufficient capacity for current operations and were unable to meet the demands of the growing pipeline.

  • Spare parts for the existing systems were either not available or becoming more difficult to source.

  • Existing systems were costly to maintain, not energy efficient, not reliable, and had insufficient redundancy, increasing the potential for unscheduled production downtime.

The primary objective for the project was to provide water system(s) generation reliability with the following additional requirements:

  • Deliver more environmental sustainable systems (i.e., lower water and energy usage)

  • Increase supply and storage capacity

  • Replace obsolete equipment

  • Have no impact on ongoing GMP operations.

GSK engaged Hargrove Life Sciences to complete the design for this project. The conceptual design phase included evaluation of new equipment technologies, a sustainability evaluation including energy and operating cost comparisons, and visiting other recently installed GSK water systems at other locations. Also, because the pilot plant was landlocked, with no available space inside the facility for new equipment, investigation and analysis of where to install the new water systems was required.


Selecting the equipment and technology for the HPW and WFI systems was based on evaluations that took into consideration sustainability goals including reduction of energy, water, and carbon footprint. It was also determined that the best way to achieve system generation reliability was to have complete redundancy for all mechanical equipment (i.e., essentially two of everything).

Redesigning the high-purity water system

The existing HPW distribution system provided a continuous flow of 30 gallons per minute (gpm) of purified water at 25 °C. Purified water was constantly circulated to the utility systems in the basement mechanical room and also to the three GMP operating floors using dual, sanitary variable-frequency drive (VFD) pumps rated at 150 gpm. The basement utilities supplied with HPW feedwater included a WFI still and two clean-steam generators.

This existing system produced water with a resistivity greater than 10 Mohm and total organic carbon (TOC) with less than 5 ppb, and GSK wanted to maintain this high quality for the new system while also incorporating a sustainable design.
Criteria for the new HPW system included the following:

  • Redundant mechanical equipment (i.e., two of everything). Reliability would be achieved through redundancy, with dual multimedia filters, softeners, carbon filters, reverse osmosis (RO)/continuous deionization (CDI) skids, distribution pumps, and vent filters.

  • The existing 34 gpm generation system did not always adequately maintain building operating demands. The new system(s) would require larger capacity (40 gpm) for existing building operating demands, increased WFI generation, and future building expansion or increase in users/processes.

  • Start/stop technology was a requirement to reduce electrical energy and water consumption rates.

  • The generation system must be designed for hot water sanitization at 80 °C (65 °C minimum).

  • A mixed-bed polisher would be required on the new system design to meet resistivity quality requirements.

  • A new stainless-steel, HPW storage tank must be provided that would include increased storage capacity to meet larger instantaneous demands of water from expected increases in production. The new tank would also retain a nitrogen blanket that was installed on the existing storage tank, which had proven to be successful in helping to maintain a low bioburden in the system.

Based on these criteria, it was decided that the RO system only needed to be single-pass technology to meet HPW quality requirements. The project team also decided to implement carbon filtration instead of bisulfite injection or ultraviolet light technology as the primary method to remove residual chlorine/chloramines. Although carbon filtration represents the most expensive initial cost, it is the most effective method of removing residual chlorine/chloramines. The downside of using carbon filtration is the carbon filters are an ideal breeding ground for bacteria. However, because the carbon filters would be hot water sanitizable, the concern of bacteria growth was reduced. Other design considerations included:

  • Installing a new bulk brine storage system to eliminate the need for manual material handling of the salt required for softener regeneration

  • Installing a RO reject water recovery system used for makeup water to the building’s cooling tower

  • Supplying HPW to an adjacent biopharm process development facility, thus enabling the decommissioning of a second water purification system serving this adjacent building, which further reduced site operating expense

  • Installing a new online microbial detection system.

The online microbial detection technology was released for commercial use in pharmaceutical clean utility systems just prior to the design phase of this project. After various on-site pilot testing scenarios, it was concluded that this online microbial detection technology would be installed on the HPW distribution system, greatly reducing the system’s water sampling and analysis work load. The project purchased and installed one of the first commercially available units in the United States. This online microbial detection system complements the online TOC and conductivity systems, which are typical to GSK water system designs.


Redesigning the WFI system

The existing WFI generation system was a 25-year-old multi-effect (ME) still. The unit produced a maximum of 470 gph of WFI at 82 °C. As the facility’s GMP processing manufacturing capacity increased, the WFI generation system had inadequate generation capacity, which resulted in rigid planning for various manufacturing users so that the WFI storage tank would not be completely drained during use. WFI manufacturing capacity studies indicated that the maximum WFI usage in the facility could approach 1325 gallons in a two-hour period. With the existing WFI still make-up rate, the WFI storage tank volume would fall dramatically and cause the WFI distribution system to shut down. It was determined that the existing WFI storage tank would not be replaced because of limited head room and access space into the basement utility area.

With maximum site plant steam pressure limited to 90 psig, replacing the existing still with another ME still would require an oversized (de-rated) system, as these are designed to normally operate with plant steam at 115 psig. In addition, the ME still would require an external cooling exchanger to remove excess heat, which would have an impact on the building’s process glycol system.

With reduced plant steam pressure and the desire to minimize process glycol loads, the engineering team determined that the replacement WFI stills would be vapor compression (VC) type stills. VC stills are designed to operate with 50 psig plant steam in lieu of 115 psig required for ME stills, which yields energy savings from reduced steam usage. Also, the cooling load could be virtually eliminated, which would reduce approximately 25 tons of process glycol that was required for the ME still. Although VC stills are more expensive from a capital installation cost perspective, they are more economical to operate from a utility demand.
The existing WFI distribution system consisted of three independent supply loops with a dedicated pump for each GMP operating floor of the facility. The new pumping distribution system design consisted of two new VFD-controlled redundant pumps each capable of serving all three floors simultaneously. If one pump were to fail, the remaining pump was designed with sufficient capacity to maintain WFI distribution for the entire facility.

Retrofitting the water systems

At the completion of engineering design, the project team determined that the two new water systems must be delivered using a phased construction and validation approach to minimize shutdown interruptions to the manufacturing areas.

Phase 1 included the construction, installation, and validation of a new HPW generation, storage, and distribution system in a location that would not impact the facility’s planned GMP manufacturing schedule. Phase 1A included replacement of the WFI distribution pumps and the main WFI control panel including validation testing. Phase 2 involved the demolition of the existing HPW generation and storage equipment in the building basement to establish the location for the installation of two new VC WFI stills. Phase 2 could only be completed after Phase 1 HPW generation and distribution systems were released for GMP use.

Phase 1
Phase 1 would entail the construction of a strategically placed new mechanical room in an unused courtyard that was isolated between existing facilities, as shown in Figure 1. The new mechanical room in the chosen location was required to be two stories, providing a location to install, commission, and validate the new HPW system prior to connecting to the operating pilot plant and decommissioning the existing HPW system.

Figure 1. Courtyard between two buildings that was chosen as the site of the new, two-story mechanical room. All figures are courtesy of the authors.
The redundancy requirement proved to be a challenge because two HPW generation trains would need to reside in the proposed two-story mechanical room, which was constrained in a space that was only 14 ft. wide and 110 ft. long. A thorough and iterative study was done to confirm the equipment could be installed and be serviceable, which included the development of a three-dimensional design model (see Figure 2). Once the proposed space design was proved out, the detailed design of the water system and proposed addition began. The GSK/Hargrove team worked closely together to design every aspect of the project to ensure minimal impact to the existing adjacent operating facilities.

Figure 2. Three-dimensional design model of two-story mechanical room.

Phase 1 construction of the HPW mechanical room in the unused courtyard required the removal of a portion of Building 38’s exterior glass façade, potentially exposing the pilot plant to the elements of nature (e.g., weather, insects). Therefore, prior to removing any portions of the building’s existing façade, temporary weather-proof interior walls were constructed. As the façade was then being removed, the daily limit of removal was controlled to an area that could be sealed back up in the same day with a temporary facade.

Installation of equipment into the new HPW mechanical room

With the new mechanical room completely surrounded on all sides by existing buildings with no access for bringing in large pieces of equipment, it was necessary to install all the major pieces of equipment using a hydraulic crane. The challenge with doing this is that the room construction took place prior to equipment delivery. Therefore, the project team staged the construction of the new room to essentially have a section of the mezzanine floor and roof to be constructed after equipment installation. Once the missing building sections were constructed, equipment could then be shifted to its final location, as shown in Figure 3.

Figure 3. Installation of equipment into the high-purity water mechanical room.

Phase 1A
Phase 1A included the installation of two new WFI distribution pumps and the new WFI control system. The existing WFI distribution system consisted of three independent loops for each GMP operating floor of the building, each with its own pump. The new distribution system consists of two new VFD-controlled redundant pumps each capable of serving all three floors. This phase was the most critical to ongoing operations because there was no backup plan (i.e., everything had to go right the first time to bring the WFI system back on-line in the shortest possible time frame). Therefore, extensive planning was required including: the development of process operational descriptions for the new control system, which included 86 instruments and 106 input/output points and the development of the WFI system hydraulics using Fathom Modeling for the entire building’s distribution system, which turned out to be extremely valuable for vetting the design. The plan to control the distribution system was to use flow control on the return from each parallel floor loop to maintain minimum velocities. Also, the third-floor return pressure, which was the most remote location in the system, would be used to adjust the speed of the pumps to maintain minimum loop pressures.

Once Phase 1 and Phase 1A were completed, the pilot plant would be supplied with high purity water from the new HPW generation system and supplied with WFI using the existing ME still, new redundant WFI distribution pumps, and new WFI control panel. Phase 2 could then commence, which included the demolition of the existing HPW system in the basement mechanical room and the installation of two new WFI VC Stills.

Phase 2
Phase 2 construction activities were conducted similarly to Phase 1 given that the new WFI stills would be installed, commissioned, and validated while the Pilot Plant was fully operational. Due to the challenge with bringing the new equipment into the basement mechanical room through a constrained access door, the new stills were designed to be disassembled and shipped in multiple components that would then be reassembled onsite.


Utility tie-ins for the different project phases

All supporting utility system tie-ins (e.g., nitrogen, air, clean steam) and the connecting of generation systems to the building’s existing distribution systems were completed during preplanned windows when the pilot plant was not operating and had no water demand.

Leveraged FAT qualification approach

The project team decided early to use a leveraged factory acceptance test (FAT) qualification approach to reduce the overall project validation schedule. All GSK equipment FATs are executed and documented in a manner that may allow GSK to “leverage” the FAT execution documentation (FAT turnover package) during future commissioning and validation on-site at their facility by referring back to FAT-executed approved testing and results. The FAT is a GMP activity approved by all stakeholders including the equipment vendor, GSK Engineering and Validation group, Quality Assurance group, and Facilities Operations group to eliminate the need for redundant testing once the system has arrived on site.
Testing that has been successfully completed at the FAT and poses a low risk of being impacted by transport of the system from the factory to the site in a manner that would change the results of the testing will not be repeated. This testing is detailed and leveraged under a later lifecycle qualification document. Testing that has not been successfully completed within the FAT or that poses a higher risk of impact during transport is either repeated or conducted for the first time at the site under a later lifecycle qualification document. This approach reduced the qualification schedule by three weeks.

The total duration of construction was approximately 60 weeks. The majority of this time (44 weeks) was spent on Phase 1, which was the construction of the new HPW mechanical room and installation and validation of the HPW system. Phase 2, which was the removal of the existing HPW equipment in the building’s basement and the installation and qualification of the new WFI generation systems, took another 16 weeks to construct and validate.

Project challenges and results

The main challenges included the small area to construct the new HPW mechanical rooms and the limited utility shutdown opportunities available to tie-in two new water systems. Another major concern was installing the new WFI distribution pumps with a new WFI control panel because there was no turning back when the old pumps and control panel were removed. The WFI concern was reduced after full programmable-logic-controller simulations were completed along with hydraulic modeling of the WFI distribution system. In fact, the installation and validation of the WFI pumps and controls went exactly as simulated.

The phased approach for construction, validation, and testing provided for the installation of state-of-the-art water generation systems with minimal plant operation downtime and no impact to ongoing production. Since the release of the project for GMP use, benefits were realized quickly by the business, such as ample supply of water, reliable supply of water, lower energy usage, and lower overall operating costs. The GSK Engineering design team worked with selected equipment vendors to minimize water and electrical usage wherever possible, which has resulted in significant environmental sustainability benefits and operational cost savings. Annual operational expenses were reduced by approximately $170k per year, and annual carbon emissions were reduced by 348 tonnes CO2.

The two new redundant HPW generation systems (see Figures 4and 5) represent the latest technology for energy and water efficiency in the production of USP grade water. They are designed using commercially available technology that enables one of the systems to automatically be brought off-line to be sanitized and put back on-line when required. This control technology also consumes significantly less water and energy and produces significantly less waste-water compared to conventional systems. This system translates into operating savings, with more environmentally responsible, energy-efficient, purified water generation processes. In addition, customized software programming requirements were implemented to reduce potable water usage during the softener regeneration cycles, and the RO membrane cleaning is now determined on normalized differential pressure software monitoring in lieu of a traditional totalized flow rate approach.

Major benefits included in the implementation of the new HPW generation systems included the following.

  • Water make-up and sewer savings were approximately 11,000 gallons per day (or 4M gallons per year).

  • Electrical costs were reduced by 88% because the HPW generation system shuts down if there is not a demand for HPW storage tank make-up.

  • The on-line microbial detection system has reduced manual quality grab samples by 20%.

  • The bulk brine tank system reduced site labor costs because the manual salt replenishment process has been eliminated.

  • Additional operational savings were realized by having the new HPW system supply the biopharmaceutical development pilot plant, thus eliminating a costly vendor service ion-exchange contract.

The two, new, redundant VC stills provide additional capacity and ensure reliability. The VC stills were also designed and validated to operate using variable compressor speeds to reduce electric demands when providing WFI to the storage and distribution system. For example, if the required WFI fill rate is minimized, the still compressor can operate on a slower speed to lower electrical operating costs.

Major benefits included in the implementation of the new WFI generation systems include:

  • Electrical costs reduced by 66% (VC does not have a feed water pump)

  • Plant steam consumption reduced by 65%

  • Chilled water consumption reduced by 97% (not required by VC for cooling)

  • Blow down of high purity water to drain reduced by 91% (VC does not blow down in standby mode).

The water systems were installed with minimal planned downtime, and the two water-generation systems were constructed and validated in approximately one year. All project objectives were exceeded.


Figure 4. First floor of the new water treatment room with the new high-purity water generation system (reverse osmosis and continuous deionizer skids).


Figure 5. Second floor of the new water treatment room showing the top of the high purity water storage tank with multimedia, softeners, and carbon filters in the background.

Article DetailsBioPharm International
Vol. 30, No. 5
Pages: 42–46,49

When referring to this article, please cite it as B. Lipko, S. Walter, and B. Termine, " Case Study: Retrofitting Two New High Purity Water Systems," BioPharm International 30 (5) 2017.

About the Authors
Brian Lipko, PE is leader of Projects and Steve Walter,CPIP is Process Technology leader, both with Hargrove Life Sciences,, Tel. 1.215.789.9662; Brian Termine, PE is Maintenance Engineering manager at GSK.