Laboratory water systems: Cost-effective generation and distribution Part 2: System design and operations By Norman H. Toussaint, PE, and Lauren M. Goodfellow This two-part article outlines the challenges and some solutions for designing a robust yet cost-effective lab water system. The first installment, published last month, discussed quality requirements and demand calculation; the second installment, below, reviews specific design choices, equipment, and operational issues. Central vs. local systems After determining required system capacity, the next step is determining the approach to generation and distribution. Two different approaches are available for generating high-purity water: • Installation of a central system supplying to all areas. • Installation of a local system at each point of use. Each approach offers technical and financial advantages and disadvantages. Centralized systems, where installed, are often sized to meet the demands of multiple uses, including non-laboratory uses such as boiler feedwater or closed-loop mechanical system makeup, in addition to laboratory uses. Product capacities can vary from tens to hundreds of gal/min (gpm). A central system is typically most cost-effective if the total facility laboratory water demand is constant, or if there is a continuous need for treated water for other applications. A central system also simplifies operation and maintenance with only one system to manage. Local systems (in-lab point of use systems) are available from a variety of water treatment system vendors in a range of capacities. Although typically sized for one laboratory or one application, they can provide up to 1 gpm product water capacity. Depending on the treatment methods employed, these systems are capable of producing very high-purity water. One advantage of a localized system is a customized quality specification for low-volume demand. For example, if a laboratory facility has only 10% wet labs, a local system costing $5,000 to $8,000/gpm can be a very cost-effective way to meet pure water demand when compared to $25,000 to $40,000/gpm for a centralized system. On the down side, dispersal of equipment can increase maintenance complexity, resulting in additional labor costs. A combination of central and local systems can be a cost-effective solution in cases where the demand for ultra-high-purity water is limited and variable, but when there is a sufficient constant demand for a lower-grade water. The local system can work as a “step-up” unit focusing on certain contaminants and providing quality delivery. Treatment equipment The selection of equipment, whether central or point-of-use, depends on the water quality objectives (e.g., removal of trace minerals or metals, or removal of organic and microbial materials), and to a lesser extent, on the quality of the incoming water supply. Treatment systems are conventionally broken down in two groups: pre-treatment (or makeup) and polishing systems. Pre-treatment systems are designed to remove the bulk of suspended, dissolved, and organic contaminants, providing feedwater suitable for makeup to a variety of facility systems, scientific equipment cooling, or feed to additional pre-treatments or polishing systems. Without pre-treatment, the equipment used to produce high-purity water could see significant contaminant loads, resulting in lower efficiency and higher operating costs. Pre-treatment systems typically include one or more of the following steps: • Filtration. • Activated carbon. • Reverse osmosis. • Continuous ion exchange (electrodeionization). Depending on inlet water quality and the performance of RO or EDI equipment selected, product water from pre-treatment systems can meet ASTM Type II “Reagent Grade” water standards without further processing. Fig. 1. Typical makeup/polishing system components. All figures: HDR Architecture.Click to enlarge. Polishing systems are designed to further process the pre-treated water to meet the final water quality objectives. Depending on these objectives, the technology can include: • Distillation. • Mixed bed ion exchange. • Ultra- or nano-filtration. • UV oxidation. • Sub-micron filtration. An example of a makeup and polishing system is illustrated in Fig. 1 (above, left). It is important to recognize that achieving one water treatment objective may have negative impacts on another. For instance, ion exchange resins selected to remove dissolved minerals and achieve high resistivity values can contribute to elevated levels of organics and bacteria. Treatment system components must be sequenced to progressively reduce undesirable contaminants without introducing new ones into the product stream. Typical sizing criteria for treatment equipment (such as RO systems and deionized water units) are well-established, and are beyond the scope of this discussion. It is worth repeating that pre-treatment or makeup systems may supply building uses other than laboratory water, e.g., makeup to closed-loop systems or humidification systems, and these demands need to be included in any preliminary estimates. System sizes (and the supporting infrastructure) also need to consider adequate capacity for backwash and rinsing operations, chemical dilution, and so on. Local or in-lab systems typically combine pre-treatment and polishing steps in one or more treatment modules. They are available as on-demand units or with storage capability. These systems can be fed from the building potable water supply or from a pre-treated source (central water softener or pre-treatment system); pre-treating the local treatment system supply usually results in a longer service life and can also improve treatment system performance. While most, if not all, water quality objectives can be accomplished using local treatment units, the fact that the system can be idle when there is no demand may lead to problems with bacteriological contamination; this needs to be carefully reviewed with the users. Storage and distribution of laboratory water The primary considerations in design of laboratory water distribution systems are straightforward: The system must meet the average and peak demands from all users while delivering water that meets the required quality criteria. From an engineering perspective, this means selecting and sizing piping systems to maintain velocity criteria and avoid stagnant flows and dead legs, while providing sufficient flexibility to allow for laboratory renovations or expansions. In a complex laboratory facility, however, this is not always a simple exercise. Most laboratory water systems, even point-of-use systems, include provision for storage of product water. Providing storage allows makeup systems to be right-sized to take advantage of water production during off-peak hours. For central plant storage, the capacity should be carefully determined so that all demands can be satisfied without providing unnecessary volume, which increases vessel cost and required space. It should also be noted that storage of high-purity water requires particular attention to storage vessel construction materials and venting or nitrogen blanketing provisions, to minimize the re-contamination of the water due to infiltration or contamination imparted from the interior vessel surface. Fig. 2. Distribution loop configurations.Click to enlarge. There are a number of ways to route laboratory water system piping from the point of generation to the point of use. To maintain water quality, the distribution loop should be as short as possible and sized to maintain proper flow conditions at normal and peak demands (typically, 5- to 9-ft/sec in supply piping with minimum 3 ft. per second in the return flow). This is also the most cost-effective approach. The ultimate configuration depends on the layout of laboratories in the building, including the stack configuration in the case of multi-floor facilities. If water quality is critical to laboratory operation, the water distribution piping concepts should be provided during initial building planning (along with other critical services decisions like air management) to ensure that sufficient space and access is provided. Several of the most common loop configurations are shown in Fig. 2 (left). The number and location of service connections should be carefully planned with the user, and should correlate to the demand and diversity estimates. Aside from adding to the overall system cost, each outlet introduces a flow disturbance where contamination such as microbial growth can occur. This can be a particular concern when connecting to fixtures provided with laboratory furniture such as fume hoods, unless the fixtures are designed for high-purity water applications (e.g., recirculating tap fittings). Selection of piping system materials can also have a significant impact on total cost, although the primary consideration should be maintenance of water quality. Plastics such as polyvinyl chloride (PVC) and acrylonitrile butadiene styrene (ABS) are appropriate for makeup systems but can contribute significant contaminants to high-purity water. Unpigmented polymer materials such as polypropylene and polyvinylidine difluoride (PVDF), particularly when joined by heat fusion welding with minimal internal beads or crevices at the joints, are appropriate for ultra-high-purity water distribution. Fig.3. Relative material costs.Click to enlarge. For high-purity applications, the selection of valves and fittings (for example, diaphragm valves with zero dead leg) is as important as piping material specification and piping system design. These high- purity valves and fittings can be significantly more expensive than comparable components of PVC or ABS. An example of relative piping system costs in various materials of construction is provided in Fig. 3 (above). Cost considerations Equipment for production of high-purity water can greatly vary in cost (in the range of $5,000 to $40,000/gpm of product flow). This is one reason why careful analysis by the design team is needed to properly match the technology and capacity to the need. With a centralized distribution system, a significant portion of system cost can be spent on the treatment equipment; therefore we recommend considering alternate delivery strategies early in the project. Alternatives include: • Point-of-use treatment units. • “Modular” design of central plant systems. • Build/own/operate arrangements. It should be noted that these alternative approaches do not minimize the importance in proper system planning, selection and specification as described in this article. Designing a central treatment plant with modular treatment units can be a useful strategy to minimize initial cost and deal with uncertainties in capacity growth and/or increasingly stringent water quality criteria. If the capacity is sized for short-term requirements, the distribution system must still consider the full build-out capacity and the maximum and minimum flow rates at initial and future conditions. A common trend in scientific facilities is “shelled” laboratory space. When construction dollars are at a premium or when detailed laboratory requirements cannot be determined in time to support the construction schedule, the project team may decide to “shell out” portions of a new facility for future fit-out, and offset the initial costs. The laboratory water system designer must be careful to specify distribution piping headers for the maximum anticipated build-out density, leaving branches or laterals until later; otherwise the optimum flow and velocity criteria may be unachievable. One does not always save full ft2 costs when deferring lab water systems to a future build-out phase. Operational issues Consideration of performance, efficiency, and cost does not end at completion of laboratory water system installation. Throughout the life of laboratory water systems, numerous decisions and actions that must be taken, and the two major cost drivers are materials and labor. Troubleshooting, maintenance, and repair efforts can be reduced if proper attention is paid during planning and design. Installation of flow metering and continuous quality monitoring instrumentation (for example, at the point of supply from the polishing system) adds to initial system capital cost but can ultimately reduce operating cost by providing information to allow efficient system balancing and early indication of preventative or corrective maintenance. Placement of monitoring instruments after regenerable or replaceable treatment components like ion exchangers can provide the operator with a more accurate indication of capacity breakthrough than predictive methods (time or cumulative flow), resulting in dynamic treatment and reductions in chemical cost. Planning for easy access to all system components can reduce maintenance time and expense; the impact on budget is a one-time additional area allocated to the laboratory water system, in facilities where providing usable research space is the primary objective. This is another feature that can be provided at reasonable cost if carefully planned. It is often possible to share maintenance space with other uses. Similarly, providing isolation valves (or even piping headers) for future use can be seen as extravagant until the costs of a future retrofit are taken into account. In summary, the design of laboratory water systems involves accurately establishing the quality and sizing criteria to meet specific laboratory needs, developing schemes to build in flexibility (for example, to accommodate future changes in laboratory use), selecting materials and fixtures, and specifying system components. System design is made more challenging by the various laboratory water standards and treatment options available. Only when the user’s real quality criteria and mode of operation are understood can a cost-effective solution be realized. Norman H. Toussaint, PE, LEED AP, is a senior chemical engineer in the London, U.K., office of HDR Architecture Inc. Lauren M. Goodfellow is a laboratory planner in the firm’s Mountain View, Calif., office. HDR offers a broad array of architectural, engineering, consulting, and management services for various sectors, with science and technology as a core market group.