In 1995, the Colorado Department of Corrections embarked on a pioneer program for rehabilitating teenage criminals, called the Youth Offender System (YOS). Instead of simple incarceration, selected teenagers are put through a rigorous, "tough love"-type training regimen, which has been likened to Marine boot camp.

The site chosen for this program was an unused portion of the Colorado Mental Health Institute (Pueblo, CO), an extensive, campus-type mental hospital facility. Although the state hospital is still active, many dormitories and related facilities have been shut down for almost 10 years.

RNL Design, an architectural-engineering firm based in Denver, was chosen to design the renovation project, which consists of upgrading four dormitories and four adjunct buildings serving administration, food service, visitation, and related services. Also on the docket is construction of a new health clinic, high-security housing, and a gymnasium.

RNL Design's Engineering Department undertook the task of designing the upgrade of the campus heating-cooling systems. The project is to be carried out in at least two construction phases. Construction for Phase I is currently nearing completion.

Design Restrictions

When the engineering team started its preliminary analysis, it soon became apparent the design would be subject to some tight constraints, including:

  • Keeping within a rigid, low-cost budget imposed by the state, with little money available for any "extravagant" systems;
  • Keeping the areas requiring access for maintenance to a minimum, due to the nature of maintenance operations within a prison environment;
  • Accommodating the owner's desire to keep operating costs (i.e., energy and equipment maintenance costs) to a minimum;
  • Maintaining the aesthetic qualities of the existing architecture as much as possible; and
  • Dealing with tight space restrictions in nearly all of the existing buildings, which had been constructed before ducted air systems were commonplace; furthermore, interior space for new mechanical equipment, such as chillers or air-handling units, would be practically nil because the YOS program needed to accommodate as many inmates as possible.

With all this in mind, RNL engineers evaluated several candidate systems, including:

  • Decentralized, packaged rooftop air-handling units on each building with evaporative cooling, an option often used in Colorado's arid climate;
  • Packaged rooftop units with DX cooling instead of evaporative cooling; and
  • A campus-wide system using a chilled-water distribution loop from building to building for the cooling side, with centralized chiller plants using evaporative cooling doing the heat rejection (i.e., cooling towers).

    The evaporative cooling option, although appropriate for maintaining low energy costs, would require ductwork too large for the space available, in addition to having a high initial cost. The DX option, although the least expensive from a first-cost perspective, presented energy cost, noise, maintenance, and aesthetic issues. The chilled-water loop could work within the constraints' parameters, but only if the installed costs could be minimized.

    RNL's engineering team came up with a "novel" chilled-water distribution design that is believed to be the first of its kind in Colorado: a single-pipe, primary-secondary chilled-water system.

    A What?

    Although single-pipe, primary-secondary (also known as hydraulically decoupled) systems are common for heating system design, they are rare in chilled-water applications, although some have been applied successfully in other areas of the United States.

    The more common primary-secondary system is a two-pipe system with separate supply and return lines, in which the separate supply piping can maintain a more constant chilled-water supply temperature. With a single-pipe system, every time a branch loop reconnects with the main, the return water mixes with the supply water and raises its temperature. Figure 1 shows the difference between these two kinds of primary-secondary systems.

    If a single-pipe system consists of a small number of proportionately large loads, then at each secondary takeoff, the temperature in the primary will rise abruptly to a temperature that is too warm for the following secondary-loop cooling systems.

    If, on the other hand, the primary system is feeding a large number of relatively small secondary loads, the incremental temperature rise at each succeeding set of takeoffs will be small, and the coils at the later legs of the system can be sized for higher entering water temperatures with smaller temperature rises.

    By the time the water temperature has risen to a maximum temperature that is still acceptable for that last air-handling unit, either the primary piping system should be at the last takeoff and headed back to the main chiller plant, or an intermediate chiller should be inserted at that point to lower the water temperature for the next set of secondary loops.

    Application To YOS

    For the YOS project, the design team determined that two intermediate chillers spaced out along the primary loop would be the optimum solution, rather than using more chillers, or even just one large central chiller plant.

    One of the advantages of this design for the YOS application was that the design allowed for adding chillers and extending the primary loop to more distant parts of the campus in later expansions. Figure 2 shows a site plan of the campus with the primary chilled-water loop in the center, and the chillers at roughly opposite sides of the loop.

    Another advantage of this design approach was that by splitting the overall temperature rise into two separate portions, there was no need for an unusually low temperature at the start of the chilled-water circuit to accommodate the last secondary takeoff. It was determined that supply water temperature of about 40 degrees F leaving either of the chillers would meet design criteria. This is within the parameters of normal chiller operation using standard chillers.

    Figure 2 also shows where the loop will most likely be extended in future construction phases. Although most of the calculated loads are fairly small (on the order of 40 tons out of a roughly 500-ton loop), resulting in fairly low temperature rises over the stretch of the primary loop, some of the loads are considerably larger. (See Table 1.)

    The gallon per minute (gpm) flow in the primary piping that would be needed to satisfy all the secondary loops would differ, depending on whether the water was flowing clockwise or counterclockwise through the primary loop. It was important to locate the chillers so that the largest loads were at the ends of the runs, so that the largest temperature rises would occur at a point where the chillers could bring the temperatures down again.

    Pumping Considerations

    One of the big advantages of primary-secondary systems is the simplified pumping design. Because of the low pressure drop within the portion of piping that is common to both loops, the primary and secondary loops are hydraulically decoupled; that is, the operation of the primary pumps have practically no effect on the operations of the secondary pumps.

    The central distribution pump was sized at 1,200 gpm with only 60 ft of head pressure for a 3,000-ft-long piping loop, requiring a 40-hp pump. A standard system would have required nearly twice as much hp. However, for the primary-secondary system, the only purpose of the main distribution pump is to circulate the water around the main loop. It does not have to push the water through the building loops with their air-handling unit coils, control valves, etc. The primary-secondary system's main distribution pump only "sees" the pressure drop associated with piping friction over 3,000 ft.

    In this case, the firm increased the pump motor size to handle additional piping friction for future construction phases, in which the main loop would be extended. It was not necessary, however, to increase the pump gpm capacity for future phases, as would have been the case if the firm had used a standard supply-return design.

    Piping Considerations

    PVC plastic piping was chosen for several reasons:

    • Its low friction factor;
    • The fact that no internal scaling problems would occur over time; and
    • The inherently lower heat transmission through the piping walls.

    Two major issues arise when deciding whether or not to use plastic piping in general: its strength, and its high rate of thermal expansion. The issue of strength is important with larger pipe sizes because the momentum of the water at larger mass flows can create higher stresses at elbows and other fittings. Plastic cannot take the kinds of stresses that steel can.

    The design firm minimized those stresses by keeping the water velocity at 6 ft/sec. This, incidentally, kept the friction factor at just over 1 ft/100 ft (using a 10-in. pipe), thereby keeping pumping power requirements low.

    The issue of thermal expansion is also a major consideration, since plastic piping expands much more than steel. Over a length of 3,000 ft, the expansion is on the order of several inches and can seriously stress branch connections.

    The design firm opted for piping with premanufactured sleeves and internal neoprene gaskets. The firm specified that each 10-ft section of piping would be connected to the next section, with at least 1/4 in. to spare before butting tightly to the next pipe within the sleeve. With the pipe buried 6 ft underground, and carrying chilled water with its relatively low delta T, the resulting expansion that arises for each 10-ft section is on the order of magnitude of 0.01 in., which is easily accommodated by the neoprene gaskets.

    Other design issues concerning piping included heat gain from the soil in the main underground distribution loop, and how the secondary building piping loops would connect to the primary distribution loop.

    Regarding heat gain, the "District Cooling" chapter in the 1992 ASHRAE Systems and Equipment Handbook outlines the computational procedure for this application, using an example similar to the YOS project. The design firm found that even with bare pipe buried 6 ft underground, the heat gain from the soil would be on the order of 3%. It turned out, however, that the additional cost of using manufactured insulated pipe was small, so the firm opted for that.

    As for the secondary connections, Figure 3 shows a typical detail that illustrates the firm's approach. In example A, the design intent was to avoid any short-circuiting of the chilled water back into the chiller loop instead of into the main distribution loop, where it belongs. Classical primary-secondary design methods advocate keeping branch tees as close together as possible (within 12 in. or so), to minimize pressure drop in the common piping, thereby providing hydraulic decoupling.

    The firm's concern was that keeping the tees this close might increase the possibility of "ghost flows," as illustrated in Figure 4, due to the impact of the flow from the branch driving the water into both directions at the main. The firm, therefore, separated the branch tees by about 10 pipe diameters, so that the momentum of any ghost flows would be overcome by the primary flow.

    Example B in Figure 3 shows how the firm connected the secondary branches when two buildings were close together. If the return loop in Building 1 was just upstream of Building 2, then Building 2 would have to use water at a higher temperature. By placing both supply and takeoffs upstream of the returns, both buildings get the same water temperature.

    Chiller Considerations

    Although the single-pipe, primary-secondary system offered so many design benefits for this application, it also imposed some design constraints and energy penalties with respect to the chillers.

    First of all, the primary piping loop could only support a 6 degree delta T at design (Table 1), compared to an 8 degree to 12 degree delta T in more standard systems. Referring back to the earlier discussion on the large number of small loads on the primary piping loop, if the last load before the chiller had been a larger one, the chiller would have seen a more normal delta T. However, the layout of buildings on the campus, as well as restrictions on where the firm could locate the chillers, left the firm with no other option.

    This delta T dictated a 4-gpm/ton flow through the evaporator of the larger chiller (1,200 gpm total flow through both the primary and the 300-ton chiller). With the smaller, 200-ton chiller, the firm had two options:

    • Pushing 6 gpm/ton through the evaporator for a flow equal to the primary side, with a resulting 4 degree delta T; or
    • Pushing only, say, 4 gpm/ton with a higher delta T (39 degree F chilled water leaving the chiller), but the same resulting 41 degree F mixed-water temperature on the primary.

    With the second option, the decrease in chiller efficiency just about canceled out the pump energy savings. The firm eventually decided on the 800-gpm option, so as to reduce the required piping size to the chiller.

    Control Considerations

    Determining how to control the system turned out to be remarkably easy, once the firm took a step back and examined the requirements.

    As long as the chilled-water loop starts with at least 42 derees F on a design day, then the capacity is present to satisfy the furthest load before the next chiller. No sophisticated control algorithms are necessary, and neither chiller needs to know what the other is doing. The only requirement is that a sensor downstream of each chiller keeps the water in the primary loop at 42 degrees F maximum.

    However, since the firm was already using digital direct controls (ddc), it was easy to add primary-loop temperature reset based on the greatest differential in the buildings, to keep the primary loop flow constant.

    In the end, the YOS project demonstrates that given the right circumstances, this application can offer distinct advantages for campus distribution systems with respect to piping system cost, pump power requirements, and simnplicity of controls.ES

    EDITOR'S NOTE: The images associated with this article do not translate to this website.