1、OR-05-1 2-4 Reliability Engineering for Datacom Cooling Systems Donald L. Beaty, PE Member ASHRAE ABSTRACT A metric that is paramount in todays datacom (data processing and telecommunications) facilities is reliability. It is a requirement that is often quant$ed by the facility decision makers. As l
2、oad densities increase, the challenges to plan, implement, and operate reliable cooling systems to satis there is little focus on investing in designs that minimize human error. QUALITATIVE VARIABLES THAT IMPACT COOLING SYSTEM RELIABILITY Overview and Baseline To illustrate the extent of the impact
3、that qualitative vari- ables have on the reliability of a cooling system, let us consider a project design needing CRAC units to handle a 1 O-ton sensi- ble load. In the subsections below, we will look at groupings of scenarios derived from changing the emphasis of a single variable and how that cha
4、nge impacts the overall cooling system reliability. The groupings will focus on the following variables: Equipment location Commissioning process Implementation of redundant units Maintenance and operation The focus behind the description of the scenarios in these groupings is to demonstrate how imp
5、roving the overall cool- ing system reliability is not a simple challenge that is met by implementing a single solution but, rather, that it requires an engineered approach including a holistic consideration of a wide variety of variables and an implementation strategy to ensure that each one is pro
6、visioned with the common goal in mind. Group 1: Impact of Equipment Location in this first group ofscenarios (Group i), we will compare the impact that the location of the equipment being used may have on the overall cooling system reliability. The first scenario under this group will also be used a
7、s a baseline for ali subsequent groups. Scenario la (Baseline). A new CRAC unit, optimally located and commissioned, and currently has been in operation for 80 hours under full load (i.e., broken in). Scenario lb. Same as Scenario la but the unit is installed in an undesirable location, resulting in
8、 less than recommended clearances and very difficultkomplex piping and wiring configurations (Figure 3). Scenario IC. Same as Scenario la but with physical obstructions and other producers of turbulence to air- flow paths on the supply and/or return airflow side (Fig- ure 4). Figure 3 Vertical stanc
9、hions blocking front access to CRAC unit. ASHRAE Transactions: Symposia 949 Figure 4 Piping blocking supply airflow path. Scenarios la through IC are simplistic but are helpful to demonstrate the variation in reliability and performance that can commonly occur independently from a focus solely on “s
10、ingle point of failure.” As indicated by the scenarios, the location of the equipment from both a connectivity, clearance, and airflow path standpoint. all can have a significant impact on the reliability of the cooling system as a whole. Group 2: Impact of Commissioning Process Commissioning is oft
11、en thought of as simply “extensive start-up and testing,” but true commissioning is far broader and more important than simply testing. ASHRAE defines commissioning as: The process of ensuring that systems are designed, installed, functionally tested, and capable of being oper- ated and maintained t
12、o perform in conformity with the design intent. This process could easily be renamed as the “critical process” or “roadmap to successful reliability.” ASHRAE provides a good source for information on the process of commissioning, including ASHRAE Guideline 1- 1996, The HVAC Commissioning Process. In
13、 this guideline, commissioning begins with the initial planning phase and continues on through the design, construction, start-up, accep- tance, and training. In other words, it should be applied throughout the life of the building. The following is another group of scenarios (Group 2) where Scenari
14、o 2a is exactly the same as Scenario la in the previous group (Group i). However, the variable that impacts the reliability in this group has to do with the level of the commissioning process that is performed. Scenario 2a (Baseline). A new CRAC unit, optimally located and commissioned and currently
15、 has been in operation for 80 hours under full load (i.e., broken in). Scenario 2b. Same as Scenario 2a including a complete commissioning process but the unit started serving criti- cal load without the benefit of any full-load hours of break in or bum in (therefore exposed to the risk of infant mo
16、rtality as described by Galm 2003). Scenario 2c. Same as Scenario 2a but the commission- ing process did not include a load bank to apply a full cooling load for the prefunctional and functional testing that occurs with the ASHRAE-defined commissioning process. Scenario 2d. Same as Scenario 2a but n
17、o commission- ing was performed; instead, just the traditional factory start-up procedures and standard punchlist were used. Scenarios 2a through 2d are also simplistic but are helpful to demonstrate the variation in reliability and performance that can commonly occur simply by varying the commissio
18、ning effort from comprehensive to none. Group 3: Impact of Implementation of Redundant Units The following is another group of scenarios (Group 3) where Scenario 3a is exactly the same as Scenario la in Group 1 but now is used as a baseline to compare a still different set of variations that impact
19、reliability. This time, the variations have to do with the addition of redundant units and why that method does not automatically equate to an improvement in the overall reliability of the system. Scenario 3a (Baseline). a new CRAC unit, optimally located and commissioned, and currently has been in
20、operation for 80 hours under full load (i.e., broken in). Scenario 3b. Two new CRAC units each equal in size to the CRAC unit in Scenario 3a but the locations of the units are such that a failure of one unit will not provide adequate capacity to the intended load. Scenario 3c. Two new CRAC units, ea
21、ch equal in size to the CRAC unit in Scenario 3a, but the units are nut calibrated to work cohesively and do not have sufficient controls or mechanisms to effectively operate as a reli- able, redundant system. Scenario 3d. Two new CRAC units, each equal in size to the CRAC unit in Scenario 3a, but n
22、o commissioning was performed and, instead, just the traditional factory start-up procedures and standard punchlist were used. Scenarios 3a through 3d provide some sense of the vari- ation in reliability even if redundant units are added. Redun- dant units are often thought of as the simple solution
23、 to increasing the reliability of a system, but, as indicated in the scenarios above, the decision to utilize a redundant system still requires an implementation process focused on meeting the overall goal of improved reliability. 950 ASHRAE Transactions: Symposia Group 4: Impact of Maintenance and
24、Operation In the final group of scenarios (Group 4), once again Scenario 4a is exactly the same as Scenario 1 a in Group 1 but now is used as a baseline to compare how maintenance and operation variations can impact system reliability. Scenario 4a (Baseline). A new CRAC unit, optimally located and c
25、ommissioned, and currently has been in operation for 80 hours under full load e., broken in). Further, it receives award-winning preventative mainte- nance by superbly trained operators and technicians. Scenario 4b. Same as Scenario 4a but maintenance is only performed reactively on an as-needed bas
26、is or with a minimal focus. Scenario 4c. Same as Scenario 4a but preventative maintenance is performed by below-average technicians or by average technicians who are pushed to restrict time and cost to where their performance is impacted. Scenario 4d. Same as Scenario 4a including the pre- mium prev
27、entative maintenance schedule, but operators are not technical and/or have inadequate training on how to reliably operate the cooling system (e.g., do not understand the impact of a raised floor tile being tempo- rarily removed or the actions to take if a unit goes off- line for service or fails). S
28、cenario 4e. Same as Scenario 4a but one or more shifts have operators that are not capable of effectively operat- ing the cooling system when a failure occurs (e.g., turn other units on, change valve or damper positions, etc.). Scenarios 4a through 4e provide some sense of the vari- ation in reliabi
29、lity based on operations and maintenance vari- ations. Summary of Scenarios The purpose of describing Scenario Groups 1 through 4 was not to pick nor debate which scenario is best or which of the variable groupings require the most consideration but to highlight that one possible logical choice from
30、 a reliability perspective could be that of the baseline scenario. Certainly there is room for much debate and also the opportunity to develop a more reliable configuration than the baseline scenario described above, but it supports the point that there are no easy answers and, further, that redunda
31、ncy or avoiding “single points of failure” is only one issue in the road- map to reliable cooling systems. The scenarios support that an equal focus should be on the pre-design, design, commissioning, implementation, and operation phases, and each phase must include ASHRAE Transactions: Symposia cle
32、arly defined and agreed upon objectives, metrics that can be used to measure performance, a check-and-balance system to validate results. RELIABILITY ENGINEERING INFRASTRUCTURE The reliability engineering approach includes provisions for the reduction of risk at an infrastructure level. Reducing ris
33、k sounds so simple-just focus on systems or facilities that are easy to design, bid, implement, commission, operate, and change. However, just like the adage “sorry for the long letter, no time to write a short one,” making things easy or simple requires considerable effort. Some influences on the c
34、omplex- ity involved include tight space constraints, lack of rnodularitylorganization, excessive interdependencieslsequences. use of nonstandard pieces of equipment or components, Provisions can be made at an infrastructure level that can improve the reliability. Following are examples of such prov
35、i- sions. Space Constraints It is a common habit to consider the walls, floor, and ceiling as fixed elements, resulting in accepting them “as is.” Although this is for the most part the actuality in an existing building, this is unfortunately often consid- ered to be the case in new construction as
36、well. The basic floor plan layout and spatial definition is initially established prior to equipment layout and cooling sys- tem design. Seldom is that initial layout revisited with a critical eye after progress has been made on the equip- ment layouts and cooling system design. As a result, the phy
37、sical constraints dictated by the ini- tial floor plan layout may create conditions or environ- ments that are more difficult to design, construct, and operate. Although spatial or building costs (e.g., shell, envelope, or space) are a significant percentage of an overall project cost for commercial
38、 projects, this is not the case with datacom facilities where the infrastructure and equipment costs are typically more dominant. The cost impact for the basic space is further deemphasized for higher density cooling applications since the increased power and cooling infrastructure cost is so high.
39、Providing ample room volume or three-dimensional space is typically a relatively small cost item and a good investment. Even when spatial issues are identified at an early stage as potential obstacles, often there is a ratio- nalization that occurs preventing a change, perhaps due to time pressures
40、and a resistance toward change in gen- eral, etc. Such a reaction is shortsighted and typically results in time, cost, and reliability issues later that are far more severe than if it had been effectively addressed immediately. Modularity Modularity can be defined as a process utilizing standard- iz
41、ation and repetition in order to improve consistency and effi- 951 ciency. A modular approach to a large project or problem would involve breaking up the whole into smaller components e., modules), which could then be replicated to apply to the larger project. Design, construction, commissioning, an
42、d operation all usually benefit from modularity. Across all phases, modularity is beneficial since it creates an opportunity to focus more time in a concentrated area and then replicate it with relatively little effort rather than proportionally spending the time across the entire project or system.
43、 This modular approach also promotes consistency. From a design standpoint, the modular approach poten- tially frees up enough time in the schedule and man-hours to allow for more analysis, simulations, andor modeling to vali- date the concept. This can also create the opportunity for the design to
44、be scrutinized by more people since it is just one module (a subset), and this, in turn, can lead to further opti- mization of the design. From a construction standpoint, modularity creates the opportunity to prefabricate (fabricate in the shop) sections and assemblies. Shop fabrication takes place
45、in a more controlled environment with optimum machinery and resources, which, in turn, can achieve better, more reliable results. It can also reduce cost and allow for a shorter construction schedule by potentially allowing the use of larger or multiple construction crews or even separate companies
46、(e.g., one team per module or group of modules). From a commissioning standpoint, there is more time available to refine the prefunctional and functional testing as well as benchmarking one module against the other. From an operation standpoint, both normal operation and troubleshooting are simplifi
47、ed due to the repeatability. Here again, the potential to develop and understand the procedures better exists since it only needs to occur for a single module and not uniquely address the whole system and all of its subsets. By selecting standard components and equipment (i.e., wherever possible usi
48、ng off-the-shelf or one-day shipping items), the risk or impact of human error diminishes. Further, it reduces or eliminates the concern of the quantity of each item to be kept in inventory. Experienced technicians are critical to reliability and successful operation. Selecting unique or custom comp
49、onents and equipment can result in having few experienced techni- cians available or none locally available (e.g., having to be flown in). Material and Component Selection Focused on Reliability The correct approach needs to include consideration for the selection of components down to a detailed level to ensure an optimum decision is made. For example, although the selection of the components and materials for a piping system seems like a straightforward topic, there are many things that can impact its reliability, including type of piping material, expansion and contraction, supports, c
