Snow Melt System Performance
By ROBERT BEANHVAC Systems Hydronics
Understand the influence of sensitivity and storm characteristics.
I would speculate that if one were to do peer reviews of a thousand snow melt designs over the past decade, it is likely less than five per cent would have included a discussion with the client about the influence of storm characteristics on system performance. Despite the most current published research by ASHRAE, which incidentally is now 15 years old, designers still default exclusively to the mid-1950s Class Method for determining loads without considering how “load sensitivity” and “storm characteristics” might conflict with client expectations and the designer’s chosen design and recommended system.
For those unfamiliar with snow melt design, the historical Class Method stems in part back to the work of Adlam, Chapman and Katunich, who assigned a surface flux (Btu/hr ft2 (W/m2)) to maintain a percentage of the slab surface free of snow accumulation. The percentage was simplified by defining it as a free area ratio stated as 0.0, 0.5 and 1; this sequence was then later contracted respectively to the Class I, Class II and Class III system types.
Generally what this meant is at some combination of outdoor temperature, wind speed and snow fall rate, a Class I system with a 0.0 free area ratio would tolerate complete coverage of snow on the surface. That is 0 per cent melting on contact with an expectation that the surface would clear after several hours of operation. A Class II system at 0.5 free area ratio would tolerate 50 per cent accumulation (essentially slush). A Class III system at 1.0 free area ratio would accept no accumulation, that is 100 per cent melting on contact.
Class I was assigned by practitioners to low risk residential applications, Class II for moderate risk commercial projects and Class III for high risk institutional/industrial systems. Though this method served the state of knowledge of the time, ASHRAE research carried out in 1999 at the University of Minnesota by Ramsey et al and in 2001 by Spitler et al at Oklahoma State University (OSU) has since
reevaluated the calculation procedure, which includes a study of load sensitivity and storm characteristics on performance.
So, what changed? In lieu of the Class Method, Ramsey and his team introduced a design methodology around sensitivity tables defined as “Frequency Percentiles” in a project titled, “Development of Snow Melting Load Design Algorithms and Data for Locations around the World.” Using a refined set of algorithms, results now reported in the ASHRAE handbooks are presented in terms of “…frequency distributions that indicate the percentage of time that the required snow melting load does not exceed the reported value.” These frequencies were reported for percentages of 75, 90, 95, 98, 99 and 100.
Translation: if a maximum load calculated for an area ratio of 1 and represented by 100 per cent, and a design load is picked as 75 per cent, then it communicates to the client that the design load won’t exceed the maximum load 75 per cent of the time or it can be stated that 25 per cent of the time the design load will be exceeded.
Following this work, Spitler and his team looked at the methodology and noted, “Design loads (surface heat fluxes) have been calculated by taking the instantaneous weather conditions and calculating the flux required at the surface to provide a given free area ratio. In this type of calculation, no account is taken of the history of the storm up to the point of interest, and no account is taken of the dynamic response of the heated slab. However, this design heat flux can never be provided at the surface instantaneously.”
From this position ASHRAE funded the “Development of a Two-Dimensional Transient Model of Snow-Melting Systems, and Use of the Model for Analysis of Design Alternatives.” A sample screen shot is shown in Figure 1.
By incorporating the work from U of M with the work from OSU it is possible to look at what could happen to system performance in light of storm characteristics or what I like to call its “personality.” Such information becomes very useful in communicating to clients the potential results given methods of slab construction, design and assembly of the system and choices in control strategies.
Let’s look at one example using Figure 1. Make note from the upper part of the graph that the wind speed recorded at the beginning of the storm was at approximately four fps (4.4 km/hr) and increased over a 45 to 50 hour period to a peak of approximately 47 fps (51.6 km/hr) tapering off to approximately 10 fps (10.9 km/hr) after 77 hours. Make note of the gusts where the wind picked up and then dropped down. During this same period the dry bulb temperature gradually rose from 25F to 32F (-4C to 0C) and then declined to just above 0F (-18C) with the exception of three significant drops ranging from 25 to 30F (14C to 17C) below melting temperature. Note that the precipitation rate, which for the first 30 hours was rather mild with five instances of snow, was followed by a steadily increasing snow fall lasting for about 10 hours. It then tapered off to the 52-hour mark at which point no further snow fall was recorded.
Once you are comfortable understanding the transient personality of this particular storm, ask yourself how choices in control strategies such as “on/off” or “idle/on” might behave from an energy consumption and performance perspective? Would it matter if the slab was coupled to the ground or decoupled with insulation? Would it make a difference if tubes were located deep in the slab or midway? How would these and other design choices along with the storm characteristics fit in with various clients and their expectations? How likely would client’s expectations be met if these elements were not discussed?
Consider this storm with an on/off control strategy and no insulation. For the first seven hours and possible hours prior to the data, the upper portion of the slab would be at or closer to ambient conditions and the bottom of the slab would be at or closer to the temperature of the earth (this can be the same or not). Then sometime around the seven-hour mark the building operator or controls turned on the system in response to snow fall.
At the time of “on” there would exist a “pick up” load where the downward flow of heat to ground and upward flow of heat to surface heat would have to be of such intensity for some time to bring the slab up above the melting temperature during the snow fall. Because this first instance of a snow load was very low during rising ambient temperature and mild winds, it is quite possible that somewhere between hour 10 and 12 the system would be shut off. Shortly thereafter another snow fall occurs with greater intensity followed by an aggressive but short lived drop in temperature.
Given the following snow fall instances at hour 20 and 25, it is likely the system would be left in the “on” mode at least until hour 28, at which point it is possible the system would be turned off (or possibly left on). The critical change in characteristic occurs at hour 30 where there is a simultaneous increase in snow fall, increase in wind speed and drop in temperature lasting for 10 hours. If hour 30 occurred on a Friday evening when the operator was absent and unable to turn the system on there would be an accumulation of snow over the weekend.
Given the previous operation, if a control activated the system would the designed and delivered power be able to maintain the free area ratio without knowing if the snow in
tensity would continue during uncertain changes in temperatures and wind speed and wind direction? If the slab served a commercial building using a Class II rating for a free area ratio of 0.5 at a 150 Btu/hr ft2 (473 W/m2) but during hours 35 to 41 the actual load was 230 Btu/hr ft2 (725 W/m2), then it is very likely that the slab would accumulate 100 per cent snow coverage rendering it into a Class I system rather than the slush expected from a Class II system.
Perhaps the consequences for this client would be minimal for a flat slab but what if the slab was sloped steeply toward a pedestrian sidewalk and street? We would now have a health and safety liability issue to consider.
How would this change if an “idle/on” strategy was applied. With such a system the pickup load would be minimal and with sufficient capacity it would be able to maintain the system to meet client expectations. But what if the storm characteristics were such that only a drop in temperature occurred without snow fall, then the “idle/on” system would be consuming energy even though there was no load.
Neither the “on/off” or “idle/on” strategy is right or wrong but the choice will have an influence on energy consumption and how the system performs when exposed to different storm personalities.
In the Frequency Percentile method, the designer is required to evaluate and explain to the client the risks when a maximum system load of, say 100 per cent is designed to deliver something less than 100 per cent. In this example, if the client decided to design for 100 per cent load at 230 Btu/hr ft2 (725 W/m2) using an “idle/on” control the system would maintain a free area ratio of 1.0 during this storm. Every drop in percentile thereafter would increase the risk of the system developing slush or having a complete accumulation of snow.
The Class Method lacks the sensitivity analysis provided by the Frequency Percentile Method and was inadequate in explaining to owners the effects of transient loads, variable expectations, appropriate systems, with associated capital and operating costs. With an understanding of
sensitivity and storm characteristics it becomes easy to explain to clients. <>
Bean, R. 2012. Thermal to Hydraulic Calculation Procedure for Snow Melting Systems, ASHRAE Conference San Antonio. Seminar: Energy Efficient Snow Melt System
Ramsey, J.W., M.J. Hewett, T.H. Kuen, S.D. Petersen. T.J. Spielman, A. Briefer, (1999), Development of Snow Melting Load Design Algorithms and Data for Locations Around the world (ASHRAE 926-RP), Final Report. Atlanta: American Society of Heating Refrigerating and Air-conditioning Engineers Inc.
Spitler, J.D., S.J. Rees, X. Xiao, M. Chulliparambil. 2001. Development of a Two-Dimensional Transient Model of Snow-Melting Systems, And Use of the Model for Analysis of Design Alternatives (RP-1090), Final Report, Atlanta, GA: American Society of Heating Refrigerating and Air-conditioning Engineers Inc.
XIAO, X. 2002. Modeling of Hydronic and Electric-Cable Snow-Melting Systems for Pavements and Bridge Decks, M.Sc. Thesis, Oklahoma State University