Simulation and experimental investigation of condensation in residential venting.
A major market barrier to upgrading the efficiency of or fuel-switching with a heating system is the cost associated with the chimney liner and its installation. According to one utility's cost breakdown, the chimney liner constitutes 25% of the total conversion cost of fuel-switching. This conversion is subject to provisions of both NFPA 54 National Fuel Gas Code (NFGC) and NFPA 211 Standard for Chimneys, Fireplaces, Vents and Solid Fuel Burning Appliances. This is, however, in contrast to the costs brought upon by pitting corrosion, vent failure, and potential for unsafe venting of products of combustion (DeWerth, 1983). With respect to pertinent codes, it is the responsibility of both the regulators and regulated community to periodically re-examine the underlying assumptions and technical process behind standing code guidance. The research community may aid in the process, as with any unresolved technical issues that arise in that applicable code and installation practice evaluation can be addressed by modeling and possibly laboratory investigation to justify those code compliance interpretations.
In the current study, the project team conducted a technical evaluation and laboratory investigation of factors affecting venting requirements for gas vent systems in a representative utility service territory fuel-switching, thus vent system upgrades, are more common, the U.S. Northeast. Focusing on the potential for relining requirements during these upgrades, this study considers venting systems with masonry chimneys. Hydronic (hot water) boilers are prevalent in this region, which were treated as equivalent to furnaces in the original development of the NFGC (Philips, 1994):
"The differences between a boiler and a furnace for the purposes of a vent system analysis are twofold:
* The cycle rate at 50-percent load is 2 cycles per hour for a boiler (this directly affects furnace on-times as well), and
* A boiler is typically equipped with a draft hood and a stack damper.
Based upon these results, there is no significant difference between a boiler and a furnace in terms of exterior masonry chimney wet-times predicted by VENT-II (for the same steady-state efficiency)."
Higher efficient gas-fired appliances in Category I venting systems are more likely to result in condensation within exterior masonry chimneys. The use of vent dampers with hot water boilers reduces the off-cycle flow significantly, reducing the ability of the vent system to dry out between on-cycles. Additionally, boilers with vent dampers have different off-cycle flow characteristics than the furnaces used to develop the Category I appliance venting tables in the NFGC (hereafter referred to as "venting tables").
Performance is assessed primarily through numerical modeling, both using VENT-II and Computational Fluid Dynamics (CFD). VENT-II software was developed for the Gas Research Institute (GRI) during the original development of the venting tables (DeWerth, 1983). FLUENT version 6.3 is used for CFD, modeling fluid flow, heat transfer, and phase change. As VENT-II is a simpler 1-D modeling tool with a Graphical User Interface (GUI), it has been a useful tool in venting research. With its finer resolution and more sophisticated algorithms, CFD is used to assess the relative veracity of VENT-II in modeling venting systems for varied boiler efficiency, firing rate, ambient condition, number of exposed chimney walls, and presence of chimney liner. Simulation results are compared with experimental data from a full-scale test masonry chimney, using the metric Boolean continuously "wet" or "dry" conditions as used in the NFGC.
MODEL DEVELOPMENT AND ASSUMPTIONS
The objective in using CFD modeling is to use a sophisticated tool selected both to compare to VENT-II simulation results and experimental data, while exploring the venting dynamics of gas-fired boilers versus relative to that of fan-assisted furnaces, the subject of prior study and the venting tables. VENT-II is a one dimensional nodal model, solving a reduced form of the Navier-Stokes equations and a semi-empirical condensation model at the interior flue surface. In addition to cycling and ambient conditions, boundary conditions at the appliance outlets characterize flue gas properties relative to user-defined cycling, temperature profiles, firing rates, and percentage of excess air. FLUENT version 6.3 is the CFD simulation software used, with turbulence modeled using the k-[epsilon] model. The fluid domain includes the chimney interior and the ambient surroundings, primarily included for solution stability, and the solid domain is the chimney itself for heat transfer modeling. Sensitivity of condensation results to meshing of the chimney interior was explored, which is shown in Figure 1 in addition to this computational domain. Each simulation and chimney configuration utilized this mesh.
[FIGURE 1 OMITTED]
Initial screening CFD results that do not directly model condensation show order of magnitude agreements with the VENT-II analysis, there are known gaps in modeling physics concerning the mass and heat transfer associated with condensation. The source code of FLUENT is not accessible for modification; however it offers many so-called 'access points' where users may program additional physical models to be included with those built-in. This is how condensation is directly modeled, with customizable physical models. To directly model surface condensation in CFD, a so-called User Defined Function (UDF) is programmed which models condensation as a volumetric reaction rate coupled with the latent energy source/sink. To prevent numerical instabilities and solution divergence in the implementation of a UDF during CFD simulation, a simplified approach is taken to the UDF formulation. In brief, the UDF is defined as follows:
1. A volumetric reaction is defined through macros in the source code, which is preferential over a surface reaction to minimize the numerical instability caused by large intra-cell gradients. As water vapor is a small fraction of the overall flue gas and the fraction that condenses
is even smaller, gradients from cell face to cell face will be very large, causing solver oscillations.
2. Reviewing similarly employed condensation models in the literature, they fit into one of three categories in increasing complexity: simple mass flux, mass diffusion via an empirical diffusivity, and heat diffusion based upon film or drop-wise condensation on various geometries. After performing extensive modeling of flue gas condensation, Perujo et al. (2004) found that for a sufficiently fine mesh, the mass flux and mass diffusion models yield similar results within 10% of one another, which due to its simplicity was selected (Perujo, 2004). This model states conservatively that water vapor in excess of the saturated cell condition will condense. This is determined after calculating the water vapor saturation pressure at each cell (ASHRAE 2009), the actual and saturated water vapor concentration is determined.
3. Proportional to the water vapor mass condensing, there is a latent heat release for that cell.
In the baselining of this condensation modeling methodology, a non-compliant common venting of a fan-assisted furnace and natural-draft gas-fired water heater into both interior and exterior masonry chimneys is simulated with VENT-II and CFD. Comparing these simulation results, the interior and exterior masonry chimney simulations result in net condensation gains within the chimney segment above the roofline within 43% and 22% of each other (with CFD showing consistent overestimation), showing order of magnitude agreement. As the primary criteria of simulation-based venting table development was the Boolean "wet" or "not wet", this agreement is sufficient, and this modeling method was adopted.
Simulation Strategy & Boundary Conditions
Following initial confirmation that CFD modeling techniques to be used showed good agreement with the software used in the development of the venting tables, VENT-II, a parametric matrix for CFD modeling, was developed and refined in cooperation with the primary UTD sponsor. The metric used to determine agreement is the presence of net surface condensation on the chimney interior, continuously "wet" or "dry" conditions. The parametric modeling matrix incorporates typical physical and operating conditions seen in areas where oil-fired appliances are prevalent, namely the Northeastern U.S.
To reflect typical installations of this region, the space heating appliance of concern is a hydronic (hot water) boiler, as opposed to a furnace. Although previous studies involved with the development of the venting tables have stated that boilers and furnaces of equal capacity and efficiency result in approximately equivalent amount of condensation within the chimney, assumptions and boundary conditions will be based upon data from boiler operation. The focus was placed on exterior masonry chimneys, as the venting tables are more restrictive than for interior masonry chimneys.
A parametric matrix of modeling conditions is used for the comparison of simulation methods over a range of installation and operational conditions. Parameters and their variations are as follows: chimney construction with 1 or 3 exterior sides exposed, with a clay tile liner or an unlined double-brick chimney, an outdoor ambient "1,000-hour" outdoor temperature corresponding to Department of Energy (DOE) Region IV or V (26[degrees]F or 13[degrees]F, -3[degrees]C or -11[degrees]C), and a 78% steady state efficiency (SSE) natural draft or 83% SSE fan-assisted boiler. The following parameter choices were initially modeled and following discussion with the Technical Advisory Group (TAG), were eliminated: chimney height relative to roofline and varying the boiler firing rate. In addition to the marginal difference in chimney condensation observed in modeling results, utility representatives within the TAG confirmed anecdotally that 200,000 Btu/hr (59 kW) boilers and chimneys terminating at 3 feet (0.9 m) above the roofline are very common. Additionally, previous studies show that condensation is much less affected by firing rate (flue mass flow rate) than SSE (flue gas dew point temperature) (Philips, 1994).
To facilitate direct comparisons between simulation results and prior work in development of the venting tables, similar equipment characteristics are used. In all cases the natural draft gas-fired water heater simulated has: a steady state efficiency of 78%, a four hour cycle firing for 13 minutes, 45% excess air, and a maximum flue gas temperature of 497[degrees]F (258[degrees]C). For the boiler simulated with set point water temperature of 140[degrees]F (60[degrees]C), the following assumptions are made:
* The total cycle is 30 minutes, firing for 12 minutes - For a load factor of approximately 40%, corresponding to outdoor winter design conditions considered between 5[degrees]F and -10[degrees]F (-15[degrees]C and -23[degrees]C), participating manufacturers agreed this was a representative cycle. At design conditions, these systems are assumed to be sized appropriately.
* The starting flue temperatures for 78% and 83% SSE boilers are 140[degrees]F and 180[degrees]F (60[degrees]C and 82[degrees]C) respectively - The former reflects a worst case scenario for an atmospheric boiler reaching the target water temperature during off-cycle flow and the latter is based upon the experimental datasets provided by participating manufacturers.
* The 78% and 83% SSE boilers operate with 30% and 60% excess air respectively - The former is an estimate based upon performance of atmospheric furnaces of the same SSE (Philips, 1994) and the latter is based upon the manufacturer datasets which measured 12% [O.sub.2].
* The 78% SSE boilers have Electric Butterfly dampers within the first vent connector segment - These boilers are commonly fitted with vent dampers and Electric Butterfly dampers are used over other types due to their fast vent restriction, leading to cooler off-cycle flows, and resulting in conservative condensation estimates.
* The 83% SSE Boilers have off-cycle flows at 40% the magnitude of the on-cycle flow - This is an estimate consistent with the ASHRAE 103 & ANSI Z21.47 AFUE test and surveys of manufacturers and installers, covering the breadth of boiler system pressure drops.
The CFD simulation performed for each parametric case are two-fold, deriving both transient and steady state solutions, which are performed in parallel to maximize computational resources. The transient simulation will determine if chimney interior conditions are continuously wet at 1,000-Hour ambient temperature conditions. This transient simulation of a typical boiler cycle will have variable boundary conditions reflecting boiler cycling, as shown in Figure 2 for exiting flue gas temperature. Outdoor conditions reflect the "1,000 Hour" approach used in the prior development of NFGC venting tables (Philips, 1994), which is the heating season bin temperature for Regions IV and V at 1,000 cumulative hours. Similarly adopted from this prior analysis, the chimney mass will be pre-heated over an hour of full-firing followed by 12 hours of continuous cycling. If condensation occurs, in the form of supersaturated conditions at the chimney surface, the model will be considered to be continuously wet. This transient simulation is followed by a steady state simulation directly modeling the flue condensation at the chimney surface using the burner on conditions at the end of the firing (at the 12th minute), referring to Figure 2. The boundary condition properties are taken from the flue temperature curves and VENT-II results are averaged over the burner on portion of the cycle.
[FIGURE 2 OMITTED]
Boundary conditions for CFD modeling are developed through the following: (1) determine the exiting flue temperature and mass flow versus time, firing cycle, and overall cycle time for a given boiler efficiency, (2) import these into the VENT-II appliance model, (3) use the burner firing rate, thermal efficiency, selected water heater configuration, and ambient conditions to perform a VENT-II simulation, (4) import values for mass flow, static pressure, temperature, and species mass fraction at inlets to chimney into UDFs and CFD boundary conditions in transient and time-averaged form, dependent on CFD simulation type, and (4) initialize CFD model and preheat the chimney steady state firing boundary conditions.
Table 1 shows the matrix formed by these modeling parameter choices. Note that a combination of parameters is referred to as a "case", while individual modeling runs for each case are referred to as "simulations". For the cases shown, VENT-II and CFD modeling agree qualitatively on presence of net surface condensation (continuously "wet" or "dry" conditions), which is consistent with the relining guidelines of the NFGC venting tables. To utilize modeling resources efficiently and provided additional dimensions of comparison, a screening method was used for each case whereby (1) a fully transient simulation models the complete 13 hour cycle and models chimney condensation as a departure of relative humidity followed by (2) a steady-state model that determine the peak condensation rate during full-fire through direct modeling of surface condensation. Numerical disagreement exists between models on the peak steady state condensation rate, however they do not result in disagreement of net surface condensation over equipment on/off cycles modeled and later tested.
Table 1. Parametric Modeling Results Case No. CFD Simulations VENT-II SSC * Transient Cond. Rate oz./[ft.sup.2]-s (g/[m.sup.2]-s) 1 Y 2.7E-4 (8.3E-2) 6.9E-4 (2.1E-1) 2 N 1.1E-4 (3.4E-2) 4.9E-4 (1.5E-1) 3 ** Y 3.1E-4 (9.4E-2) 5.5E-5 (1.7E-2) 4 Y 7.6E-4 (2.3E-2) 1.1E-4 (3.3E-2) 5 N 1.2E-4 (3.7E-2) 6.5E-5 (2.0E-2) 6 Y 9.9E-5 (3.0E-2) 3.1E-4 (9.3E-2) 7 N Neg. *** 1.8E-5 (5.6E-3) 8 Y 8.7E-5 (2.7E-2) 1.7E-4 (5.3E-2) 9 N Neg. *** 0 10 ** N Neg. *** 0 11 N 3.3E-7 (1.0E-4) 1.1E-5 (3.2E-2) 12 N 1.9E-6 (5.7E-4) 1.1E-5 (3.2E-2) 13 N 2.2E-6 (6.7E-4) 1.1E-5 (3.3E-3) 14 N 4.0E-6 (1.2E-3) 0 15 N 2.4E-8 (7.4E-6) 0 16 N 3.7E-8 (1.1E-5) 0 17 N 1.1E-8 (3.3E-6) 0 18 N 1.1E-8 (3.4E-6) 0 Case No. Chimney Liner Exterior Sides Outdoor DOE Boiler SSE Exposed Region Yes No 1 3 IV V 78 83 1 1 1 1 1 2 2 2 2 2 3 ** 3 3 3 3 4 4 4 4 4 5 5 5 5 5 6 6 6 6 6 7 7 7 7 7 8 8 8 8 8 9 9 9 9 9 10 ** 10 10 10 10 11 11 11 11 11 12 12 12 12 12 13 13 13 13 13 14 14 14 14 14 15 15 15 15 15 16 16 16 16 16 17 17 17 17 17 18 18 18 18 18 * Steady State Condensation ** Input is 160 kBtuh (49 kW) not 200 kBtuh (59 kW) *** Negligible condensation rates are below 1e-8 oz. [H.sub.2]O/[ft.sup.2]-s (3.1e-6 g/[m.sup.2]-s)
The first 10 cases, for a double brick chimney (no clay tile liner), show that the most influential parameter with respect to net surface condensation (i.e. continuously wet conditions) was the boiler efficiency, as 83% SSE boilers created continuously wet conditions and 78% SSE boilers did not. A continuously wet condition is defined as where the chimney interior surface does not dry out completely following a one hour heat up and 12 hours of cycling (Philips, 1994). Minor influences are observed from: chimney configurations, outdoor temperatures, and input rates.
For the last 8 cases, with a clay tile-lined chimney, no transient simulations reported measureable condensation. The corresponding steady-state simulations reported steady state surface condensation rates that were between two to four orders of magnitude less than those of the prior, un-lined, 10 simulations. Following the reporting of these results, a member of the TAG took exception with the results for case 11, which based upon anecdote and field experience, continuously wet conditions were anticipated for this case. This case is further examined through experimental testing with a full-scale exterior masonry chimney. Note that Case 17 is the example graphical output shown in Figure 1.
As constructed, the chimney is 27' 8" (8.4 m) tall and has an 8" (20 cm) nominal clay tile liner. The clay tile liner is modified to accommodate 40 interior thermocouples (see Figure 3), one at each cardinal direction spaced vertically every 2 feet (61 cm), with an additional set just below the chimney exit. The hydronic boiler tested had an input of 195,000 Btu/hr (57 kW) and an AFUE of 80.3%, with the manifold pressure increased to match the name-plate firing rate. The natural draft gas-fired storage water heater had 40 gallons of storage, an Energy Factor (EF) of 0.59, and an input of 36,000 Btu/hr (10.5 kW). Once outside temperatures dropped to levels comparable to those simulated, testing commenced. The boiler tested was of a different efficiency than those modeled, thus the boundary conditions and User-Defined Functions (UDFs) programmed were modified, through direct measurement of the boiler and water heater flue gas temperature profiles.
[FIGURE 3 OMITTED]
The CFD simulation shows that surface temperatures are lower than local dew points and remain so during the on-cycle, producing condensing conditions without dry out. The off-cycle sees near-surface temperatures above the dew point; however the off-cycle flow is 15% of the on-cycle flow, which is insufficient to pick up the accumulated condensate. This results in a net gain of condensate and continuously wet conditions. These results qualitatively match the VENT-II simulation of a boiler equipped with a vent damper reducing off-cycle flow to 15% of on-cycle flow.
Comparing the on-cycle mid-chimney surface temperatures of another 13 hour test with an average outdoor temperature of 25[degrees]F (-4[degrees]C) with the previous transient CFD modeling results of Case 13, Figure 4, shows good agreement. As off-cycle flow modeled in Case 13 was larger, by approximately a factor of 2.8, the heat capacity of the off-cycle flow is increased, in part explaining the discrepancy in off-cycle temperatures. The relative agreement in chimney surface temperatures during the on-cycle, despite a difference of maximum boiler flue gas exit temperatures of over 40 [degrees]F (4[degrees]C), highlights the importance in off-cycle flow with respect to conditions within the chimney.
[FIGURE 4 OMITTED]
CFD modeling of Case 11, for an exterior masonry chimney with three sides exposed, a clay tile liner, outdoor temperature of 13 [degrees]F (-10.5[degrees]C), and an 83 % efficient boiler; reported dry conditions for the transient simulation and near negligible condensation for the steady state simulation, despite anecdotal expectations of continuously wet conditions. Three full 13 hour tests were performed, with photographs in Figure 5, which were different combinations tests with or without common venting with water heater and equipping the boiler with a vent damper: (1) both, (2) no water heater, and (3) no vent damper. Cases 1 and 2 were continuously wet, highlighting the importance of off-cycle flow magnitude. Prior expectations were likely based upon observations of boilers with a vent damper, reducing the off-cycle flow rate close to that previously measured 0.6 lb/min (0.3 kg/min). The initial CFD modeling of Case 11, like all cases simulated, used an estimate of 40% off-cycle flow consistent with the ASHRAE 103 & ANSI Z21.47 AFUE test and surveys of manufacturers and installers, covering the breadth of boiler system pressure drops. As measured, test 3 without a vent damper had an off-cycle flow of approximately 90% of on-cycle flow, compared to 40% and 15% of CFD run for Case 11 and test 1 respectively.
[FIGURE 5 OMITTED]
Historically, the NFGC venting tables were concerned with furnaces with respect to chimney and vent connector condensation. The development focused on fan-assisted furnaces, as modeled during the venting table development, which were not equipped with vent dampers. Boilers were determined to have an equivalent impact on venting systems with respect to the potential for in-flue condensation, despite differences in cycling, flue gas dew point temperatures, and presence of vent dampers. As boilers are prevalent in the U.S. Northeast, the focus of this investigation into the veracity of this assertion is with exterior masonry chimneys, while performing a comparison of two simulation tools, VENT-II and CFD. Following a direct comparison of results, the underlying physical models of VENT-II provide sufficiently accurate predictions of potential for condensation as compared to CFD modeling. VENT-II, CFD, and laboratory experimentation confirm the relining recommendations in the NFGC venting tables for the cases studied, using assumptions from their original development.
From this study, the primary factors influencing condensation within exterior masonry chimneys identified are the appliance efficiencies and magnitude of off-cycle flow from hot water boilers. With respect to causing continuously wet conditions, higher efficiencies reduce flue gas dew point temperatures, leading to condensing during the on-cycle, and a reduced off-cycle flow limits the drying of the chimney interior. In this study, test data of a hot water boiler and natural draft water heater venting common vented into an exterior masonry laboratory chimney, and subsequent model validation, have highlighted the relative importance of off-cycle flow magnitude over the appliance efficiency. Exploring this issue followed a discrepancy between Simulation Case 11 and expectations of continuously wet conditions, in which this discrepancy lied not in the underlying physics either modeling tool, but rather the conservative assumption of off-cycle flow per the AFUE test guidelines. The effectiveness of the vent damper and boiler off-cycle control, which provided a flow range of 15% to 40% of the on-cycle flow, is key to prevention of continuously wet conditions.
ASHRAE. 2009. ASHRAE Handbook: Fundamentals.
DeWerth, D. W. et al. "Venting Requirements for High Efficiency Gas-Fired Heating Equipment" Battelle, for Gas Research Institute (1983).
National Fire Protection Association Standard 31 "Standard for the Installation of Oil-Burning Equipment" (2006)
National Fire Protection Association Standard 54 "National Fuel Gas Code" (2009)
National Fire Protection Association Standard 211 "Standard for Chimneys, Fireplaces, Vents and Solid Fuel-Burning Equipment" (2006)
Jakob, F. E. et al. "U.S. Government Rulemaking Regarding the Efficiency of Furnaces, Boilers, and Direct Heating Equipment" Battelle, for Gas Research Institute (1997).
Perujo, M. P. "Condensation of Water Vapor and Acid Mixtures from Exhaust Gases". Ph.D. Dissertation, Technische Universitat Berlin, (2004).
Philips, D. B. et al. "Venting Gas Appliances into Exterior Masonry Chimneys: Analysis and Recommendations" Battelle, for Gas Research Institute (1994).
Paul Glanville, PE
Associate Member ASHRAE
|Printer friendly Cite/link Email Feedback|
|Author:||Glanville, Paul; Brand, Larry; Scott, Shawn|
|Date:||Jan 1, 2011|
|Previous Article:||Prediction of air mixing from high sidewall diffusers in cooling mode.|
|Next Article:||Energy efficiency design options for residential water heaters: economic impacts on consumers.|