Introduction *

Pedagogic note: This module addresses several areas where field data is not commonly supportive of general concepts of stormwater behavior. These areas of "misbehavior" can and have lead to improper stormwater management strategies. This module therefore presents some quantitative and non-quantitative information from numerous field monitoring activities that portray relationships concerning urban hydrology and pollutant concentrations that can lead to a more general approach in stormwater management.

Source: The material in this module was mostly extracted from the following final draft EPA report:

Pitt, R., M. Lilburn, S. Nix, S.R. Durrans, S. Burian, J. Voorhees, and J. Martinson Guidance Manual for Integrated Wet Weather Flow (WWF) Collection and Treatment Systems for Newly Urbanized Areas (New WWF Systems). U.S. Environmental Protection Agency. 612 pgs. Expected publication in 2000.

Previous discussions have also been presented at Bill’s stormwater modeling and management conferences (Conference on stormwater and urban water systems modeling) in Toronto over the past several years, as published in the conference proceedings:

Pitt, R. and J. Lantrip. "Infiltration through disturbed urban soils." In: Advances in Modeling the Management of Stormwater Impacts, Volume 8. (Edited by W. James). Computational Hydraulics International, Guelph, Ontario. 1999.

Pitt, R. "Small storm hydrology and why it is important for the design of stormwater control practices." In: Advances in Modeling the Management of Stormwater Impacts, Volume 7. (Edited by W. James). Computational Hydraulics International, Guelph, Ontario and Lewis Publishers/CRC Press. 1998.

Pitt, R. "Unique Features of the Source Loading and Management Model (SLAMM)." In: Advances in Modeling the Management of Stormwater Impacts, Volume 6. (Edited by W. James). Computational Hydraulics International, Guelph, Ontario and Lewis Publishers/CRC Press. pp. 13 – 37. 1997.

Some topics included in this module have also briefly appeared in previously. They have been repeated here because only a few students are studying all of the posted modules.

There are many assumptions that stormwater researchers have relied on over the years in their analyses and designs. Unfortunately, some of these are incorrect and have lead to improper and inefficient strategies. Misconceptions and improper data evaluations have also become incorporated into many planning and design models. As an example, Module 7 discussed a number of problems associated with common methods to evaluate accumulation and washoff of pollutants and showed how these improper representations can lead to gross errors in evaluating source controls for stormwater management. In this module, additional information is presented concerning a number of other misconceptions that have become "institutionalized." The topics in this module address some urban hydrology issues and relationships of runoff concentrations with hydrologic parameters. This module also presents a suggested general approach for stormwater designs that may be applicable for a wide range of conditions.

Different drainage design criteria and receiving water use objectives often require the examination of different types of rains for the design of urban drainage systems. These different (and often conflicting) objectives of a stormwater drainage system can be addressed by using distinct portions of the long-term rainfall record. Several historical examinations (including Heaney, et al. 1977) have also considered the need for the examination of a wide range of rain events for drainage design. However, the lack of efficient computer resources severely restricted long-term analyses in the past. Currently, computer resources are much more available and are capable of much more comprehensive investigations (Gregory and James 1996). In addition to having more efficient computational resources, it is also necessary to re-examine some of the fundamental urban hydrology modeling assumptions (Pitt 1987). Most of the urban hydrology methods currently used for drainage design have been successfully used for large "design" storms. Obviously, this approach (providing urban areas safe from excessive flooding and associated flood related damages) is the most critical objective of urban drainage. However, it is now possible (and legally required in many areas) to provide urban drainage systems that also minimizes other problems associated with urban stormwater. This broader set of urban drainage objectives requires a broader approach to drainage design, and the use of hydrology methods with different assumptions and simplifications.

Runoff volume is usually the most important hydrology parameter in water quality studies, while peak flow rate and time of concentration are usually the most important hydrologic parameters for flooding and drainage studies. The relationships between these different hydrologic parameters and rain parameters are significantly different for different classes of rains. Runoff models for water quality investigations should therefore be different than the runoff models for flooding and drainage investigations. Similarly, flooding and drainage investigations should normally not use a hydrology model developed for water quality investigations.

The importance of different areas in a watershed as pollutant sources is dependent on accurate hydrology predictions. One also need to know the variations of each source area’s importance for different rains. Many control practice designs also depend on inflow hydrology. If one incorrectly predicts the sources of pollutants or flows, then one will not get expected stormwater control benefits. The material in this module briefly describes a method to accurately predict the sources of urban runoff source flows during important small rains. This method is fundamental to the Source Loading and Management Model (SLAMM). A preliminary interface program now allows SLAMM to be used in conjunction with the SWMM model by replacing SWMM’s RUNOFF module with SLAMM. This allows the more accurate site description and control information available in SLAMM to be used in conjunction with the more detailed representation available in SWMM’s EXTRAN module.

Most existing stormwater models incorrectly predict flows associated with small rains in urban areas. This is important because common small storms are responsible for most of the annual urban runoff discharge quantities throughout North America (EPA 1983, Pitt 1987). Most existing urban runoff models originated from drainage and flooding evaluation procedures that emphasized very large rains (several inches in depth). These large storms only contribute very small portions of the annual average discharges. Obviously, the pollutant shock loadings and habitat destruction caused by a large storm may create significant receiving water use impairments, but a number of years will be available for recovery before another massive rain occurs. However, moderate storms, occurring several times a year, are responsible for the majority of the pollutant discharges. The effects caused by these frequent discharges are mostly chronic in nature (such as contaminated sediment and frequent high flow rates) and the interevent periods are not long enough to allow the receiving water conditions to recover (Pitt and Bozeman 1982).

Simplifying the assumptions concerning runoff losses for impervious and pervious areas for small rains has little significance on the accuracy of the predictions of runoff volumes for large rains. These same assumptions, however, cause dramatically large errors when predicting runoff associated with small rains, the rains of most importance for water pollutant discharges. The significance of small rains as important pollutant generators is then missed and controls are then designed for wrong storms and wrong source areas. The hydrology prediction method described here is a simplified procedure used to predict runoff volumes from individual homogeneous areas for a wide variety of rains. It requires knowledge of certain development characteristics of the urban area.

The following information starts with a comparison of general storm drainage design procedures as they were used in the mid 1960s compared to the mid 1990s in the US. Some historical comments are also provided. (see Burian, et al. 1999 and the web page at: http://www.eng.ua.edu/~awra/download.htm for a more complete description of the historical development of urban wet weather flow issues.) Drainage design objectives that may be applicable for a wide range of conditions are also presented. The controversial topic (at least in the US) of the potential use of combined sewers in high density areas is also discussed. Designs for the future, as envisioned by a group of engineers, scientists, regulators, and planners attending an Engineering Foundation conference in Malmo, Sweden in 1997 are also summarized. Detailed information from field observations is presented in the appendix describing many of the concepts associated with urban hydrology and water quality that lead to the development of the recommended stormwater management strategy presented earlier in this module.

Several interesting correlations can be made when comparing answers obtained in this recent survey with those gathered during the previous survey. Of particular notice, in the 1967 survey, 70% of the reporting cities supported the use of 5- to 10-year design storms. Those cities with significantly different responses used smaller, rather than larger, storms. In the 1997 survey, the majority of participants also used storms in approximately the same range, but with most stating that they used a 10-year storm.

The earlier survey demonstrated that "practically all" responding cities used the rational method for design. However, most cities were not using this procedure correctly. Typically, either the runoff coefficient, or the rainfall intensity, were determined incorrectly. The most significant problem was the use of the 24-hr averaged storm rain intensity instead of the rain intensity associated with the drainage area time of concentration. This error can cause gross under-designs of drainage systems. In the recent survey, a majority of engineers still use the rational method for design, but correct use could not be measured during the simple survey. However, there seemed to be a good understanding of time of concentration methods. Newer methods, such as those developed by NRCS (such as TR-55), are beginning to be used more in design practices. These methods found significant use in larger watersheds, a positive indication of the realization of the limitations of the rational method by engineers.

System failure indicators were another factor examined in both surveys. In the earlier UW survey, it was determined that the most common indicator of system failure was water ponding at inlets. Although this was a concern of engineers during the recent survey, it was not as prevalent. The second leading sign of system failure in the 1967 survey was water ponding in back yards. Again, this was not a high priority for current design engineers today.

Answers obtained in the two surveys give a similar picture of stormwater pollution. The same constituents were mentioned in both groups of responses. Reoccurring answers included sediment, oil and grease, salts, and fertilizers. It appears that the same body of common knowledge concerning stormwater pollution was present thirty years ago as it is today. However, there has been little use of stormwater pollution control measures during the past 30 years, even though recognition of the problem was common.

Much can be learned from observing past WWF management practices. Indeed, the review of the literature has provided helpful insights that should prove useful in developing future WWF management strategies. The following characteristics were often observed in successful strategies or were conspicuously missing from unsuccessful strategies. The list provided below indicates considerations that should be incorporated into future WWF management strategies:

• technology transfer
• user friendly design methods and tools
• political, social, and economic ramifications
• sustainability of design
• goal of wet weather system should be to mitigate impacts on the environment
• designs should be optimized in terms of pollutant control, receiving water impacts, and cost

McPherson (1975; 1978) voiced concerns more than 20 years ago and offered suggestions to reduce the technology transfer (development to implementation) lag time. Professional societies have published monographs with the purpose of bridging the gap between research and practice (Kibler 1982). History has displayed examples of the technology transfer time lag. Take the prediction of runoff from a watershed as an example. The formula methods, such as McMath, Roe, and Burkli-Ziegler, dominated sewer design in the late 1800s. The rational method of determining stormwater runoff was introduced to the United States by Emil Kuichling in 1889, but it did not become a widely utilized method until much later. A paper by Charles Buerger (1915) states:

It [the rational method] is not widely used, however, and the formula methods, of which the Burkli-Ziegler and the McMath are the most popular, are generally used, in spite of the common realization of the fact that the results given by them lack consistency, and are very erratic and unreliable.

This statement can be applied today, except now the rational method might be considered the method that engineers are continuing to embrace while the new technology that has been introduced recently is not being implemented. The reasoning Buerger offered in 1915 for the lack of implementation is even more interesting. He stated that the rational method had not received the widest use because it was relatively laborious, and required a material exercise in judgment. This again is a popular reason expressed today for the lack of application of other techniques.

An advantage of developing user friendly design methods and tools is the reduction in the time lag between development and implementation. Practitioners generally embrace technology that is simple to understand while still providing the means to perform the job in the most cost effective manner possible. The methods and tools that have gained application through history have been simple to implement and easy to understand, although not necessarily the most accurate or appropriate.

Another consideration noticed during the review of the literature is that past design engineers and planners were forced to consider the socioeconomic, political, and legal ramifications associated with their plans and designs. These topics can be the primary inhibitors to the implementation of innovative technology and in the future must be addressed for progress to be made (Berwick, et al. 1980). Berwick, et al. (1980) and others have reviewed the reasons for lack of implementation and attribute it to a variety of problems. Some of the problems have been identified as the regulatory framework surrounding development, risks associated with development, public attitudes, and others. A future design methodology for WWF management will have an advantage if it considers the socioeconomic, political, and legal implications of system implementation.

Considering the other points listed above, sustainable development will have the benefit of significantly reducing the environmental impacts over time associated with a project; while promoting economic stability as well. The literature is replete with examples of entire systems (Paris in the middle ages) or parts of systems that were designed without considering the long-term sustainability of the project. The systems performed poorly and resulted in additional money being contributed to rehabilitate and maintain the design.

Even though domestic sewage collection systems are not a major topic for this research, the topic cannot be ignored when addressing wet weather flow. The continued use of combined sewer systems is common in many parts of the world, and the U.S. has many existing combined systems still in use. In addition, separate sewer overflows (SSOs) are also common in many urban areas that only have separate systems. Overflows of raw sewage during wet weather is therefore unfortunately common in many areas of the U.S. Overlooking these wet weather problems can badly distort efforts in stormwater management. In addition, there is renewed interest in the use of combined sewer systems in the U.S. under specific conditions, where their use (in conjunction with improved treatment facilities) may result in reduced, and more cost-effective, WWF discharges. Heaney, et al. (1997) for example, found that combined systems may discharge a smaller pollutant load to a receiving water than separate systems in cases where the stormwater is discharged untreated and where the sanitary wastewater is well treated. They present an example in southern Germany where combined sewer systems are being designed with extensive infiltration components to reduce the inflow of stormwater to the drainage system, reducing the frequency and magnitude of CSO events. Similar systems are also used in Switzerland and in Japan with comparable results.

Some of the important issues facing the use of combined sewers in the future include:

· the use of separate versus combined sewers and under what watershed/demographic conditions and

characteristics warrant separate versus combined systems;

· the concept of larger size combined sewers providing for inline storage and flushing cells with or without

steeper slopes and bottom shapes to alleviate antecedent dry-weather flow solids deposition; and

· taking advantage of new construction for larger capacity of dry-weather flow treatment and sludge

handling facilities to accommodate additional flow during wet weather conditions.

· solids deposition in sewerage and prevention of solids from entering sewerage

Conditions for the use of Combined Sewers

The debate on the use of combined sewers has been long. As noted above, Hering (1881) visited Europe and made recommendations to the U.S. National Board of Health concerning the use of combined sewers. He recommended that combined sewers be used in extensive and closely built-up districts (generally large or rapidly growing cities), while using separate systems for areas where rainwater did not need to be removed in underground drainage conveyance systems. His recommendations were largely ignored. Combined sewers were extensively used in many of the older U.S. cities because of perceived cost savings. Of course, the existing combined sewer systems in the U.S. are now mostly located in the most dense portions of central cities, along with some of the older residential areas. Many newer separate sanitary sewer systems also connect to downstream combined systems. In addition, current separate sewer systems actually may operate as combined systems due to excessive infiltration of sewage into stormwater systems, or by direct connections of sewage into stormwater systems.

Current Separate Systems that are actually Combined Systems. Unfortunately, many separate sanitary sewage collection systems in the U.S. are in poor repair, resulting in inappropriate discharges of sewage into receiving waters. Pitt, et al. (1994) developed a method for cities to identify and correct inappropriate discharges. The following discussion is from this user guide.

Current interest in illicit or inappropriate connections to storm drainage systems is an outgrowth of investigations into the larger problem of determining the role urban stormwater runoff plays as a contributor to receiving water quality problems. Urban stormwater runoff is traditionally defined as that portion of precipitation which drains from city surfaces exposed to precipitation and flows via natural or man-made drainage systems into receiving waters. Urban stormwater runoff also includes waters from many other sources which find their way into storm drainage systems. For example, Montoya (1987) found that slightly less than half the water discharged from Sacramento's stormwater drainage system was not directly attributable to precipitation. Sources of some of this water can be identified and accounted for by examining current NPDES (National Pollutant Discharge Elimination System) permit records, for permitted industrial wastewaters that can be discharged to the storm drainage system. However, most of the water comes from other sources, including illicit and/or inappropriate entries to the storm drainage system. These entries can account for a significant amount of the pollutants discharged from storm sewerage systems (Pitt and McLean 1986).

Three categories of non-stormwater outfall discharges were identified by Pitt, et al. (1994): pathogenic/toxicant, nuisance and aquatic life threatening, and clean water. The most important category is for stormwater outfalls contributing pathogens or toxicants. The most likely sources for this category are sanitary or industrial wastewaters. Section 402 (p)(3)(B)(ii) of the 1987 reenactment of the federal Clean Water Act (CWA) requires that National Pollutant Discharge Elimination System (NPDES) permits for municipal separate storm sewers shall include a requirement to effectively prohibit problematic non-stormwater discharges into storm sewers. Pitt, et al. (1994) developed a scheme to identify and correct problem outfalls to allow compliance with these CWA requirements. Outfall analysis surveys should have a high probability of identifying all of the outfalls in this most critical category. High probabilities of detection of other contaminated outfalls are also likely when using these procedures. After identification of the contaminated outfalls, their associated drainage areas are then subjected to a detailed source identification investigation. The identified pollutant sources are then corrected.

Sanitary sewage finds its way into separate storm sewers in a number of ways. Direct cross-connections may tie sanitary lines directly to storm drains (relatively rare), or seepage from leaking joints and cracked pipes in the sanitary collection system can infiltrate storm sewers (much more common). Surface malfunctions and insufficiently treated wastewater from septic tanks may contribute pollutants to separate storm sewers directly or by way of contaminated groundwater infiltration. Seepage of sewage or septic tank effluent (septage) into underground portions of buildings may be pumped into separate storm sewers by sump pumps (EPA 1989).

Due to indifference, ignorance, poor enforcement of ordinances, or other reasons, a stormwater drainage system may have sanitary wastewater sewerage direct connections. Obviously, the sanitary wastewater entering the storm drain will not receive any treatment and will pollute a large flow of stormwater, in addition to the receiving water. If the storm drain has a low dry-weather flow rate, the presence of sanitary wastewater may be obvious due to toilet paper, feces, and odors. In cases of high dry-weather flows, it may be more difficult to obviously detect raw sanitary wastewaters due to the low percentage of sanitary wastewater in the mixture. Even though the sanitary wastewater fraction may be low, the pathogenic microorganism counts may be exceedingly high.

Corrective measures involve undertaking a program of disconnecting the sanitary sewer connections to the storm drainage system and reconnecting them to a proper sanitary wastewater sewerage system. The storm drainage system then has to be repaired so that the holes left by the disconnected sanitary sewer entrances do not become a location for dirt and groundwater to enter. However, there are situations in which the sanitary system is so connected to the stormwater system that good intentions, vigilance, and reasonable remedial actions will not be sufficient to solve the problems. In an extreme case, it may be that while it was thought that a community had a separate sanitary sewer system and a separate storm drainage system, in reality the storm drainage system is acting as a combined sewer system. When recognized for what it really is, the alternatives for the future become clearer: undertake the considerable investment and commitment to rebuild the system as a truly separate system, or recognize the system as a combined sewer system, and operate it as such, without the disillusionment that it is a problem-plagued storm drainage system which can be rehabilitated.

It would be best to correct at least the sanitary sewer if only one drainage system can be corrected. This would have the dual advantage of preventing infiltration of high or percolating groundwaters into the sanitary sewerage and preventing pollution of stormwater with exfiltrating sanitary wastewater. Rehabilitation of either drainage systems by use of inserted liners, or otherwise patching leaking areas, are possible corrective measures. It is important that all drains with infiltration problems be corrected for this corrective action to be effective. This would also include repairing house lateral sanitary wastewater lines, as well as the main drainage runs. However, these corrective measures are more likely to be cost effective when only a relatively small part of the complete drainage systems require rehabilitation.

Normally, widespread failure of septic tank systems might necessitate the construction of a sanitary sewer to replace the septic tanks. Also, identifying and disconnecting sanitary sewers from the storm drainage system is usually undertaken. Connections (whether directly by piping or indirectly by exfiltration or infiltration) of sanitary sewers to the storm drainage system may be so widespread that the storm drainage system has to be recognized as a combined sewer system. This could also be the case when the prevalence of septic tank failures leads to widespread sanitary wastewater runoff to the storm drainage system. One usually thinks of a combined sewer system as having all of the sanitary sewer connections to the same sewers that carry stormwater, but, there are degrees of a storm drainage system becoming a combined sewer system. Prior to these actions taking place, the storm drainage system operates to some degree as a combined sewer system. It may be that the sanitary sewerage system is not capable of handling the load that would be imposed on it if a complete sewer separation program were undertaken. Or, in an extreme case, no sanitary sewer system may exist.

By recognizing that a combined sewer system does in fact exist may help to focus attention on appropriate remedial measures. The resources may not be available to undertake construction of a separate sanitary wastewater drainage system. One should then focus on how to manage the combined sewer system that is in place. Conventional CSO end-of-pipe storage/treatment needs to be investigated, in addition to methods to reduce the entry of stormwater into the drainage system (through upland infiltration, for example). Also, the combined sewer system may be tied into other combined sewers so that more centralized treatment and storage can be applied. While operation of a combined sewer system is not a desirable option, it may be preferable to having the stormwater and the large number of sanitary entries receive no treatment.

An early identification and decision to designate a storm drainage system a combined sewer system, will prevent abortive time and costs being spent on further investigations. These resources can then be more effectively used to treat the newly designated combined sewer system. In essence, recognition of a system as being a combined sewer system provides a focus in the regulatory community so that it may be possible to operate the system so as to minimize the damage to the environment. Plans can then be developed to provide the resources to separate the system.

Conditions where New Combined Systems may be Appropriate. As noted above, it may be more cost-effective and result in the least pollutant discharges to operate separate drainage systems that are badly in need of repair as actual combined sewer systems, compared to costly and ineffective repairs to the separate systems. However, proposed construction of new combined sewer systems would be very controversial in the U.S. and it would be very difficult to overcome resistance to their construction. The main areas of resistance relate to the massive efforts expended in the last several decades in reducing the number and severity of combined sewer overflows (CSOs), usually under court order. In addition, current interest and massive correction efforts to control separate sewer overflows (SSOs) in many cities would also result in a great deal of resistance from engineers, municipalities, regulatory agencies and environmental groups to the construction of new combined sewer systems. The political resistance to the construction of new combined sewer systems in the U.S. is therefore considered almost insurmountable. However, it may be interesting to note where they may be appropriate from a technical viewpoint.

As pointed out by Hering in 1881, combined sewer systems may be suitable in dense urban areas, where the sanitary sewage flow is relatively high per area. Of course, any use of a combined sewer must be accompanied with provisions to reduce any untreated overflows to almost zero. In reality, the current level of untreated sanitary sewage discharges in urban areas from badly functioning separate systems is likely much higher than anyone acknowledges or considers when conducting wet weather flow management projects. The major concern with combined sewer systems is the overflow discharges of dangerous levels of pathogenic microorganisms, and nuisance conditions associated with floatable debris and noxious sediment accumulations. Discharges of potentially dangerous medical wastes and drug paraphernalia is also of great concern. However, it may be possible to construct a new combined sewer system that would operate with fewer annual untreated discharges of sewage than many currently separate systems, plus provide treatment of stormwater. The following attributes would be helpful for any new sewerage system, especially a combined system:

· The major goal of any new WWF collection system should be the minimization of stormwater runoff and sanitary wastewater entering the system. As noted previously, there are many beneficial uses of stormwater that could account for substantial fractions of the annual runoff. Similarly, household water conservation (especially low-flow toilets and reduced flow showerheads, etc.) can also substantially reduce wastewater flows to the sewerage.

· The conveyance system could be either a conventional combined system, or one of two possible new scenarios that would reduce the flows in the sewerage that could cause CSOs or SSOs. These new options include: 1) utilize a flow storage tank at each household to retain sanitary wastewater during wet weather, or 2) prohibit the entry of stormwater into the sewerage at a level that would cause overflows. The effective use of an existing conventional combined sewer system would require extensive modifications to provide adequate storage and increased treatment capacity to reduce overflows. These new options are briefly described below:

The first option may be termed a shared sewer system as the two flows (stormwater and sanitary wastewater) are not co-mingled at the same time in the single drainage system, but are kept separate as much as possible. This option, commonly used in England in the later part of the last century, and recently re-introduced by Pruel (1996) would require an adequately sized storage tank that could hold household wastewater for specific periods of time (depending on rain durations, conveyance capabilities, and treatment rate available). Figure 1 (Reyburn 1989) shows a old drawing of sanitary fittings and drains from a catalogue from Thos. Crapper & Co., Ltd., Sanitary Engineers, Chelsea, England. The house connections are all directed to an intercepting chamber which receives the branch drains from the house. This chamber is vented and is fitted with a trap. The large intercepting chamber is connected to the public sewer. In this drawing, the roof runoff is also directly connected to the intercepting chamber, possibly as an aid in flushing the chamber.

The intercepting chamber would normally be empty, with the wastewater flowing across the bottom of the tank in a small-flow channel (for an in-line installation), or the tank could be off-line. During wet weather, a flapper valve or other fitting at the connection to the full-flowing sewer would prevent additional water from entering the drainage, causing wastewater to back up into the intercepting chamber. When the wet weather flow subsided, the tank would empty into the sewerage. In a modern application, tank flushing could be accomplished (possibly using captured stormwater) with a tipping bucket or sprays to remove any settled solids in the tank. The flushing mechanisms would not need to be very complex. The initial higher flows (less than the capacity of the treatment facility) in the sewerage would therefore be mostly stormwater and would be used to flush solids, that accumulated during the low-flow sanitary wastewater flow conditions, to the treatment facility. This "first flush" would therefore be captured, along with a sizeable amount of stormwater, for treatment. As the WWF exceeded the capacity of the treatment facility, overflows of stormwater, with little sanitary sewage, would occur. There are many options available that can be used to temporarily increase the capacity of the treatment facility, or to provide temporary storage before treatment. In addition, many end-of-pipe stormwater treatment options are available to treat the smaller quantities of stormwater that would be discharged through the overflows.

Preul (1996) calculated the needed on-site storage volumes for this "shared sewer" concept. His "combined sewer prevention system" (CSPS) was investigated for locations in Cincinnati, Ohio, and in Toronto, Ontario. He found that storage tanks capable of detaining household sanitary wastewater on-site for 6 hours in Cincinnati would prevent about 90% of the CSO occurrences. The Toronto location would only require on-site detention capabilities of 3 hours for similar benefits. He has predicted an expected domestic wastewater production of about 60 to 80 liters per person per day in the future, with the required use of low water use plumbing fixtures. For a typical 2.8 person household, the daily sanitary wastewater flow in Cincinnati would be about 170 to 220 L per day per household. Therefore, a household storage volume of 55 L would provide 6 hours of average storage and 90% control of CSO occurrences. A 220 L storage capacity per household would virtually eliminate all CSOs in Cincinnati. Required household storage capacities in Toronto would be even less, with 30L storage tanks providing almost complete control. These are all relatively small volumes and would cost only a very modest amount, if designed and constructed at the time the housing units are built.

Another option is basically a separate sanitary sewerage system that is constructed to be very water-tight. This would be a less complex option than above, in some ways, but does require very good construction and maintenance practices. The sanitary sewerage system may be best a vacuum or small diameter pressurized system, both having been used for many years at numerous locations throughout the U.S. The stormwater would be conveyed separately, emphasizing on-site reuse and infiltration, through either open channels if compatible with the land use, or through a separate drainage system. Critical source area controls would be utilized, along with end-of-pipe treatment, as appropriate. With a tight conveyance system, no extra stormwater could enter the sanitary sewerage, greatly lessening the threat of overflows during wet weather.

Use of Larger, Steeper, and More Efficient Cross-Sections for Combined Sewers

According to Field, et al. (1994), new urban areas or upstream additions to older combined sewer systems should use advanced combined sewer designs requiring larger diameter sewers having steeper slopes and more effective bottom cross-sections to add storage capacity to the system and eliminate antecedent dry weather flow pollutant deposition and resulting pollutant concentrated storm flushes (Field 1975, 1980, and 1990b; Kaufman and Lai 1978; Sonnen 1977). The additional capital cost of an advanced combined sewer system would be incrementally small,

considering the overall cost of installing a conventional combined sewer system or a two-pipe separate (storm and sanitary) sewer system, and the cost effectiveness for storm-flow pollution control.

Larger combined sewers would provide in-system storage for short periods of excessive flows, and would allow larger flows to be conveyed to the treatment facility. Inflatable dams in the sewerage could be used to selectively back up water in the sewerage, reducing excessive flows. Upland detention can also be used to significantly reduce stormwater flows. Stormwater flows can be captured and detained at many locations before entering the drainage system. Temporary rooftop storage, parking lot storage, and even limited road flooding have been used to reduce stormwater flows into combined sewers. Conventional stormwater detention facilities are also available for storage of large volumes of stormwater. However, the use of extensive stormwater infiltration, as demonstrated in Germany, Switzerland, Canada, and in Tokyo in areas having combined sewerage appears to be very effective in reducing CSO volumes and frequency. The previously described household detention of sanitary wastewater should also be considered in conjunction with increased in-line storage and conveyance capacity. Of course, in order to be effective, treatment capacity would need to be increased to allow for a greater portion of the WWF to be treated. The following discussion presents several methods for increasing the treatment facility capacity for combined sewerage systems.

Solids in Sewers

Heaney, et al. (1997) stated that historically, sanitary sewers were designed primarily based on peak sewage flow rates, assuming that solids would be carried with the sewage if simple guidelines were followed. Generally, these guidelines require sewage flow rates of between 0.6 and 3.5 m/sec. Much more can be done to more effectively accommodate solids in sewers, however. Knowledge about solids in sewers and their associated pollutants is extensive after more than a decade of detailed research in Europe and Scandinavia, and elsewhere (USA and Japan in particular) prior to that, but little of this work has been incorporated in modern sewerage design. However, there are still significant outstanding uncertainties and research is continuing worldwide. The sewer sediments working group (SSWG) of the Joint Committee on Urban Storm Drainage of IAWQ/IAHR is producing a Scientific and Technical Report entitled Solids in Sewers: state of the art, and subtitled Characteristics, effects and control of sewer solids and associated pollutants which will summarize the available knowledge, and recommend future research directions (Ashley, et al. 1996). The following briefly summarizes these solids in sewers issues covered in this special report that have dramatic effects on combined sewer and separate sanitary sewer design and maintenance.

Origins, occurrence, nature and transport of solids in sewers. The emerging importance of sewers as a part of the treatment process and interaction with treatment plants has recently led to the concept of the "sewer as a reactor" (Hvitved-Jacobsen, et al. 1995). In-sewer processes are perhaps the least understood aspect of sewer solids. The transport and movement processes and mechanisms, together with aggregation and disaggregation effects, sediment deposition, change in nature and subsequent erosion and transport are all important processes. There are particular problems which differentiate sewers from fluvial sediment transport systems, such as source limitation, rigid non-erodible boundaries and organic effects.

Effects sewer solids have on the performance of wastewater systems. Problems caused by sewer solids relate to physical effects, such as blockages, conveyance constraints, and overall effects on the hydraulics. These all affect the relative roughness of the boundary between the flowing wastewater and the pipe material. The quality and potential pollution problems of erosion and sediment flushes and associated shock loads on treatment plants are significant and control rules are as yet poorly developed. Sewer corrosion and other gas related problems are also important, especially for H₂S, VOCs and odors.

Sediment management options. It is important to integrate watershed source management opportunities with in-sewer control and treatment plant and CSO operation. Source controls can be applies prior to and at entry to sewerage systems. These include best management practices (BMPs), problems of sanitary wastes and cultural habits which may be difficult to change. For example, reductions in water usage for the promoted of conservation and/or alternative options for sanitary waste disposal may lead to inadequate flows within sewers for traditional assumptions about self-cleansing performance.

There are new ideas for the structural design of sewers and ancillary components for the minimization of sediment problems. The use of recent research results in developing controlled sedimenting sewer designs (May 1995) is considered to be a major new design option. New research is needed in this area if design guidelines are to be developed (Bertrand-Krajewski, et al. 1995). Settling basins, varieties of tanks and overflow structures and innovatory screening systems are also available to minimize the introduction of solids into sewers. Operational measures such as flushing systems, balls, vane wagons and other cleaning methods are also available for flushing solids through the sewerage.

Future requirements and research needs. Ashley, et al. (1996) identified notable new developments in sewerage design, in addition to major research needs. These include:

· the concept of sewers as reactors,

· the interaction of solids with treatment plants,

· disposal of sewer solids,

· the interaction between gross solids and other sediments and options for their control,

· physical factors such as bed-forms in sewers and their effects,

· the ideal sewer shape, and

· proper determinations of particle settling velocity and particle size.

Increasing Capacity of Treatment and Sludge Handling Facilities

The design of new POTW should include treatment of CSO and not just treatment for peak dry weather flow conditions. Larger interceptors, higher treatment flowrates, and alternative highrate treatment methods should be used in new POTW designs (Field, et al. 1994). During construction of new facilities, many new opportunities are available, compared to retrofitting modifications to existing and outdated facilities. Some of these include specialized treatment unit operations that are capable of handling a wide range of flows, utilizing parallel processes to optimize treatment for widely varying flows, and using specialized high-rate processes for polishing effluent during high flow periods. There are many possible options for enhanced wet weather flow treatment at POTWs. Some of these are listed below (from Field, et al. 1994):

· POTW operational changes. Directing increased flows through primary settling tanks is usually the cheapest option for operating a treatment facility during increased wet weather flows. Generally, increased flows would decrease the performance of the settling tanks. However, when the normally untreated CSO is considered, significant improvements in pollutant discharges can usually be achieved, especially when considering the settling characteristics of wet weather flows that enable more effective settling compared to dry weather sanitary flows.

· Numerous modifications to settling tanks are also available to enhance wet weather performance. These include the use of dissolved air floatation, the use of lamella plates, and the possible use of chemical coagulants and polyelectrolytes.

· High-rate physical/chemical processes can also be used at POTWs during wet weather flows for enhanced treatment. These could be used as polishing units that would not normally be used during dry weather. Microscreens, polymer additions, coagulants with microsand and plate separators, plus deep-bed filters have all been shown to be highly effective when treating CSOs.

· Swirl degritters and deflection separators are also useful unit processes for combined sewage treatment that have not been used in separate sanitary sewage treatment.

An idealized WWF management system would include several attributes affecting the conveyance of the stormwater. Basic to these is an understanding of the different objectives of stormwater drainage systems, and the associated rainfall and runoff conditions. The following discussion assumes four major categories of rainfall characteristics and receiving water effects that should influence drainage design objectives.

Appendix A contains detailed examples from throughout the US, but the following comments are based on the long-term monitoring results from the Milwaukee, WI (Bannerman, et al. 1983) effort conducted during the Nationwide Urban Runoff Program (EPA 1983). During this monitoring period, there were two unusually large rains that occurred. The median rain, by count, was about 0.3 inches, while the rain associated with the median runoff quantity from typical medium density residential areas was about 0.75 inches. Therefore, more than half of the runoff was associated with rain events that were smaller that 0.75 inches. The largest storms (about 3 and 5 inches in depth) distort these values because, on average, the Milwaukee area only can expect one 3.5 inch storm every five years. If these large rains did not occur, such as for most years, then the significance of the small rains would be even greater. When the accumulative loadings of different pollutants (suspended solids, COD, phosphates, and lead) are compared, the runoff and discharge distributions are shown to be very similar and it is apparent that runoff volume is the most import factor affecting pollutant discharges.

As an example, rainfall and runoff distributions for Milwaukee can be divided into four groupings:

· <0.5 inch. These rains account for most of the events, but little of the runoff volume, and are therefore easiest to control. They produce much less pollutant mass discharges and probably have less receiving water effects than other rains. However, the runoff pollutant concentrations likely exceed regulatory standards for several categories of critical pollutants, especially bacteria and some total recoverable heavy metals. They also cause large numbers of overflow events in uncontrolled combined sewers. In most areas, runoff from these rains should be totally captured and either re-used for on-site beneficial uses or infiltrated in upland areas. These rains should be removed from the surface drainage system.

· 0.5 to 1.5 inches. These rains account for the majority of the runoff volume and produce moderate to high flows. The small rains in this category should also be removed from the drainage system and the runoff re-used on site for beneficial uses or infiltrated to replenish the lost groundwater infiltration associated with urbanization. The runoff from the larger rains should be treated to prevent pollutant discharges from entering the receiving waters.

· 1.5 to 3 inches. These rains produce the most damaging flows, from a habitat destruction standpoint, and occur every several months (at least once or twice a year). These recurring high flows, which were historically associated with much less frequent rains, establish the energy gradient of the stream and cause unstable streambanks. Typical storm drainage design events fall in the upper portion of this category. Extensive pollution control designed for these events would be very costly, especially considering the relatively small portion of the annual runoff associated with the events. However, discharge rate reductions are important to reduce habitat problems in the receiving waters. The infiltration and other treatment controls used to handle the smaller storms in the above categories would have some benefit in reducing pollutant discharges during these larger, rare storms.

· >3 inches. The smallest rains in this category are included in design storms used for drainage systems in Milwaukee. These rains occur only rarely (once every several years to once every several decades, or less frequently) and produce extremely large flows. The monitoring period during the Milwaukee NURP program was unusual in that two of these events occurred. These storms, while very destructive, are sufficiently rare that the resulting environmental problems do not justify the massive controls that would be necessary for their reduction. The problem during these events is massive property damage and possible loss of life. These rains typically greatly exceed the capacities of the storm drainage systems, causing extensive flooding. It is critical that these excessive flows be conveyed in "secondary" drainage systems. These secondary systems would normally be graded large depressions between buildings that would direct the water away from the buildings and critical transportation routes and to possible infrequent/temporary detention areas (such as large playing fields or parking lots). Because these events are so rare, institutional memory often fails and development is allowed in areas that are not indicated on conventional flood maps, but would suffer critical flood damage.

There are many questions that remain concerning the "best" wet weather flow drainage and treatment systems that should be used in newly developing areas. Of course, there is no one "best" answer for all areas and conditions. A wide variety of options exist and an engineer must select from these depending on numerous site specific situations. In most cases, conventional separate sanitary wastewater and stormwater drainage systems would seem most appropriate. However, these systems have shown to be of reduced value in many cases. The most significant problems relate to the large amount of inflow and infiltration (I/I) occurring in separate sanitary wastewater systems and the lack of stormwater pollution controls in separate stormwater systems. Pertroff (1996) estimated that more than half of the annual flows treated by municipal wastewater treatment plants are from I/I. In addition, I/I is likely the major cause of SSOs in separate sanitary wastewater collection systems. Therefore, in order for separate sanitary wastewater collection systems to be effective in the future, they must be constructed to eliminate almost all I/I contributions. This is possible, as demonstrated by current vacuum and pressurized sanitary wastewater collection systems.

The municipal representatives are the real experts of the current systems and present conservative viewpoints because they will most likely be responsible for operations of drainage systems in the future. The following are some of their concerns and predictions for the future concerning urban drainage issues:

· We must start with existing systems and make slow and gradual changes.

· Future citizens will be better educated and will be willing to make life style changes that will reduce

wastewater discharges.

· We will still have centralized wastewater treatment systems in the future because of better hygienic,

health, energy, and environmental benefits, compared to de-centralized systems.

· Stormwater will be eliminated from sewerage in the future, increasing capacity for sanitary wastewater.

· I/I will be reduced considerably due to new methods of detection and prevention.

· There will be more rigid restrictions on the use of materials to prevent corrosion problems.

· Multi-disciplinary/integrated planning in urban areas will be more widespread, with clear strategies for

operations. Relationships between precipitation, sewerage, treatment facilities, and receiving waters will be

better considered.

· Urban drainage will become better integrated with other technical aspects of the infrastructure.

· Reuse of stormwater and treated wastewaters should be promoted where necessary (dual water systems,

with degraded water available for less critical uses for example). Don’t rely on highly purified domestic

water for all uses.

· There was no consensus for the uniform use of either combined or separate systems in the future.

Regulators stressed the need to live within the carrying capacity of the planet (water, food, housing, and industry). The central focus here was on water quantity and quality and the need to enhance water resources in the broadest context, such as at planet, country, catchment, community, and citizen levels. The principles of ideal regulations for urban drainage include the following:

· Self regulation is preferred. Too much regulation stifles innovation.

· Regulations must be balanced against risk.

· Only regulate that which is not managed in other ways.

· Good legislation is the least amount. Financial support and positive enforcement is needed most.

However, effective punishment is also needed.

· Related resources (air, land, and water) should be regulated in one agency.

· Regulatory consistency, not uniformity, is needed most.

· Must have appropriate time scales for action considering needed planning.

· Education is the key component of what regulators should do. Designers are a key group for education.

They should be linked with citizens for political and financial support. Politicians are short-term and

typically have few long-term goals. Polluters need to know the objectives and problems.

The planners felt there must be a better agreement between all parties on the definition of sustainability. Planners encouraged the need to move away from urban stormwater management by drains and towards urban waterways. They also felt there are better ways to manage stormwater pollutants besides transport of the pollutants by water. Other issues that the planners brought up included:

· Much more effort should be spent on source control (prevention) than on treatment (cure).

· Emphasis should be placed on keeping stormwater on site instead of transporting it downstream.

· Soil characteristics need just as much consideration as transportation elements when selecting sites for

new development.

· The planning for urban development should be holistic by integrating water supply and drainage, for

example. Currently, the developer does the planning.

· Only a small portion of the total domestic water needs require the highest quality water. Reuse of gray

water on site, plus storage of stormwater for use on site needs to be considered.

· Greater emphasis should be placed on increasing density of urban development and making high density

areas more comfortable, in order to preserve more open space.

· A multi-disciplinary approach in planning is critically needed. Developers and citizens should be brought

together to examine new development scenarios.

· Better communication is needed between planners, developers, citizens, and politicians.

· Improved building techniques and materials are needed.

· Must convince politicians of the importance of long-term goals.

· Catchment planning is needed to increase building density in order to decrease impervious density.

The lack of a universal definition for sustainability was recognized by the researchers and consultants. Many local considerations make a universal definition impractical. However, there are many acceptable criteria for sustainability; the most basic being that sustainable actions would be acceptable over long periods of time. The urban area needs to consider both the built-up area plus the surrounding natural area. Similarly, the urban water cycle needs to consider water supply, stormwater, and sanitary wastewater together. Guiding principles of sustainable urban water resources include the following:

· Water is renewable on a large scale. We can have sustainable use of water if we are careful.

· We must accept multiple objectives and use a multi-disciplinary approach.

· Source control (especially pollution prevention) should be a top priority.

· We must not transport our problems downstream.

Technological aspects of the sustainability of urban stormwater resources include:

· "Best management practices" (BMPs) are not yet proven to be sustainable (functionally or economically).

· BMPs are more sustainable in new growth areas.

· It is barely possible to counterbalance new problems related to new growth if we impose high levels of

effective controls in areas of new development, and simultaneously use high levels of retro-fitted controls

in existing areas. It will be difficult to improve or fix existing problems with existing resources.

· Retro-fitting is possible, but much less effective and much more expensive than using controls in new

development.

· Combined sewers will eventually function adequately.

· Future urban drainage approaches are not likely to change radically or quickly.

· Urbanization will continue in a manner similar to recent trends.

· There will be a gradual acceptance of source control of stormwater pollution.

· The urban water cycle may eventually include: bottled water for all consumptive uses, piped water for

cooking and water contact, and recycled graywater and stormwater for other uses (such as irrigation and

toilet flushing).

The following list indicates some likely effective wastewater collection scenarios for several different conditions for the future:

· low and very low density residential developments (<2 acre lot sizes). Sanitary wastewater should be treated on site using septic tanks and advanced on-site treatment options. Domestic water conservation to reduce sanitary wastewater flows should be an important component of these systems. Most stormwater should be infiltrated on site by directing runoff from paved and roof areas to small bio-retention areas. Disturbed soil areas should use compost-amended soils and should otherwise be constructed to minimize soil compaction. Roads should have grass swale drainage to accommodate moderate to large storms.

· medium density developments (¼ to 2 acre lot sizes). Separate sanitary wastewater and stormwater drainage systems should be used. Sanitary wastewater collection systems must be constructed and maintained to eliminate I/I, or use vacuum or pressurized conveyance systems. Again, most stormwater should be infiltrated on site by directing runoff from paved and roof areas to small bio-retention areas. Paved areas should be minimized and the use of porous pavements and paver blocks should be used for walkways, driveways, overflow parking areas, etc. Disturbed soil areas should use compost-amended soils and should otherwise be constructed to minimize soil compaction. Grass swale drainages should be encouraged to accommodate moderate to large storms for the excess runoff in residential areas, depending on slope, soil types, and other features affecting swale stability. Commercial and industrial areas should also use grass swales, depending on groundwater contamination potential and available space. Wet detention ponds should be used for controlling runoff from commercial and industrial areas. Special controls should be used at critical source areas that have excessive pollution generating potential.

· high density developments. Combined sewer systems could be effectively used in these areas. On-site infiltration of the least contaminated stormwater (such as from roofs and landscaped areas) is needed to minimize wet weather flows. On-site storage of sanitary wastewaters during wet weather (using Preul’s CSPS), plus extensive use of in-line and off-line storage, and the use of effective high-rate treatment systems would minimize the damage associated with any CSOs. The treatment of the wet weather flows at the wastewater treatment facility would likely result in less pollutant discharges in these areas than if conventional separate wastewater collection systems were used.

There are many comprehensive sources of actual monitored stormwater characteristics that can be used to evaluate design procedures. The US EPA’s Nationwide Urban Runoff Program (EPA 1983) and the EPA’s Urban- Rainfall-Runoff-Quality Data Base (Heaney, et al. 1982) contain much information, for example, and much of that data can be downloaded from the Internet at: http://www.eng.ua.edu/~awra/download.htm

Most of the EPA’s "Data Base" is for 2 locations in Broward County, FL; 1 site in Dade County, FL; 2 sites in Salt Lake City, UT; and 2 sites in Seattle, WA. Most of the data were obtained during the 1970s. These sites had the best representation of data of interest for these analyses and the sites were well described. Parameters examined included simultaneous rainfall and runoff depths, plus peak rain and flow rates. The following plots were prepared using this data:

· runoff depth versus rainfall,

· volumetric runoff coefficient (Rv) versus rainfall,

· NRCS curve number (CN) versus rainfall, and

· ratio of reported peak flow/peak rainfall versus rainfall.

In a similar manner, information from the EPA’s NURP program (EPA 1983) was also investigated. A wider variety of information was collected during NURP, enabling additional relationships examining stormwater quality. Most of the data examined is from 5 sites in Champaign, IL; 2 sites in Austin, TX; 5 sites in Irondequoit Bay, NY; 1 site in Rapid City, SD; plus additional observations from Tampa, FL, Winston Salem, NC, and Eugene and Springfield, OR. Most of this data were obtained during the early 1980s and was subjected to rigorous quality control. Besides the four plots listed above, the following plots were also constructed examining potential water quality concentration relationships:

· total suspended solids concentration versus rainfall,

· COD concentration versus rainfall,

· phosphorous concentration versus rainfall,

· lead concentration versus rainfall,

· peak flow/peak rain versus rainfall, and

· peak flow rate versus peak rain intensity.

The basic rainfall versus runoff plots (Figure A-1) were made to indicate the smoothness of this basic relationship. A large scatter instead of a smooth curve may indicate measurement errors or uneven rainfalls over the catchment, or highly variable infiltration characteristics (due to changing soil moisture before the different rains). As shown on these plots, the runoff depth increases with increasing rain. However, several plots do show substantial scatter, mostly for sites having relatively small runoff yields. In addition, in some cases, more runoff was observed than could be accounted for by the rain. Errors in these measurements may be significant and would vary for the different sites. The author of this module was involved in several of the monitoring projects that are included in these analyses, and also served on EPA technical committees overseeing others. In addition, I have many years experience in monitoring these parameters in many locations and recognize many of the past problems and current attempts to correct them. The following list therefore shows possible measurement errors that may have affected this data:

· variable rainfall over a large test catchment that was not well represented by enough rain gages

(Although several of the test catchments had multiple rain gages, most did not, and few were

probably frequently re-calibrated in the field.),

· poorly calibrated monitoring equipment (Many flow monitoring equipment relied on using the

Manning’s equation in pipes, with assumed roughness coefficients, without independent calibration,

while other monitoring locations used calibrated insert weirs.)

· transcription errors (Many of these older monitoring activities required manual transfer from field

equipment recorders to computers for analysis. In many cases, obvious "factor of ten" errors were

made, for example.),

· newly developed equipment was used that had not been adequately tested, and

· difficult locations in the sewerage or streams that were monitored.

It is expected that the measurement errors were probably no less than about 25% during these monitoring activities.

The plots of rainfall versus the volumetric runoff coefficient plot (Figure A-2) shows the ratio of the runoff volume, expressed as depth for the watershed, to rain depth, or the Rv, for different rain depths. This is a related plot to the one described above. If the Rv ratio was constant for all events, the rainfall versus runoff depth plot described above, would indicate a straight diagonal line, with no scatter. It is typically assumed that the above described relationship would indicate increasing Rv values as the rain depth increased. Figure A-1 shows a slight upwards curve with increasing rain depths. This is due to the rainfall losses making up smaller and smaller portions of the total rainfall as the rainfall increases, with a larger fraction of the rainfall occurring as runoff. The plot of Rv versus rainfall (Figure A-2) would therefore show an increasing trend with increasing rain depth. In most cases, the plots of actual data indicate a large (random?) scatter, making the identification of a trend problematic. The use of a constant Rv for all rains may also be a problem because of the large scatter. In many cases, the long-term average Rv for a residential area may be close to the typically used value. In Figure A-2, the values appear to center about 0.2 (somewhat smaller than the typically used value of about 0.3 for medium density residential areas), but the observed Rv values may range from lows of less than 0.04 to highs of greater than 0.5, especially for the smallest rains. The small rains probably have the greatest measurement errors, as the rainfall is much more variable for small rains than for larger rains, plus very low flows are difficult to accurately measure. Obviously, understanding what may be causing this scatter is of great interest, but is difficult because of measurement errors masking trends that may be present. In many cases, using a probability distribution to describe this variation may be the best approach.

Figure A-3 is a plot of the NRCS curve number (CN) versus rainfall depth (SCS 1986). The NRCS assumes that the CN is constant for all rain depths for a specific site. However, they specify several limitations, including:

· the CN method is less accurate when the runoff is less than 0.5 inch. It is suggested that an

independent procedure be used for confirmation,

· the CN needs to be modified according to antecedent conditions, especially soil moisture before an

event, and

· the effects of impervious modifications (especially if they are not directly connected to the drainage

path) needs to be reflected in the CN.

Few of these warnings are considered by most storm drainage designers, or by users of NRCS CN procedures for stormwater quality analyses. Figure A-3 shows the typical pattern obtained when plotting CN against rain depth. The CN for small rain depths is always very large (approaching 100), then it decreases as the rain depth increases. At some point, the observed CN values equal the NRCS published recommended CN. During rains smaller than this matching point, the actual CN is greater than the NRCS CN. Predicted runoff depths would therefore be much less than the observed depths during these rains. Very large differences in runoff depths are associated with small differences in CN values, making this variation very important.

Figure A-4 shows the observed peak runoff flow rate versus the peak rain intensity. If the averaging period for the peak flows and peak rain intensities were close to the catchment time of concentration (t_c), the slope of this relationship would be comparable to the Rational coefficient (C). The averaging times for the peak values probably ranged from 5 minutes to 1 hour for the different projects. Unfortunately, this averaging time period was rarely specified in the data documentation. Most urban area t_c values probably range from about 5 to 15 minutes. As indicated in this figure, the relationship between these two parameters shows a general upward trend, but it would be difficult to fit a statistically valid straight line through the data. As noted above for the other two drainage design procedures, actual real-world variations (coupled to measurement errors) add a lot of variation to the predicted runoff flow and volume estimates. Most drainage designers do not consider the actual variations that may occur.

Figure A-5 shows an example plot of the ratio of the peak runoff flow rate to the average runoff flow rate versus rain depth. These values can be used to help describe the shape of simple urban area hydrographs. If the hydrograph can be represented by a simple triangular hydrograph, then the peak flow to average flow ratio must be close to 2. As shown on these figures, this ratio is typically substantially larger than 2 (it can never be less than 1 obviously), indicating the need to use a somewhat more sophisticated hydrograph shape (such as a double triangular hydrograph that can consider greater flows). These plots indicate if this ratio can be predicted as a function of rain depth. In most cases, values close to 2 are seen for the smallest rains, but they ratio increases to 5, or more, fairly quickly, but with much variability.

Stormwater Receiving Water Problems

Reviews of numerous urban receiving water studies from throughout the U.S. have identified the following diverse list of receiving water problems that may be caused by stormwater (Pitt 1995):

· Sedimentation damage in stormwater conveyance systems and in receiving waters.

· Nuisance algae growths from nutrient discharges into quiescent waters.

· Inedible fish and undrinkable water caused by toxic pollutant discharges.

· Shifts to less sensitive aquatic organisms caused by contaminated sediments and habitat destruction.

· Property damage from increased drainage system failures.

· Swimming beach closures from pathogenic microorganisms.

· Water quality violations, especially for bacteria and total recoverable heavy metals.

The first four problem areas are mostly associated with slug (mass) discharges (not instantaneous concentrations or rates), while the last three are mostly associated with instantaneous concentrations and high flow rates.

In order to predict receiving water problems caused by stormwater, accurate flow estimates and pollutant mass discharges must be known. Knowing where the potentially problem pollutants originate in the watershed is also valuable in order to select appropriate stormwater control candidates. Accurate knowledge of runoff volumes during different storms has been shown to be necessary when predicting pollutant discharges.

Most of the Annual Rain is Associated With Many Small Individual Events. This discussion reviews actual monitored rainfall and runoff distributions for Milwaukee, WI (data from Bannerman, et al. 1983), and examines long-term rainfall histories and predicted runoff from 24 locations throughout the U.S. The Milwaukee observations show that southeastern Wisconsin rainfall distributions can be divided into the following categories, with possible management approaches relevant for each category of rain:

· Common rains having relatively low pollutant discharges are associated with rains less than about

0.5 in. (12 mm) in depth. These are key rains when runoff-associated water quality violations, such as for bacteria, are of concern. In most areas, runoff from these rains should be totally captured and either re-used for on-site beneficial uses or infiltrated in upland areas. For most areas, the runoff from these rains can be relatively easily removed from the surface drainage system.

· Rains between 0.5 and 1.5 in. (12 and 38 mm) are responsible for about 75% of the runoff pollutant discharges and are key rains when addressing mass pollutant discharges. The small rains in this category can also be removed from the drainage system and the runoff re-used on site for beneficial uses or infiltrated to replenish the lost groundwater infiltration associated with urbanization. The runoff from the larger rains should be treated to prevent pollutant discharges from entering the receiving waters.

· Rains greater than 1.5 in. (38 mm) are associated with drainage design and are only responsible for relatively small portions of the annual pollutant discharges. Typical storm drainage design events fall in the upper portion of this category. Extensive pollution control designed for these events would be very costly, especially considering the relatively small portion of the annual runoff associated with the events. However, discharge rate reductions are important to reduce habitat problems in the receiving waters. The infiltration and other treatment controls used to handle the smaller storms in the above categories would have some benefit in reducing pollutant discharges during these larger, rarer storms.

· In addition, extremely large rains also infrequently occur that exceed the capacity of the drainage system and cause local flooding. Two of these extreme events were monitored in Milwaukee during the Nationwide Urban Runoff Program (NURP) project (EPA 1983). These storms, while very destructive, are sufficiently rare that the resulting environmental problems do not justify the massive stormwater quality controls that would be necessary for their reduction. The problem during these events is massive property damage and possible loss of life. These rains typically greatly exceed the capacities of the storm drainage systems, causing extensive flooding. It is critical that these excessive flows be conveyed in "secondary" drainage systems. These secondary systems would normally be graded large depressions between buildings that would direct the water away from the buildings and critical transportation routes and to possible infrequent/temporary detention areas (such as large playing fields or parking lots). Because these events are so rare, institutional memory often fails and development is allowed in areas that are not indicated on conventional flood maps, but would suffer critical flood damage.

 

Obviously, the critical values defining these rain categories are highly dependent on local rain and development conditions. Computer modeling analyses from several representative urban locations from throughout the U.S. are presented in this paper. These modeled plots indicate how these rainfall and runoff probability distributions can be used for more effective storm drainage design in the future. In all cases, better integration of stormwater quality and drainage design objectives will require the use of long-term continuous simulations of alternative drainage designs in conjunction with upland and end-of-pipe stormwater quality controls. The complexity of most receiving water quality problems prevents a simple analysis. The use of simple design storms, which was a major breakthrough in effective drainage design more than 100 years ago, is not adequate when receiving water quality issues must also be addressed.

This discussion also reviews typical urban hydrology methods and discusses common problems in their use in predicting flows from these important small and moderate sized storms. A general model is then described, and validation data presented, showing better runoff volume predictions possible for a wide range of rain conditions.

Figure A-10 includes cumulative probability density functions (CDFs) of measured rain and runoff distributions for Milwaukee during the 1981 NURP monitored rain year (data from Bannerman, et al. 1983). CDFs are used for plotting because they clearly show the ranges of rain depths responsible for most of the runoff. Rains between 0.05 and 5 in. were monitored during this period, with two very large events (greater than 3 inches) occurred during this monitoring period which greatly distort these curves, compared to typical rain years. The following observations are evident:

· The median rain depth was about 0.3 in.

· 66% of all Milwaukee rains are less than 0.5 in. in depth.

· For medium density residential areas, 50% of runoff was associated with rains less than 0.75 in.

· A 100-yr., 24-hr rain of 5.6 in. for Milwaukee could produce about 15% of the typical annual runoff volume, but it only contributes about 0.15% of the average annual runoff volume, when amortized over 100 yrs.

· Similarly, a 25-yr., 24-hr rain of 4.4 in. for Milwaukee could produce about 12.5% of the typical annual runoff volume, but it only contributes about 0.5% of the average annual runoff volume, when amortized over 25 yrs.

Figure A-11 shows CDFs of measured Milwaukee pollutant loads associated with different rain depths for a medium density residential area. Suspended solids, COD, lead, and phosphate loads are seen to closely follow the runoff volume CDF shown in Figure A-10, as expected. Since load is the product of concentration and runoff volume, some of the high correlation shown between load and rain depth is obviously spurious. However, these overlays illustrate the range of rains associated with the greatest pollutant discharges.

The monitored rainfall and runoff distributions for Milwaukee show the following distinct rain categories:

· <0.5 inch. These rains account for most of the events, but little of the runoff volume, and are therefore easiest to control. They produce much less pollutant mass discharges and probably have less receiving water effects than other rains. However, the runoff pollutant concentrations likely exceed regulatory standards for several categories of critical pollutants, especially bacteria and some total recoverable metals. They also cause large numbers of overflow events in uncontrolled combined sewers. These rains are very common, occurring once or twice a week (accounting for about 60% of the total rainfall events and about 45% of the total runoff events that occurred), but they only account for about 20% of the annual runoff and pollutant discharges. Rains less than about 0.05 inches did not produce noticeable runoff.

· 0.5 to 1.5 inches. These rains account for the majority of the runoff volume (about 50% of the annual volume for this Milwaukee example) and produce moderate to high flows. They account for about 35% of the annual rain events, and about 20% of the annual runoff events. These rains occur on the average about every two weeks during the spring to fall seasons and subject the receiving waters to frequent high pollutant loads and moderate to high flows.

· 1.5 to 3 inches. These rains produce the most damaging flows, from a habitat destruction standpoint, and occur every several months (at least once or twice a year). These recurring high flows, which were historically associated with much less frequent rains, establish the energy gradient of the stream and cause unstable streambanks. Only about 2 percent of the rains are in this category and they are responsible for about 10 percent of the annual runoff and pollutant discharges.

· >3 inches. This category is rarely represented in field studies due to the rarity of these large events and the typically short duration of most field observations. The smallest rains in this category are included in design storms used for drainage systems in Milwaukee. These rains occur only rarely (once every several years to once every several decades, or less frequently) and produce extremely large flows. The 3-year monitoring period during the Milwaukee NURP program (1980 through 1983) was unusual in that two of these events occurred. Less than 2 percent of the rains were in this category (typically <<1% would be), and they produced about 15% of the annual runoff quantity and pollutant discharges. During a "normal" period, these rains would only produce a very small fraction of the annual average discharges. However, when they do occur, great property and receiving water damage results. The receiving water damage (mostly associated with habitat destruction, sediment scouring, and the flushing of organisms great distances downstream and out of the system) can conceivably naturally recover to before-storm conditions within a few years.

These rainfall and pollutant mass distributions are not unique for Milwaukee. Appendix B of this report contains many examples of similar plots of monitored rainfall, runoff, and pollutant mass distributions for other NURP projects from throughout the country (including 5 sites in Champaign, IL; 2 sites in Austin, TX; 5 sites in Irondequoit Bay, NY; 1 site in Rapid City, SD; plus additional observations from Tampa, FL, Winston Salem, NC, and Eugene and Springfield, OR).

In addition, long-term continuous simulations were made using SLAMM (incorporating the small storm hydrology components described in this report section) for 22 representative locations from throughout the U.S. (Figure A-12). These locations represent most of the major river basins and much of the rainfall variations in the country. These analyses are only intended to show the importance of these smaller rains for many different regions and conditions. They are not intended to be used for design purposes. As noted earlier, the recommended approach for design is to continuously model long rain records for site specific conditions. These locally derived runoff distributions, reflecting site conditions and actual rains, can then used for evaluating alternative drainage and water quality designs.

These simulations were based on 5 to 10 years of rainfall records, usually containing about 500 individual rains. The rainfall records were from certified NOAA weather stations and were obtained from CD-ROMs distributed by EarthInfo of Boulder, CO. Hourly rainfall depths for the indicated periods were downloaded from the CD-ROMs into an Excel spreadsheet. The files were slightly modified (by eliminating the daily total rainfall column) and saved as a comma delineated file. This file was then read by an utility program included in the SLAMM package. This rainfall file utility combined adjacent hourly rainfall values into individual rains, based on user selections (at least 6 hrs of no rain was used to separate adjacent rain events and all rain depths were used, with the exception of the "trace" values). These rain files for each city were then used in SLAMM for typical medium density and strip commercial developments. The outputs of these computer runs were then plotted as shown on Figure A-13.

Table A-1 summarizes these rain and runoff distributions for different U.S. locations, while Figures A-14 through A-19 plot some of the important values on a U.S. map. Lower and upper runoff distribution breakpoints were identified on all of the individual distributions. The breakpoints separate the distributions into the following three general categories:

· less than lower breakpoint: small, but frequent rains. These generally account for 50 to 70 percent of all rain events (by number), but only produce about 10 to 20 percent of the runoff volume. Figure A-15 shows that the rain depth for this breakpoint ranges from about 0.10 in. in the Southwest arid regions of the country, to about 0.5 in. in the wet Southeast. These events are most important because of their frequencies, not because of their mass discharges. These rains are therefore of great interest where water quality violations associated with urban stormwater occur. This would be most common for bacteria (especially fecal coliforms) and for total recoverable heavy metals which typically exceed receiving water numeric criteria during practically every rain event in heavily urbanized drainages having separate stormwater drainage systems.

· between the lower and upper breakpoint: moderate rains. These rains generally account for 30 to 50 percent of all rains events (by number), but produce 75 to 90 percent of all of the runoff volume (Figure A-19). Figure A-17 shows that the rain depths associated with the upper breakpoint range from about 1 to 2 in. in the arid parts of the U.S. to up to 5 or 6 in. in wetter areas. As shown earlier for actual monitored events Milwaukee and elsewhere, as shown in Appendix B, these runoff volume distributions are approximately the same as the pollutant distributions. Therefore, these intermediate rains also account for most of the pollutant mass discharges and much of the actual receiving water problems associated with stormwater discharges.

· above the upper breakpoint: large, but rare rains. These rains include the typical drainage design events and are therefore quite rare. During the period analyzed, many of the sites only had one or two, if any, events above this breakpoint. These rare events do account for about 5 to 10 percent of the runoff on an annual basis, as shown on Figure A-18. Obviously, these events must be evaluated to ensure adequate drainage.

Because of the importance of these small and moderate rains, it is important to review typically used urban hydrology methods that have been commonly used to predict runoff from urban areas. These tools have been reasonably successful when evaluating drainage capacity for large "design storm" events. However, the following paragraphs will indicate their short-comings when used for evaluating the common smaller events. A general urban runoff model is also presented that has been shown to be useful to predict runoff volumes for a wide range of rain events, especially the small and moderate rains of greatest interest in water quality evaluations.

Varying Contributing areas are Important in Urban Hydrology. Figure A-21 shows the components of a hypothetical hydrograph for an urban area. For small rains, most of the runoff observed at the outfall originates from street surfaces and other directly connected impervious areas. However, as the rain depth increases, runoff from pervious areas become important. The critical problem is being able to predict when these component areas contribute significant runoff volumes (and pollutants). SLAMM (Pitt 1986 and 1992) was developed to enable predictions of runoff contributions (and source area controls), using a simplified urban hydrology approach appropriate for important small rains.

Observed Runoff Volumes Do Not Compare Well With Commonly Used Urban Runoff Prediction Methods. Some of the most commonly used stormwater design methods utilizes the NRCS curve number (CN) method, especially TR-20 and TR-55 (SCS 1986). The NRCS recommends against the use of the curve number procedure for rains less than one-half inch. Unfortunately, this warning is ignored in many urban runoff models that have been developed. As shown previously, small rains are very significant when analyzing urban runoff. In addition, the NRCS recommends that the curve number method should be used for individual components of the drainage area, if CN values differ by more than 5, instead of using a composite CN for the complete area. Unfortunately, many users of the CN method ignore these two basic warnings, and many urban stormwater models use composite CN values for all storms. The CN method is a suitable tool if properly used, unfortunately, it is frequently used for small storms and for water quality evaluations, well beyond its intended use addressing drainage design for conveyance objectives for large rains.

Figure A-22a shows rainfall-runoff plots for eight monitored areas in Milwaukee. The curve is similar to the US. Natural Resources Conservation Service (NRCS) curve number (CN) rainfall-runoff plot contained in TR-55 (SCS 1986). This figure also shows the NRCS CN values calculated using actual P (precipitation) and Q (runoff quantity) data. CNs vary greatly with rain depth.

The EPA report shows numerous similar plots for other monitored locations from throughout the U.S., collected during the EPA’s NURP projects in the early 1980s (EPA 1983), and from the EPA’s rainfall-runoff-quality data base (Huber, et al. 1982). Figures A-23 through A-26 contain CN versus rain depth plots for many of these cities, including: 2 locations in Broward County, FL; 1 site in Dade County, FL; 2 sites in Salt Lake City, UT; and 2 sites in Seattle, WA (from the rainfall-runoff-quality data base), plus 4 sites in Champaign, IL; 5 sites in Irondequoit Bay, NY; 2 sites in Austin, TX; and 1 site in Rapid City, SD (from the NURP data). Figure A-23 contains plots for areas with little urbanization, Figure A-24 contains plots for medium density residential areas and mixed common urban areas, Figure A-25 contains plots for high density and commercial areas, and Figure A-26 contains plots for catchments having only major roadways. In all cases, the general pattern is the same: observed curve numbers are all very high for small rains, tapering off as the rains become large. All of the test watersheds are typical for these land uses and do not contain any unusual drainage designs or stormwater controls.

Table A-2 is a summary of these observed curve numbers at several different rain depths, compared to typical curve numbers presented by the NRCS (SCS 1986) for these land uses. Several of the sites had adequate descriptions to enable curve numbers to be estimated, based on their directly connected impervious areas and soil texture. The following list shows these sites, with the NRCS recommended curve numbers, and the approximate rain depth where these curve numbers were observed:

· Broward Co., FL, residential land use (40% imperv., with sandy soils). NRCS CN = 61, observed at about 3.5 in. of rain.

· Champaign-Urbana, IL, single family residential land use (18% imperv., with silty, poorly drained soils). NRCS CN = 84, observed at about 1.2 in. of rain.

· Champaign-Urbana, IL, single family residential land use (19% imperv., with silty, poorly drained soils). NRCS CN = 84, observed at about 1.2 in. of rain.

· Dade Co., FL, high density residential land use (almost all impervious, "D" soils). NRCS CN = 92,

observed at about 1.3 in. of rain.

· Champaign-Urbana, IL, commercial land use (40% imperv., with silty and poorly drained soils). NRCS

CN = 87, observed at about 1.1 in. of rain.

· Champaign-Urbana, IL, commercial land use (55% imperv., with silty and poorly drained soils). NRCS

CN = 91, observed at about 0.8 in. of rain.

· Broward Co., FL, transportation catchment (54% imperv., with sandy soils). NRCS CN = 73, observed at

about 1.7 in. of rain.

· Salt Lake City, UT, roadway land use (mostly paved, sandy loam). NRCS CN = 89, observed at about 0.3

in. of rain.

· Salt Lake City, UT, transportation catchment (imperv. Raods, clay loam). NRCS CN = 95, observed at

about 0.15 in. of rain.

For the rains less than the matching point (rain depth where the NRCS recommended CN was observed), the actual CN is larger than the recommended CN and the predicted runoff using the NRCS methods would be less than actually occurred. Similarly, for rains larger than the matching point, the actual CN is smaller than the recommended CN and the predicted runoff using the NRCS CN method would be greater than actually occurred. The magnitude of the runoff differences varies greatly, depending on the CN values and the rain depth. As an example, if the recommended NRCS CN was 84, but the actual CN was really 98 for a 0.2 in. rain (similar to the Champaign, IL, medium density residential sites), the percentage error is infinite. For a 1 in. rain, the actual CN at this site was about 86 and the recommended NRCS value remains at 84. The difference now is much smaller, as the rain depth being examined is close to the matching point depth of 1.2 inches. If the rain depth of concern was much larger, say 3 inches, the errors would be in the other direction, as summarized below:

The overall annual runoff depth error associated with using the NRCS recommended CN method depends on the frequency of rains having the different errors. Because the matching point rainfall depths are close to the rain depth associated with the median runoff depth, as shown previously on A-1, the annual errors may be within reason. However, the errors associated with individual events, and for the three classes of rain depths described earlier, are likely very large. This is a significant problem with stormwater quality management where accurate representations of the sources of the runoff are needed in order to evaluate control practices and development options. If the relative sources of the runoff flows are in great error, inappropriate and wasteful expenditures are likely.

Actual Volumetric Runoff Coefficients (Rv) Vary With Storm Size. Figure A-27 shows how the volumetric runoff coefficients (the ratio of runoff depth to rainfall depth) change with rain depth. After subtracting initial abstractions, continuous losses can be assumed to be mostly infiltration. After a sufficient amount of rain has occurred, all losses have been satisfied. Each unit increase in rain then results in a unit increase in runoff volume.

Small rain depths result in runoff that have small Rv values. As the rain depth increases, the Rv increases. Rv values are only "constant" over a small range in rain depths. During many urban runoff monitoring projects, only small ranges of rains are typically represented. Therefore, "averaged" Rv values are incorrectly used with the understanding that they are useful over a wider range than justified. Appendix B includes rainfall-runoff plots and Rv-rainfall plots for many locations throughout the U.S. Few of these plots are as smooth as indicated for the Milwaukee data. The NURP data was collected in the early 1980s, while the rainfall-runoff-quality data base information was collected much earlier. There was significant variation in the accuracies of monitoring rainfall and runoff for the different locations. This is most evident at test sites having large amounts of directly connected pavement. Many of the measured runoff events had greater runoff volumes than the measured rainfall volumes (Rv values greater than 1.0 and calculated CN values greater than 100). This of course cannot occur in the absence of other flow sources and was likely associated with random measurement errors. The best measurements were probably made with errors approaching 25%, while some test sites used newly available equipment and errors may have been greater. These errors are much more obvious at high density and commercial sites than at the more commonly monitored medium density residential sites.

Figure A-28 shows a plot of runoff depth versus rain depth and another plot of the NRCS CN versus rain depth for a set of artificial rain and runoff data. These plots were prepared to visually show the relationship between Rv and NRCS CN values. If the data has relatively constant Rv values for all rains, as indicated for most of the sites shown in Appendix B, the CN plots will naturally decrease substantially with increasing rain depth (again, as indicated in almost all of the measured data). It is interesting to note that the calculated NRCS CN is always very close to 100 for very small rain and runoff values, irrespective of the Rv ratio. The Rv values likely increase with increasing rain depth, which is evident if the observations can be obtained with small measurement errors and if the range of rains observed is large. Flow and rainfall measurement errors are much more obvious on the Rv plots, especially for the small rains, than on the CN plots.

Runoff Process for Paved Surfaces

When rain falls on an impervious surface, much of it will flow off the surface and contribute to the total urban runoff. With the exception of infiltration, these losses are mostly associated with the initial portions of the rain and are termed initial abstractions. Water may also infiltrate through pavement, or through cracks or seams in the pavement. For small rains, a much greater portion of the rain will be lost to these runoff loss processes than for large rains.

Paved surfaces are usually considered impervious, implying no infiltration. However, some researchers have concluded that paved surfaces do indeed experience infiltration losses. Falk and Niemczynowicz (1978) found that smooth paved surfaces had lower infiltration losses, compared to poorly maintained surfaces which had losses of about 7 percent of the total rain. Pratt and Henderson (1981) were asked after their presentation at the Second International Conference on Urban Storm Drainage if the variation of the runoff coefficient that they observed for pavement could be due to infiltration through the surface which is commonly considered to be zero. They agreed that this variation was likely due to the difference in the permeability of the "impervious" catchment surfaces. They found that gaps between concrete sections in the curbs and gutters were the principal means of runoff losses. Willeke (1966) found that cracks in gutters could allow significant amounts of water to infiltrate, especially if sandy soils underlaid concrete. Davies and Hollis (1981) found an average runoff loss from a paved road surface to be about 85 percent of the rain depth. This loss was considered about evenly divided between detention storage and infiltration through the pavement, especially through cracks in the gutter. Cedergren (1974) measured infiltration rates through typical "sealed" seams of about 20 mm per hour (with pavement seams located about every 8 meters).

Infiltration of Rain Water Through Pavement Can be a Substantial Portion of the Total Rain for Most Events. Initial abstractions are dependent of pavement texture and slope, while infiltration is dependent on pavement porosity and pavement cracks. Pavement is relatively porous. It is the pavement base course that is much more resistant to percolation. Infiltrated water is therefore forced to flow laterally towards the pavement edges. If the flow path is long, then the resulting infiltration is limited. Figure A-29 is an example from a typical pavement runoff test (Pitt 1987). Initial abstractions may be about 1 mm for pavement, while the total infiltration may be between 5 and 10 mm. The maximum losses may occur after about 20 mm of rain.

Variable Runoff Losses as a Function of Time Indicate Very Different Infiltration Values for Different Rain Intensities. Figure A-30a shows that high infiltration rates are associated with high rainfall intensities (Pitt 1987). The Horton equation predicts a single infiltration relationship as a function of time, irrespective of rain intensity. When variable runoff losses are plotted against total rain depth (Figure A-30b ) a single relationship is seen (rain intensity multiplied by time duration gives rain depth). Horton actually recommended infiltration as a function of rain depth, but current practice of using double-ring infiltrometers to calibrate the Horton equation does not allow infiltration measurements to be made as a function of rain depth, only as a function of time for the ponded test conditions.

Disturbed Urban Soils Do Not Behave as Indicated by Typically Used Models. More rain infiltrates through pavement surfaces and less rain infiltrates through soils than typically assumed. Double-ring infiltrometer test results from Oconomowoc, WI, urban soils (Table A-3) indicated highly variable infiltration rates for soils that were generally sandy (NRCS A/B hydrologic group soils). The median initial rate was about 3 in/hr, but ranged from 0 to 25 in/hr. The final rates also had a median value of about 3 in/hr after at least two hours of testing, but ranged from 0 to 15 in/hr. Many infiltration rates actually increased with time during these tests. In about 1/3 of the cases, the observed infiltration rates remained very close to zero, even for these sandy soils. Areas that experienced substantial disturbances or traffic (such as school playing fields) had the lowest infiltration rates, typically even lower than concrete or asphalt! These values indicate the large variability in infiltration rates that may occur in areas having supposedly similar soils. Obviously, these variations can significantly affect site specific runoff predictions. The lowest infiltration rates were observed in areas having heavy foot traffic and in areas obviously impacted by silt, while the highest rates were in relatively undisturbed areas.

In an attempt to explain much of the variation shown in the above early tests, recent tests of infiltration through disturbed urban soils were conducted in the Birmingham, AL, area by the author and UAB students. Eight categories of soils were tested, with about 15 to 20 individual tests conducted in each of eight categories (comprising a full factorial experiment). Numerous replicates were needed in each category because of the expected high variation in infiltration rates. The eight categories tested were as follows:

Figure A-31 contains plots showing the interactions of moisture and compaction on infiltration for both soil texture conditions. Four general conditions were observed to be statistically unique:

· noncompact sandy soils

· compact sandy soils

· noncompact and dry clayey soils

· all other clayey soils

Compaction has the greatest effect on infiltration rates in sandy soils, with little detrimental effects associated with soil moisture. Clay soils, however, are affected by both compaction and moisture. Compaction is seen to have about the same effect as moisture on these soils, with saturated and compacted clayey soils having very little effective infiltration. In most cases, the mapped soils were similar to what was actually measured in the field. However, significant differences were found at many of the 146 test locations. Table A-4 shows that the 2-hour averaged infiltration rates and their COVs in each of the four major categories were about 0.5 to 2. Although these COV values are generally high, they are much less than if compaction was ignored. These data are being fitted to conventional infiltration models, but the high variations within each of the four main categories makes it difficult to identify legitimate patterns, implying that average infiltration rates within each event may be most suitable for predictive purposes. The remaining uncertainty can be considered using Monte Carlo components in runoff models. More detailed analyses of these data will be presented in the Toronto stormwater modeling conference next year.

Very large errors in soil infiltration rates can easily be made if published soil maps and typical models are used for typically disturbed urban soils. Knowledge of compaction (which can be mapped using a cone pentrometer, or estimated based on expected activity on grassed areas) can be used to much more accurately predict stormwater runoff quantity.

Basic Characteristics of the Small Storm Hydrology Model

Figure A-29 earlier showed the small storm hydrology model which describes the shape of the relationship between rainfall and runoff. Both small-scale and large-scale tests, described by Pitt (1987), obtained data to calibrate and verify this model for homogeneous impervious and pervious areas. The runoff response curve shown on Figure A-29 departs from the x-axis at the rainfall depth when runoff begins (r₀). This depth lag corresponds to initial runoff losses. After some rain depth (r₁), runoff losses become insignificant. For impervious areas, this is when the detention storage volume becomes filled, evaporation becomes insignificant due to pavement cooling, infiltration through the pavement or through cracks slows practically to nothing, and dirt and debris become saturated. Between these two rain depths, infiltration losses occur.

Both small-scale and large-scale tests, described by Pitt (1987), obtained data to calibrate and verify a model for homogeneous impervious and pervious areas. The runoff response curve departs from the x-axis at the rainfall depth when runoff begins. This depth lag corresponds to initial runoff losses (detention storage, evaporation losses due to pavement cooling, and dirt and debris absorbing moisture for pavements). After some rain depth, infiltration into the ground (or pavement or through cracks) slows practically to nothing, and each additional increment of rainfall results in a similar increment of runoff. Between these two rain depths, infiltration losses occur. Figure A-33 shows the model describing these infiltration losses. This figure plots cumulative variable runoff losses (F, inches or mm), ignoring the initial losses, versus cumulative rain (P, inches or mm), after runoff begins. The slope of this line is the instantaneous variable runoff loss (infiltration) occurring at a specific rain depth after runoff starts. A simple nonlinear model can be used to describe this relationship which is similar to many other infiltration models. For a constant rain intensity (i), total rain depth since the start of runoff (P), equals intensity times the time since the start of runoff (t). The small storm hydrology nonlinear model for this variable runoff loss (F) is therefore:

F = bit + a(1 – e^-git) or F = bP + a(1 – e^-gP)

Three basic model parameters were used to define the model behavior, in addition to initial runoff losses and rain depth: "a", the intercept of the equilibrium loss line on the cumulative variable loss axis; "b", the rate of the variable losses after equilibrium; and "g", an exponential coefficient. If variable losses are zero at equilibrium, then "b" would be zero. Because this plot does not consider initial runoff losses, the variable loss line must pass through the origin. This model reduces to the SCS model when the "b" value is zero and "a" is S’, and when Ia is 0.16 (80% of 0.2) of "a". This general model also reduces to the Horton equation when cumulative rain depth since the start of the event is used instead of just time since the start of rain.

Observed runoff data from both small- and large-scale tests were fitted to this equation to determine the values for a, b, and g for observed i and t (or P), and F values. In addition, outfall runoff observations from many different heterogeneous land uses were used to verify the calibrated model (Pitt 1987).

Comparison of the Small Storm Hydrology Model with the Horton Infiltration Equation

The Horton equation is used in many urban runoff models to predict infiltration losses (Skaggs, et al. 1969). The small storm hydrology model can be directly compared to the Horton infiltration equation. The total storm infiltration rate is:

where F(t) is an instantaneous infiltration rate. The instantaneous infiltration rate is then:

F(t) = df/dt.

From the small storm hydrology model:

F(t) = bi + agi(e^-git).

Therefore, the Horton infiltration equation is:

F(t) = Fc + (Fo - Fc)(e^-kt),

where Fc is the final equilibrium infiltration rate, Fo is the initial infiltration rate, k is the decay coefficient, and t is the time since the rain began. Therefore the small storm hydrology model and the Horton equation are equivalent if the following relationships are simultaneously true:

bi = Fc, or b = Fc/i

-git = -kt, or g = k/i

agi = Fo - Fc, or a = (Fo - Fc)/gi, or a = (Fo - Fc)/k.

Rearranging gives:

Fc = ib (if Fc is zero, then b is also zero),

Fo = ib + aig = i(b + ag), and

k = ig.

Based on these relationships, it is seen that the time since runoff began (t) is not a factor in determining any of the Horton infiltration parameters; but rain intensity (i) is a factor.

During the small-scale pavement runoff tests (Pitt 1987), the measured accumulative infiltration rates for the high rain intensity tests were much greater than for the low rain intensity tests for the same time since the start of the rain. The infiltration rates (depth per time) were therefore much greater for the high intensity tests. In urban hydrology studies, infiltration losses in pervious areas are usually considered to be the most important loss mechanism (Hromadka 1982). The previous discussion shows that infiltration is also an important loss mechanism for pavements. Simple infiltration estimation methods have received much attention in runoff analyses (Singh and Buapeng 1977). Singh and Buapeng found that errors in infiltration estimation may be large and may therefore be responsible for major errors in runoff predictions. One of the possible sources of infiltration estimation errors is the general lack of consideration of the apparent relationship between infiltration rate and rain intensity.

The relationship between rain intensity and infiltration can be related to the concept of variable contributing areas in heterogeneous watersheds. Areas having low infiltration capacities produce runoff during rains having relatively low intensities, while greater intensity rains are required to produce runoff from areas having high infiltration capacities. Therefore, an overall area infiltration rate appears to be variable and dependent on rain intensity. These variations have not been reported in the literature for homogeneous areas (such as large paved areas). However, infiltration in pavement "systems" includes infiltration through the pavement itself, infiltration through pavement cracks and seams, and infiltration through the pavement base. These different processes would have different infiltration rates; infiltration analysis for the whole system would therefore be intensity dependent.

Comparison of the Small Storm Hydrology Model with the NRCS Curve Number Procedure

The Natural Resources Conservation Service curve number procedure (SCS 1986) is commonly used in the design of storm drainage systems. The following paragraphs illustrate how the small storm hydrology model can interface with models using curve numbers. The small storm hydrology model can be used to select curve numbers, allowing the better incorporation of the mutual drainage and flood control benefits of many water quality control measures into the design of storm drainage systems (Pitt 1987).

The NRCS CN procedure can also be compared with the small storm hydrology model and the Horton infiltration equation. The small storm hydrology model can be rewritten, knowing that P = it so that F = bP + a(1 – e^-gP). However, the NRCS procedure assumes that the final equilibrium infiltration rate is zero (Fc = 0), therefore b is also zero, leaving: F = a(1 - e^-gP). When b is zero, the intercept of the runoff loss line is equal to the maximum runoff losses, ignoring initial runoff abstractions. Therefore, the NRCS S' value (maximum variable loss, without Ia, the initial abstractions) can be substituted for "a" in this equation:

F = S'(1 - e^-gP).

There is a distinct relationship between S and CN [CN = 1,000/(S + 10)], and therefore between S' (which is assumed to be equal to 0.8S by the NRCS) and CN in the NRCS procedure. Therefore, each curve number has a unique S' value. Because the NRCS CN procedure assumes zero final infiltration, the small storm hydrology model b value is zero and the "a" value is equal to S', as shown above. The small storm hydrology model g value was determined using a nonlinear computer program (the NONLIN module of SYSTAT - The System for Statistics, Version 3, 1986, from SYSTAT, Inc., Evanston, Ill.) for the specific F verses P relationships unique for each curve number (and S' value). The maximum runoff loss, S', which ignores initial abstractions, occurs after little rain for large curve numbers, but is not reached even after 90 mm of rain for curve numbers less than about 80.

Table A-5 shows the fitted small storm hydrology model equation parameter g values for several curve number values, using SYSTAT’s NONLIN module. This table also shows the NRCS S' values and the Horton initial infiltration rate (Fo) and decay coefficients (k) for these curve numbers. According to the small storm hydrology model, the Horton equation parameters are all related to rain intensity for impervious surfaces, and the small storm hydrology model g parameter is directly related to the curve number (Pitt 1987).

Volumetric Runoff Coefficients can be Calculated for Different Surfaces and Rains using the Small Storm Hydrology Model

Table A-6 is a summary of the volumetric runoff coefficients (Rv, the ratio of runoff to rainfall volume) for different urban surfaces and rain depths from detailed source area runoff tests and through calibrating the small storm hydrology model (Pitt 1987). Flat roofs and unpaved parking areas behave strangely similar because of similar detention storage volumes and no infiltration. Large impervious areas have the largest runoff yields because of very poor pavement under-drainage. The drainage path through the pavement base is relatively thin and very long, making it very difficult for infiltrated water to drain from the base. Street widths are much narrower than the widths of large impervious areas and the base water can drain much more effectively. Pitched roofs have no infiltration rates, but do experience limited initial losses associated with flash evaporation and sorption of moisture in leaves and other roof or gutter debris. After three inches (no longer a "small" rain) the runoff yields from all impervious surfaces are similar (within 10%), but the differences can be very large for the small rains of most concern in water quality evaluations.

Table A-6. Summary of Volumetric Runoff Coefficients for Urban Runoff Flow Calculations (Pitt 1987).

The impervious and roof area values are for directly connected surfaces. If runoff is allowed to drain across grass areas, then the runoff yield may significantly decrease. However, sufficient length of drainage across the pervious surface in good condition is needed. For a relatively small paved surface, short pervious drainage paths are all that are needed. If the paved area is large, or if the pervious area has clayey or compacted soils, then much longer drainage paths are needed before significant infiltration occurs.

Table A-6 does not accurately incorporate the effects of disturbed urban soils presented earlier, but the runoff coefficients shown generally bracket the range of likely conditions expected. Some users have had good success using an intermediate soil Rv value, half way between the clayey and sandy soil conditions shown, and only using the extreme values for more unusual cases. The four urban soil categories identified earlier better represent the conditions encountered, and appropriate coefficients are currently being developed.

The runoff coefficients and indirect connection correction values were determined from calibrating the small storm hydrology model for large urban watersheds having variable complexities in Toronto and in Milwaukee (Pitt 1987). The first calibrations were conducted for simple areas. The first area was the large parking area of a commercial shopping area. The runoff coefficients for this area were used to determine the runoff relationships from large flat roofs from another shopping area that was made of mostly paved large parking and roof areas in order to determine runoff characteristics for flat roofs. The next step was to evaluate runoff data for two high density residential areas that had very little pervious areas and had all of the impervious areas directly connected. The street runoff was subtracted from the total area runoff observations to obtain information solely for pitched roofs. Finally, two medium density residential areas were studied in areas that had clayey soils and all of the impervious areas were directly connected. Roof, street and other impervious area runoff information was subtracted to obtain clayey soil runoff coefficients. Similarly, a medium density residential area was studied in an area having sandy soils to obtain sandy soil runoff coefficients. Finally, two medium density residential areas having unconnected impervious areas were studied to obtain correction coefficients.

Excellent Verification of Small Storm Hydrology Model for Many Conditions

The final runoff coefficients were verified using additional runoff data from these same areas (that were not used in the calibration efforts) and from areas located elsewhere. Figures A-34 through A-37 show how well the small storm hydrology model works over a wide range of rain depths and for two very different land uses. The "Post Office" site was a commercial shopping center, the "Burbank" site was a medium density residential area. These sites were monitored as part of the EPA’s NURP project in Milwaukee (Bannerman, et al. 1983). Figures A-36 and A-37 are for two residential sites monitored by the WI DNR in Superior, WI, and in Marquette, MI, during 1993 and 1994. These last two sites were compared to the small storm hydrology component of SLAMM with no local calibration, demonstrating the excellent fit of observed and predicted flows.

The model was subsequently calibrated for these two sites to enable better fits for the larger events. It was originally expected that this model would not work very well for very large storms, especially in areas having appreciable pervious areas, where rain intensity was expected to have a more significant effect on infiltration than for small rains. The largest rains observed for the two Milwaukee sites were greater than three inches, a very large rain that would not be expected to commonly occur. Even these rains had runoff quantities that were well predicted by this runoff model.

Example Application using the Small Storm Hydrology Model

The small storm hydrology model can be used to predict runoff volume yields for many different land uses and development conditions. It was specifically developed to determine runoff yields and corresponding water pollutant yields for small storms for stormwater quality investigations. As shown during the verification process, it is also useful for predicting runoff yields for moderate storms that are used for drainage design. If used in conjunction with a model that can account for water losses associated with stormwater controls (such as SLAMM, the Source Loading and Management Model, Pitt 1986 and 1992) it can also be used to show the mutual drainage benefits associated with these controls. As an example, the use of roadside swales, disconnections of impervious areas from the drainage system, or using infiltration devices, can all have dramatic benefits in reducing runoff volumes, even for relatively large rains.

The small storm hydrology model can be used to predict runoff yields associated with different land uses and development practices. It can also be used to predict sources of water within the drainage area. If the variable quality of runoff from each source area is known, then runoff pollutant yield estimates (and reductions) can also be made. SLAMM uses this approach. This information is very important when determining the best management strategy for water volume and runoff pollutant reduction. This example problem shows how the runoff yield predictions and sources of water for a simple area can be predicted for different rain depths. The benefits of source area disconnections are also shown.

Predicting Runoff Yields from Different Source Areas

· Calculate runoff quantity (inches) and distributions (%) by source area for the following conditions:

- Rain depths: 0.12; 0.79; 3.2 inches

- Medium density residential area (conventional curb and gutters, all impervious areas are directly

connected to the drainage system and clayey soils are common), having the following surface area

distribution:

pitched roofs 6%

driveways 5

sidewalks 3

streets 12

front yards 45

back yards 29

· Calculations:

The Rv values are from Table A-6 for the appropriate rain depths and source area. Weighted Rv values are determined by multiplying the Rv values by the percentage of the area represented. The weighted Rv values are summed to obtain a Rv value for the whole land use area. The percentage runoff yields are the ratios of the individual weighted Rv values to the summed whole area Rv.

- runoff for the 0.12 inch rain: (0.014)(0.12in)=0.017 in runoff

- runoff for the 0.79 inch rain: (0.34)(0.79in) = 0.27 in runoff

- similar calculations for the 3.2 inch rain results in a Rv of 0.48,

therefore, the runoff for this rain: (0.48)(3.2 in) = 1.6 in runoff.

As the rain depth changes, the percentage contributions from each area also changes. For the smallest rain, all of the runoff is contributed from the directly connected impervious areas. However, pervious areas contribute almost half (44%) of the runoff for the 0.79 inch rain.

Benefits of source area drainage disconnections can also be predicted for this example. The following calculations show the effects of disconnecting half of the roof, driveway and sidewalk areas for this land use:

Original weighted Rv values:

0.12" rain 0.79" rain 3.2" rain

roofs+

driveways+ 0.084 0.11 0.13

walks

streets 0.059 0.08 0.11

yards 0 0.15 0.24

total Rv: 0.14 0.34 0.48

total runoff: 0.017" 0.27" 1.6"

With disconnections:

0.12" rain 0.79" rain 3.2" rain

roofs+

driveways+ (0)(0.084)= (0.21)(0.11)= (0.34)(0.13)=

walks 0 0.023 0.044

streets 0.059 0.08 0.11

yards 0 0.15 0.24

total Rv: 0.06 0.25 0.39

total runoff: 0.01" 0.20" 1.3"

approx. % reduction: 60 25 20

The runoff contributions from the disconnected areas are decreased by the factors shown on Table A-6 for medium density areas (with no alleys) having clayey soils. These disconnections can have significant effects on the runoff quantities generated for small rains. The runoff reductions for the larger rain will also likely be important for drainage design. Similar percentage reductions in peak runoff rates are also expected for these conditions.

Runoff volume is the most important hydraulic parameter needed for most water quality studies, while peak flow rate and time of concentration are the most important parameters for most flooding and drainage studies. Common small rains account for much more of the annual runoff volume than rare flooding events. Pitt (1987) showed that estimates of runoff volume could be made with only rain depth information. Other rain characteristics (including antecedent conditions, durations, intensities, etc.) did not substantially improve runoff volume predictions, but are likely needed for peak flow rate predictions.

The literature indicates that both initial runoff abstractions (mostly detention/storage) and continuous runoff losses (infiltration) are important for impervious surfaces. Recent work with disturbed urban soils has also shown that care must be taken when using soil maps for developed conditions. The small storm hydrology model predicts runoff from several types of paved, roofed, and disturbed soil urban surfaces.

This model was used to examine long-term rain conditions at many locations throughout the U.S. to indicate the significance of small and moderate sized rains in stormwater management. These smaller rains, compared to the typical "design storm" rains used for drainage system design, contribute the vast majority of stormwater pollutants. Stormwater control practices must therefore effectively address these smaller storms to provide effective pollutant and flow reduction schemes.

Field, R. "EPA research in urban stormwater pollution control." J. Hyd. Div. 106 (HY5), 819. 1980.

Field, R. "Urban runoff pollution control-state of-the-art." J. Environ. Eng. Div., 101, (EE1), 107. 1975.

Read the above material included as module 12. Since this is the last module in the series, students have now worked with a variety of prior modules that also contain information supporting this module. It may therefore be useful to briefly review selected portions of some prior modules also.

Allow up to 12 h for reading and doing the basic analyses, and up to 6 h for writing your web page.

Many of you also participated in Module 6 where you set up SLAMM and conducted some simple analyses. Use your previously evaluated area with your SLAMM files to address the following questions:

Rain File name: bhamsrce.RAN

Printout Date: 03-27-2000

Rain Beginning Beginning Ending Ending Rainfall Rainfall Intensity Interevent

Number Rain Rain Rain Rain Depth Duration (in/hr) Duration

Date Time Date Time (in) (days) (days)

1 01/01/99 00:00 01/01/99 03:00 0.01 0.1250 0.0033 30.8750

2 02/01/99 00:00 02/01/99 07:00 0.05 0.2917 0.0071 27.7083

3 03/01/99 00:00 03/01/99 08:00 0.10 0.3333 0.0125 30.6667

4 04/01/99 00:00 04/01/99 10:00 0.25 0.4167 0.0250 29.5833

5 05/01/99 00:00 05/01/99 12:00 0.50 0.5000 0.0417 30.5000

6 06/01/99 00:00 06/01/99 14:00 0.75 0.5833 0.0536 29.4167

7 07/01/99 00:00 07/01/99 14:00 1.00 0.5833 0.0714 30.4167

8 08/01/99 00:00 08/01/99 14:00 1.50 0.5833 0.1071 30.4167

9 09/01/99 00:00 09/01/99 14:00 2.00 0.5833 0.1429 29.4167

10 10/01/99 00:00 10/01/99 14:00 2.50 0.5833 0.1786 30.4167

11 11/01/99 00:00 11/01/99 14:00 3.00 0.5833 0.2143 29.4167