University Surface Water and Flooding Audit
 

Christina Crawford
Arturo Aguilar
Ginny Farmer
Erika Windham
 

ENVI 490 Surface Water and Flooding Group - April 15, 1998
 

Abstract

As any Rice student can attest to, the Rice University campus is prone to flooding after heavy rains. Over the course of this semester, the surface water and flooding group has investigated Rice’s watershed and has prepared a model to simulate the effects further development of the Rice will have on the campus stormwater system.

During the first part the semester, the group researched the history of the Rice’s stormwater system and collected data. From the Facilities and Engineering maps of Rice’s storrnwater system, we were able to divide the campus up into areas and calculate the area, length of pipe, and the percent of impermeable land for each sub-area. This data was used to calculate the modeling parameters, time of concentration (TC) and routing constant (R), needed for the computer watershed model HEC-1.

Five scenarios were modeled in HEC-1: 1) the current campus layout including buildings currently under construction (year-2000 model), 2) the campus layout in 2002, 3) the campus layout in 2000 with an expanded detention pond, 4) the campus layout in 2002 with an expanded detention pond, and 5) the campus layout in 2004 with an expanded detention pond. The results from the HEC-1 model showed that flooding would occur the area of campus located between Sunset and Main street down to Sid Richardson College (the Grey area) during a two-year flood and a five-year flood for both the year-2000 and year-2002 models without an expanded pond. No flooding occurred in the last three models which included the expanded detention pond. As predicted, the peak flow (Q~) increased as the amount of impermeable land increased.

Introduction

Water that runs off from the Rice campus and surroundings has always found its way to Bray’s Bayou. Much of this water originally drained into the Harris Gully, a subsidiary of the bayou. In 1907, the two hundred seventy-seven acre tract of land that Edgar Odell Lovett chose to be the site of the new Rice Institute was an expanse of treeless prairie located halfway between the southern city limits of Houston and Bray’s Bayou. Harris Gully ran along the western part of the site. A small patch of pine trees along its banks were the only trees on the property. Today, the campus contains approximately 50 major buildings and over 4,000 trees. Most of the water from the campus still continues to drain into the gully; however, the gully is now an 11.5 ft x 15 ft concrete culvert that actually flows undemeath the Rice campus.

As Rice has developed, more of its land has been rendered impervious. These impermeable areas add to the total amount of runoff from the campus, which ultimately dumps more water into the bayou and can even cause the campus to flood. The campus soil, which consists of mostly clay, is frequently unable to absorb all the water it receives. This excess water is carried to the Harris Gully through a man-made stormwater system. As the university builds more buildings, even more water is forced into the system. Once this system fills up the water backs up and floods the campus.

At first, the simple solution to Rice’s flooding problems would be to build a drainage system capable of moving the excess water off the campus as fast as possible. Although convenient for the Rice community, it is important to remember that the campus is part of a larger watershed that all reaches Bray’s Bayou. To do so would only increase flooding problems for everyone downstream of the University.

The City of Houston requires Rice to hold back part of the water it receives to decrease the rate at which it leaves the campus. In 1987, a detention pond was built near the track stadium. A large portion of the rainwater from the campus flows directly into this detention pond, from which it is eventually released into the Harris Gully culvert, and then into the bayou. Some water flows directly into the culvert, and the rest moves away from Rice, adding to the Main Street pipeline and others.

The purpose of the detention pond is to offset a fixed amount of impermeable land for future development. This creates an accounting system for land development. The current detention pond provides the University with a 278,784 sq. ft. Allowance. The allowance is reduced as more buildings are constructed. Likewise, if an impermeable surface like a parking lot is removed and landscaped, the allowance is increased. [Appendix A]

Eventually, this allowance could be used up, and there would be so much impermeable land that the detention pond would not have enough capacity to handle the resulting runoff. However, the pond was designed so that its capacity can be expanded as more construction occurs on campus. There are tentative plans to enlarge the existing pond in two more phases, once in 2002, and again in 2004. [Appendix A]

The pond is under study at the present time, because even with the planned expansions, it will reach a limit someday. Before that time, a new solution for excess surface water must be found. Possibilities include another detention pond or retention pond, maybe even exploring the reuse of storm water for irrigation purposes.

Background: Basics of Hydrology

A watershed is an area of land that drains to a single outlet and is separated from other watersheds by a watershed divide. Infiltration is the movement of water from the surface into the soil. When the rainfall rate (i) exceeds the infiltration rate (f), water infiltrates the surface soils at a rate that generally decreases with time. Rate of infiltration depends on rainfall intensity, soil type (Rice - type D — heavy plastic clay), surface condition, and vegetal cover. When rainfall strikes the land surface, it may initially distribute to fill depression storage, infiltrate to fill soil moisture and ground water, or travel as interf low to a receiving stream (i.e. Harris Gully and Bray’s Bayou). No more than 10% of a watershed contributes to overland flow. However, with urban development, the impervious cover increases and the overland flow percentage is larger.

Hydrology of urban areas is dominated by two distinct characteristics: (1) preponderance of impervious surfaces, (2) presence of man-made/hydraulically uimproved~~ drainage systems. The runoff volume is larger because there is less pervious area available for infiltration. Many engineering problems in urban hydrology include the needs to control peak flows and maximum depths throughout the drainage system. Urban hydrology consists of collection/analysis of rainfall data, computation of losses (evaporation, infiltration, etc.), conversion of rainfall excess to runoff, and flow routing down the drainage system. Thus, rainfall is the driving force in hydrologic analysis.

Alternatives for control of urban runoff quantity include detention and retention ponds. A detention pond is a reservoir that detains water for a given time and then discharges entirely down stream. A retention pond is a reservoir that retains floodwater without allowing it to discharge downstream. Instead water is removed from the pond by infiltration and evaporation. In summary, urban problems are mainly due to high imperviousness and fast response times. Thus, altematives must be made in order to control the increased overland flow.

Methods and Data

A map of the Rice University pipe and drainage system was obtained from the Facilities and Engineering Department. Using the map, the campus was divided into sixteen sub-units, one for each major pipe. [Appendix B] For each of these, the area was measured and calculated. Then, the length of the largest pipe in each sub-area was determined. The percent of impermeable ground was also determined for each sub-area. Fifteen of the sub-units were gathered into 4 larger units, based on the location where the runoff water drained into the Harris Gully culvert. (One of the sub-units was not included because it does not flow into Harris Gully.) The area of each unit was determined by weighing paper cutouts of each unit and multiplying the total area by each cutout’s fractional weight. The length of each major pipe and the amount of impermeable ground were calculated from the scale on the maps. This data plus the pipe diameters were used to calculate the Unit Hydrograph Modeling Parameters, time of concentration (TC) and routing constant (R). The equations from Table 5.13 in Hydrology and Floodplain Analysis by Bedient and Huber were used to calculate these parameters.
 

TC + R =
 

where

o = 4295[% development] O.678[% conveyancei0967
 / 1.06
IL
/
where C’ is taken from the following:

C’ = -0.0152(% development) + 2.46

R=(TC+R)-TC

where

L = length of channel (outflow to basin boundary) (mi)

= length along channel to centroid of area (mi)

So = representative overland slope (ft/mi) *Assumed to be 0.5285 ft/mi

% development = percent of area that is developed (Impermeable) (%)

% conveyance = ratio of flow in channel to total flow (%) *Assumed to be 50 %

TC = time of concentration (hr)

R = routing constant (hr)

Generally, the values of TC and R decrease as percent development and percent conveyance increase within a sub-area. The TC and R values used in our model of the Rice watershed are located in Appendix C.

After the first models were run, the slope of the pipes, which was originally considered to be the same as the overland slope, was changed to 5.285 ft/mi. This slope is the one used in the kinematic rate equation for routing. The change in slope also affected the TC and R values for the four larger units. The TC values for each of the four units were determined by taking the longest subunit TC in each area and adding the time it would take for the water in the subunit to reach Harris Gully via pipe. This time value was calculated by finding the velocity of the water traveling through the pipe using Manning’s Equation and multiplying it by the length of the pipe. Since the velocity is dependent on the slope of the pipes, a changing the slope of the pipes resulted in new TC and R values. Using the longest sub-unit TC ensures a conservative model.

Model

The TC & R modeling parameters were inputted into the HEC-1 computer watershed model. The HEC-1 model incorporates a variety of calculations, including routing, unit hydrograph synthesis, diversions, and hydrograph composition, allowing for a numerical depiction of flood events. We used kinematic routing to carry the calculated flood wave through the stormwater system. We used the Olark unit hydrograph method to create initial runoff hydrographs for each sub-area. Using the calculated values of TC and R. Diversion relationships were used to incorporate the effect of forcing all water collected in a sub-area into a single pipe of limited capacity. The order of these functions is presented in the flow chart. [Appendix D]

Four different scenarios were used to model the Rice watershed. The first scenario models the current university watershed. The second scenario models the university’s watershed in 2002 with the addition of two new residential colleges, a new humanities building, and a studio arts building. The two colleges were placed near Jones Oollege, the humanities building was placed between Rayzor Hall and Fondren Library, and the studio arts building was assumed to be somewhere near Sewall Hall where the art department is currently housed. The third scenario adds additions to Wiess College, the Jones School of Business, the library, and a new music library, assumed to be near the other library, which should be completed by 2004. The last scenario models the 2004 Rice watershed with a 135,000 square foot expansion of the existing detention pond near the track stadium.

4. Results

2-Year Flood (original slope)
 Year  Q0 (cfs) t0 (hour) Flooding
 2000  31  7.1  Yes-all
 2002  31  7.1  Yes-all

5-Year Flood (original slope)
 Year  Q0 (cfs) t0 (hour) Flooding
 2000  34  7.1  Yes-all
 2002  34  7.1  Yes - all

2-Year Flood (new slope)
 Year  Q0 (cfs) t0 (hour) Flooding
 2000  54  6.3  Yes—Grey Area
 2002  55  6.3  Yes—Grey Area
 2000*** 52  6.3  No
 2002*** 52  6.3  No
 2004*** 53  6.3  No

5-Year Flood (new slope)
 Year  Q~ (cfs) t0 (hour) Flooding
 2000  72  6.3  Yes—Grey Area
 2002  72  6.3  Yes—Grey Area
 2000*** 67  6.3  No
 2002*** 68  6.3  No
 2004*** 68  6.3  No

***denotes a 40 acre-feet increase in the existing detention pond size

Analysis and Conclusions

The predicted results of an increase in impermeable land were a higher peak flow (Q~) and a decrease in the time-to-peak (tn). This did not occur when running the model with the original slope. It is possible that the spatial and temporal scales used were too large to allow any variation to be seen. Originally the campus was divided into sixteen small sub-units to be modeled which may have given more information, but the parameters for the sub-units were too small for HEC-1 to process. Another possible reason that the results from the four models did not vary is that we modeled conditions for fairly large storms. If we had been able to model rainfall for smaller, more common storms, we might have seen more variance.

However, the problem with the first model runs was most likely caused by the fact that assuming the pipe slope is equal to that of the overland slope is not a very good assumption. In fact, it is probably safe in Houston, where the overland slope is close to zero, to assume that the pipe slope is greater than the overland slope, in order to move the water more efficiently. An increase in slope should decrease the time-to-peak, as well as increase the peak flow, because the water is moving faster. After running the model with the changes, the peak flow increased by roughly 75 %, and the time to peak decreased by 11 %. Less flooding occurred because with a higher slope, the pipes remove water at a faster rate. With the increased slope, slight changes can be seen as impermeable area is added between years. For the years before enlargement of the detention pond, the peak flow showed an increase in 1 cfs over a two-year period for the 2-year flood. No change in peak flow appeared when the data from a 5-year flood were used. A 5-year flood involves a greater amount of water than a 2-year flood, so any changes that might have occurred were too small to be noticed.

Increasing the size of the detention pond decreases the rate at which the water enters the main channel, causing the small drop in peak flow as seen in the results. Also, once the pond was enlarged, its greater capacity for water storage helps prevent flooding, which according to the model stopped completely. Small changes in the peak flow as impermeable area increased were still able to be seen between years 2002 and 2004 for the 2-year flood, and between 2000 and 2002 for the 5-year flood.

Finally, the group would recommend that anyone else who wants to try to model the Rice’s surface water system spend more than a semester working on this, explore other models besides HEC-1, and incorporate all minor and major pipes. The program was unable to perform at the level of detail which is necessary to analyze the Rice system thoroughly. It would also be useful to try to find a better estimate of the slope of the pipes, which affects the results.