By J. Cassidy

1. INTRODUCTION

If a dam is to be safe its spillway must be capable of passing extreme floods without jeopardizing the dam itself. Correct design, proper construction, and reliable operation of the spillway are critical to the safety of the dam and downstream residents. Thus it is not surprising that over the years several questions at ICOLD congresses have dealt with the subject of spillway design, construction, and operation. The first was "Methods for determining the maximum discharge that should be expected at a dam and for which it should be designed. Selection of type, capacity, and general arrangements of temporary or permanent spillways," Question 12 at the Fourth Congress in New Delhi in 1951. That question was followed by Question 41 at the Eleventh Congress, Question 50 at the Thirteenth Congress, Question 52 at the Fourteenth Congress, Question 63 at the Sixteenth Congress, and Question 71 at the Eighteenth Congress in Durban, South Africa. In each congress the questions have attracted considerable attention and many responses and contributions.

Other considerations relative to spillway design are included in ICOLD Bulletins 49a [3], 58 [4], 102 [5], and 105 [6]. In addition, a special international symposium "Dams and Extreme Floods" was held in conjunction with the 1992 Annual Meeting of ICOLD in Granada, Spain [7].

The consequences of inadequate spillways or improper operation is shown in ICOLD Bulletin 99 where problems with the operation of spillways (or insufficient spillway capacity) are given as the primary cause of failure for 22% and as a secondary cause for 5 % of the reported dam incidents [1]. For earth and rockfill dams overtopping was the primary cause of failure in 31 % and a secondary cause in 18 %, whereas overtopping was the primary cause of failure in 43 % of the reported failures of masonry dams. Clearly problems with spillways (insufficient capacity, improper operation during flood passage, or design deficiencies) could have been considered to have been a significant cause for the overtopping.

Forty three responses to Question 79 were received and those responses came from authors in 20 countries. This author found it a rewarding experience to read through those papers and particularly to see that great concern exists to improve the safe performance of spillways. In general the 43 responses can be grouped into the following three themes:

-"Selection of Gated or Ungated Spillways and their Operation." (8 responses)
-"Design and Selection of Hydromechanical Equipment." (16 Responses).
-"Operation and Reliability of Gated Structures." (19 Responses).

The responses provide a wide range of design criteria, operational information, and experiences involving the operation of ungated and gated spillways and associated equipment. The experiences include minor incidents as well as incidents of great consequence.

2. DESIGN AND CHOICE OF SPILLWAY TYPE

2.1. SPILLWAY CLASSIFICATION

Spillways can be classified into basically two types. surface spil/ways and orifice or tunnel spillways. The choice as to whether to use a surface spillway or an orifice spillway is generally governed by specific site characteristics. In a restricted canyon, it is often difficult to incorporate a surface spillway and an orifice or tunnel spillway is the only logical choice. An orifice spillway has the disadvantage that discharge through the spillway is a function of the square root of the head available while that of the surface spillway is a function of the three-halves power of the head. For this reason orifice spillways are generally found only on high dams where large heads are available. In many cases they utilize the original diversion tunnel as a discharge conduit.

Spillways can be further distinguished as gated or ungated spillways which basically provides the basis for Question 79. The characteristics of the design flood and the characteristics of the site generally govern the choice between an ungated and a gated spillway for a particular project. ICOLD Bulletin 82 [2] stated that trom 34 to 41 percent of dam failures had been caused by insufficient spillway capacity. This figure illustrates that there is considerable uncertainty in the determination of a proper design flood. Logically, the design of a spillway needs to be sufficiently conservative to allow for uncertainty in calculating the magnitude of the design flood.

Morning-Glory or shaft spillways are a special type and may be either gated or ungated. They are generally used where temporary storage space in the reservoir is large enough to significantly attenuate the incoming flood. Compared to a surface spillway, a shaft spillway has lesser maximum capacity for a given crest length since the hydraulic resistance of the shaft and the tunnel out let may become the control for large rates of flow. In general a shaft spillway should be designed keeping in mind the possibility of large trash (trees or logs) blocking the entrance. Guidelines produced by the French National Committee of ICOLD state that all shaft spillways should have a minimum throat diameter of 6 meters (R.29).

Spillway rehabilitation projects were described in several responses to Question 79 (R.2, 3, 8, 13, 15, 16, 22, 24, 30, 32, 36). Response (R.8) describes changes made to Sidi Mohamed Ben Abdallah Dam in Morocco after floods during 1990 of 1,000 and 1,200 cubic meters per second had resulted in significant damage. The dam was raised and the original three-bay uncontrolled spillway was converted to a four-bay gated spillway. The new spillway and the increased reservoir capacity enabled the project to safely pass the 10,000-year flood (peak inflow of 12,060 cubic meters per second) with a peak outflow of 5,800 cubic meters per second (R.8).

E. Cifres and R. Lopez of Spain (R.22) describe the design of a spillway for the Mora de Rebellos Dam which is both gated and ungated. In their case a gate was installed in the upstream end of a side-channel spillway in order to provide for some flexibility in operation of the spillway in order to maximize conservation storage.

For many other projects both gated and ungated spillways are provided. The gated spillway operates as a service spillway and passes smaller flood flows as necessary. An uncontrolled emergency spillway is also provided with sufficient capacity to pass extreme floods without endangering the dam. Thus, flexibility of operation is provided while still providing a greater margin of safety than might be possible with only a gated spillway. M. Miscik describes in detail the design decisions leading to the use of a gated and ungated spillway for the Ruzin I project in Slovakia (R.13).

2.2. UNGATED SPILLWAYS

Ungated spillways are without question the safest type of spillway (R.5, R.12, R.14). The ungated spillway is less likely to be obstructed by floating debris, and since there is no equipment to operate, its safe operation is not impaired by possible operator errors. However, ungated spillways are generally more expensive than gated spillways for a given maximum discharge rate since they will involve long crests and consequently wide chutes or conduit diameters. The contradiction to this general rule on relative cost is given in the ungated labyrinth crest. For cases where the reservoir surface area is relatively large compared with the inflowing flood volume and maximum head on the spillway will be relatively small, a labyrinth spillway can provide an economical and safe structure (R.5, R.23).

Regulatory authorities have not usually made specifie requirements relative to the choice of ungated or gated spillways. However in France, presumably for reasons of cost and the lack of flexibility in control, ungated spillways are commonly used only for dams in small drainage areas (less than 150 km²) (R.29).

Ungated spillways have the distinct advantage that they involve no operating equipment and, thus, require little regular maintenance.

2.3. GATED SPILLWAYS

Gated spillways are generally chosen for one of more of the following reasons:

-The site is restricted and it is not physically possible to construct the necessary
length of ungated crest.
-Control of downstream flooding, or maximization of conservation storage, requires more flexible control than would be provided by an ungated spillway.
-The total first cost of a gated spillway will generally be smaller than for an ungated spillway of equivalent capacity. T. Ionescu (R.14) states that in Romania gated spillways generally cost 25 to 30 % more than equivalent ungated crests.

Gated spillway bays should always be designed and constructed with facilities for placing bulkheads upstream of the gate in order that the gate can be serviced in the dry. An example of the problems which occur in an emergency situation can be taken trom the July 1995 failure of spillway gate number 3 at Folsom Dam in California. Stop logs could not be installed because provisions had not been made for them in the original design and construction completed in the early 1950s. Stop log slots were incorporated in hastily-fabricated steel trames for installation of the upstream side of the spillway piers [8]. Ultimately frames were fabricated and installed on all eight spillway bays. Although the trame successively provided a means for closing off the gate bay and isolating the failed gate, it now disrupts flow going to the gates and has increased the size of the vortices which form in front of the gates when they are operating at partial openings.

The need to design and install stop logs for the rehabilitation of the long-span gates on the River Indus Barrage was described in response R.40. This barrage had also been constructed with no provision for stop logs. In some cases where provision has not been made for bulkheads but their use becomes important and equipment or space is not available on the dam for handling of the bulkheads, floating bulkheads may provide a solution (R.21). The design of such bulkheads is complicated because of the control for their sinking. This solution may be particularly attractive in the case of long-span gates where the weight of the required bulkheads will be great.

A status of gated-spillways in Korea was given in response (R.41). Korean rules for design of gated surface spillways included:

-Freeboard between the top of gates and the maximum normal reservoir level should be at least 0.5 meters.
-Recent recommendations include that the Probable Maximum Flood be used as the design flood.
-Velocities of approach to gates should be limited to 4 meters per second.
-Distance between the invert of the approach channel and the crest of the spillway should be at least 20% of the design hear on the spillway crest.
-A hydraulic model study is required for design of all large dams.

2.4 ORIFICE SPILLWAYS

Orifice or tunnel spillways can be defined as spillways that have submerged inlets and are usually controlled with upstream guard gates which are frequently combined with a control gate mounted within the tunnel. They contrast with outlet works in that the opening is usually at mid level in the reservoir.

By contrast, intakes for outlet works are often at a low level in the reservoir. The most downstream gate provides flow control while the most upstream gate is a guard gate which can be closed when the downstream gate requires servicing. Special design is required for gates used for sediment sluicing in order to assure that the gate can be closed when necessary and to prevent damaging erosion (R.39). The operating control gates for the outlet works are usually top-sealing radial gates or vertical-lift wheel gates. In either case the gates are designed for upstream sealing (R.39, R.40).

Orifice or tunnel spillways must generally be designed and constructed to control and convey large flows. L. Garaulet, G. Vincent, and R. Catalina give guidelines for the design of such tunnels and describe the design of a replacement spillway for Contreras Dam in Spain where an older morning-glory spillway was converted to a mid level outlet to provide additional reservoir control (R.24). By lowering the out let from the surface, it was possible to reduce the required peak outlet flow from 800 to 720 cubic meters per second.

General guidelines for design of tunnel spillways provided in (R.24) include:

-Two inlets should be required for each tunnel.
-Two gates should be used in each outlet.
-Access to the control chamber should be provided that is separate from the
discharge tunnel should.
-Aeration downstream from the control gate must be provided. Consideration
should be given to bulking of the flow in the discharge tunnel.
-A model study is always required for final design.
- Large trash racks should be provided and should be designed assuming that 20 to 25 % of the rack is blocked with accumulated debris.

2.5. PHYSICAL AND ANALYTICAL MODELS FOR SPILLWAY DESIGN

For most large dams, physical models are used to perfect the hydraulic engineering design for the spillways. There is always a question of what scale should be used for the modal. The larger the model the greater will be the cost of constructing and testing the modal. The scale of the model used is generally governed by the minimum depths of flow that must be examined at critical points of the spillway. If the model scale is too small, the discharge coefficients determined through model testing may be too small. Results of model tests at scales of 1 :20 and 1: 15 conducted in France are quoted as underestimating the discharge coefficient by 10% or more (R.27). Unfortunately, this is an uncertainty that is inherent in any physical model.

Because of concerns about the accuracy of discharge capacity of spillways as determined by physical model studies, an Acoustic Scintillation flow meter was mounted in the discharge tunnel for the Mellanfollet Dam in Sweden (R.31). An old model study used during the original design for the spillway yielded a maximum capacity of 560 Cubic meters per second. A more recent model study, performed during design to increase the spillway capacity, yielded the maximum capacity of the original spillway as 750 cubic meters per second. Although it was not possible to measure such large flows, the flow meter indicated that the results of physical model tests are subject to some uncertainty.

New guidelines adopted for dam safety in Sweden have resulted in the need to increase the spillway capacity for many dams in Sweden (R.30). This need has in turn resulted in the need to perform model tests for many spillways. Because of the projected cost of the necessary physical model studies, attention was given to using analytical modeling in place of, or as a supplement to, traditional physical model studies. Response (R.30) describes an analytical study made for Bergefosen Power Station in Sweden where it was necessary to increase the spillway capacity from 2,000 to 3,500 cubic meters per second. Independent teams performed studies and obtained similar results. It is impressive to note that obtaining analytical solutions took from one to three weeks of fast computer time for each case studied.

<1 2 3> Next