Controlling the Oosterschelde storm-surge barrier: A policy analysis of alternative strategies. Vol. 2: Sensitivity analysis

(1)

CONTROLLING THE OOSTERSCHELDE

STORM-SURGE BARRIER-A POLICY ANALYSIS

OF ALTERNATIVE STRATEGIES

VOL. II, SENSITIVITY ANALYSIS

PREPARED FOR THE NETHERLANDS RIJKSWATERST AAT

SORREL WILDHORN, RICHARD STANTON

R-2444/2-NETH

SEPTEMBER 1979

Rand

SANTA MONICA, CA. 90406

(2)

This report was prepared with the support of The Netherlands Rijkswaterstaat under Contract No. DED-1770.

Library of Congress Cataloging in Publication Data Wildhorn, Sorrel.

Controlling the Oosterschelde storm-surge barrier. ([Report] - Rand Corporation; R-2444/2-NETH) CONTENTS: --v. 2. Sensitivity analyses. 1. Flood darns and reservoirs--Netherlands--Oosterschelde. 2. Storm surges--Netherlands--Oosterschelde. 3. Dikes (Engineering)--Netherlands--Oosterschelde. 4. Estuarine area conservation--Netherlands--Oosterschelde. I. Stanton, Richard, 1945- joint author. II. Netherlands (Kingdom, 1815- ). Rijkswaterstaat. III. Title. IV. Series: Rand Corporation. Rand report; R-2444/2-NETH.

AS36.R3 R-2444/2 [TC558.N42067] 08ls [627' .42] ISBN 0-8330-0181-7 (v.2) 79-26160

The Rand Publications Series: The Report is the principal pUblication doc-umenting and transmitting Rand's major research findings and final research results. The Rand Note reports other outputs of sponsored research for general distribution. Publications of The Rand Corporation do not neces-sarily reflect the opinions or policies of the sponsors of Rand research.

(3)

CONTROLLING THE OOSTERSCHELDE

STORM-SURGE BARRIER-A POLICY ANALYSIS

OF ALTERNATIVE STRATEGIES

VOL. II, SENSITIVITY ANALYSIS

PREPARED FOR THE

NETHERLANDS

RIjKSWATERSTAAT

SORREL WILDHORN, RICHARD STANTON

R-2444/2-NETH

SEPTEMBER 1979

Rand

(4)

PREFACE

In February 1953, a storm of unprecedented severity from the North Sea flooded much of the Delta region of the Netherlands, killing nearly 2000 people and inundating 130,000 hectares. As a result of this disaster, the Dutch government embarked on a mas.siveconstruction program called the Delta Plan to enhance protection from floods caused by the North Sea in the Netherlands and, especially, in the estuaries ofthe Delta region, southwest of Rotterdam. By 1974, the new dams, dikes, and other works were complete, or nearly so, in all the Delta estuaries except 'the largest-the Oosterschelde. There, building had barely begun when it was

interrupted by controversy.

The original plan had been to construct an impermeable dam across the mouth of the Oosterschelde, thereby closing off the estuary from the sea. This, however, threatened the Oosterschelde's rich and rare ecology and its oyster and mussel industries. In response to growing opposition, the Dutch Cabinet directed the Rijkswaterstaat, the government agency responsible for water control and public works, to study an alternative approach. But there were several possible ap-proaches, each with many variations.

It soon became clear that the process of comparing and choosing among the Oosterschelde alternatives would be difficult, for their potential consequences were many, varied, and hard to assess. To aid the decisionmaking process, the Policy Analysis of the Oosterschelde (POLANO I) Project was established, in April 1975, as a joint research project between Rand (a nonprofit corporation) 1 and the

Rijks-waterstaat.2

In April 1976 Rand presented a briefing to the Rijkswaterstaat describing the methodological framework that had been developed and summarizing the results of the POLANO analysis. The Rijkswaterstaat combined this work with several special studies of its own and, in May 1976, submitted its report to the Cabinet, which recommended the storm-surge barrier plan to Parliament. The barrier was to be a flow-through dam containing many large gates that would be closed in a severe storm. In normal weather, the gates would be open to allow a reduced tide to pass into the basin, the size of the tide being governed by the aperture in the barrier. The plan was adopted in June 1976( but no aperture size was specified for the barrier. After additional analysis by the Rijkswaterstaat to help determine the aperture size, Parliament approved an aperture of14,000 square meters in Septem-ber 1977.

The POLANO II analysis, conducted between April 1976 and April 1977, had two main thrusts. One was aimed at documenting the POLANO I study,3 the other at identifying necessary new research for the storm-surge barrier project. One of the new research areas was to specify and explore alternative barrier control

1 Rand had had extensive experience with similar kinds of analysis and had been working with the

Rijkswaterstaat for several years on other problems.

2 The Rand contract was officially with the Delta Service of the Rijkswaterstaat, which had direct responsibility for the protection of the Oosterschelde.

3 See Rand Report R-2121-NETH, Vols. I through VI, published between April 1977 and February

1978,

(5)

iv

strategies, their implications for the design of the barrier, and their other conse-quences or "impacts." This work led to the establishment of the Barrier Control (BARCON) Project in April 1977. The Rijkswaterstaat contracted with Rand for the study, and also set up a Dutch counterpart research team. The study has been a joint effort.

The primary purpose of the BARCON study was to perform research and analysis to assist the Rijkswaterstaat in a policy analysis of alternative control strategies for operating the storm-surge barrier. For each of the alternatives, the project analyzed several impacts, including the safety of the dikes along the Ooster-schelde; the effects on the ecology and the shellfish and fishing industries of the region; the impacts on water management and shipping in the basin; and the im-plications for the design of the barrier and its control system.

The methodology and results of the BARCON project are described in a series of Rand reports entitled Controlling the Oosterschelde Storm-Surge Barrier-A Policy Analysis

of

Alternative Strategies. In addition to the present volume, the following volumes 'in the series have been published:

• Vol. I, Summary Report (R-2444/1-NETH), by Louis Catlett, Sorrel Wild-horn, Richard Stanton, Ary Roos, and Jan Al

• Vol. III, Predicting North Sea Water Levels (R-2444/3-NETH), by Louis Catlett and Gaineford Hall, Jr.

• Vol. IV, Basin Response to North Sea Water Levels: The BAR CON SIM-PLIC Model (R-2444/4-NETH), by Louis Catlett, Richard Stanton, and Orhan Yildiz

Volume I describes the approach and summarizes the results of the complete analysis. It presents and compares, in a common framework, the several impacts of three promising control strategies.

Volume III describes and evaluates several models for predicting North Sea water levels outside the barrier. The models include prediction based on observed local water levels (correlation over time), on observed remote water levels (correla-tion over space), on observed weather condi(correla-tions (including short-term forecasts), and on weather forecasts up to 48 hours.

Volume IV describes the simulation model of the storm-surge barrier used to estimate the variation with time of different water levels inside the Oosterschelde basin, given specified sets of storms outside the barrier. It discusses the capabilities of the model (called SIMPLIC), the storm sets and tidal shapes used, and the model's inputs and outputs.

The present report, Vol. II in the BARCON series, describes the sensitivity analysis that was conducted. This analysis has two purposes: to show why we selected two specific strategy representations from two of the three promising barrier control strategy categories for further evaluation (presented in Vol. I), and to explore the effects on performance of varying specific elements of the control strategies.

Three comments about this series of reports are appropriate. First, although formally published by Rand, the series is a joint Rand/Rijkswaterstaat research effort; whereas only one of the reports lists Dutch coauthors, all have Dutch con-tributors, as can be seen from the a,cknowledgments pages.

(6)

v

Second, the methodology and results described in these reports are expanded . and refined versions of those presented by Rand in a February 1979 all-day briefing

to the Delta Service.

Third, Vols. II, III, and IV are not intended to stand alone, and should be read in conjunction with the Summary Report (Vol. I), which contains most of the contex-tual and evaluative material.

Thus, this report is directed to those who desire additional details on the design of control strategies and their performance.

(7)

viii

E-Level On-Off Strategies

The first class we call "E-level on-off strategies," in which the barrier is normal-ly in either a fulnormal-ly open or. fulnormal-ly closed state with a short period in between. All of the E-level strategies use observed water levels as their primary control signals. The "basic E-level strategy," the simplest, uses a single trigger. The closing rule is: Close when the observed outside water level exceeds the specified trigger E-level.

We explored two refinements of this basic strategy to obtain tighter control of peak inside water levels (between the limits of2.0 and 2.6 m discussed above) over a wide variety of storm characteristics. The first refinement adds to the basic strategy a second trigger that monitors 'observed inside water level. (Both trigger levels must be exceeded to initiate barrier closing.) This refinement reduces the scatter of lower peak inside water levels (below the 2.0-m lower boundary) and is intended to raise these levels to about 2.0 m, even in less severe storms. This strategy is called the "single-stage E-level strategy."

The second refinement is designed to reduce the scatter in higher peak inside water levels (above the 2.6-m upper boundary) and tends to lower these levels to below 2.6 m, even in the most severe storms. A third trigger criterion-the exceed-ance of a specified high value for the low slack water level before the next high tide-requires the redlJ,ction of the E-level trigger for the next high tide. We call this the "two-stage E-l~vel strategy." All E-level strategies use the same simple opening rule: Open when the inside water level exceeds the outside water level.

Attenuator Strategies

In the second class of strategies, called "attenuator strategies," the barrier is operated ordinarily in a partially closed state to permit, for example, the basin to fill gradually during a storm. (In extreme storms, however, the barrier may be closed fully at some point during the storm.) All attenuator strategies use predicted water levels as their primary control signal. And strategies that use predicted levels are inherently uncertain in their performance because of prediction errors. There-fore, they need a backup strategy that uses only observed water levels as triggers and that closes the barrier fully when such triggers are exceeded. We explored several attenuator strategies with various backup strategies, as well as pure at-tenuator strategies without a backup.

The simplest attenuator is one in which the oarrier is closed partially at low slack water to some fixed aperture, with the exceedance ofa specified outside water level (called P-level) predicted at the next high water. However, tight control over peak inside water levels is not possible with this simple strategy over a wide variety of storm characteristics. One refinement that provides some improvement is adding a simple inside water level trigger backup, so that when the observed inside level exceeds the trigger the barrier is closed fully from its partially closed (fixed aper-ture) position. In fact, for specific combinations of fixed aperture and this backup rule, tight control over peak inside water levels becomes possible.

An effective, but more complex', attenuator strategy uses a variable reduced aperture. The variable .aperture size is computed using a simple algorithm that relates reduced aperture to the specific values of low slack water before the next high water and the next predicted high water outside the barrier. But even though this variant provides tighter control of inside water levels over a wide variety of

(8)

ix

storms, combinations of prediction errors and very severe storms ofthe magnitude for which the barrier is designed can produce inside water levels above the limited dike watch level. Therefore, an ongoing backup strategy of the two-stage E-level strategy class is necessary, and it remains in on-line operation after partial closing. All attenuator strategies also use the same simple opening rule described for the E-level strategies.

RESULTS OF THE SENSITIVITY ANALYSIS

For each class of strategies, we estimated the effects on performance of varia-tion in a number of important strategy elements and of uncertainties in major assumptions. The primary analytktool used was the Rand SIMPLIC model, 1 which

computes the variation of basin inside water level with time, based on an explicit description of a storm (water level versus time) outside the barrier. SIMPLIC is a fast and inexpensive computer model designed to explore many alternative barrier control strategies in a wide variety of storms. It matches the results, within a few centimeters, of the Rijkswaterstaat's (RWS') IMPLIC computer model, a slower, more expensive, and much more complex mathematical model.

For the class of E-level strategies, we varied the E-level (between 2.0 and 3.0 m), the inside water level trigger (between 1.25 and 1.95 m), the low slack water trigger (between 0.75 and 1.0 m), gate closing time (±0.5 hours from the nominal one-hour time), closing method ("fast" and "slow" tapered closing, in which the aperture reduction is faster initially compared with the linear case, then slower), and barrier effective aperture (± 10 percent from the nominal 15,000 sq m).

For the class of fixed aperture attenuator strategies, we varied the fixed aper-ture (between 1500 and 6000 sq m) and the backup observed inside water level trigger (between 1.8 and 2.1 m), and then studied in detail the fixed J-lA = 1O,000-sq-m attenuator strategy with backup (a two-stage E-level rule with E = 2.75, IWL

=

1.8, but if LSW ~ 1.0, E

=

2.25 m). For the variable aperture strategy, we determined the effect of adding an ongoing backup two-stage E-level strategy in which the inside water level trigger varied linearly with residual aperture (from 1.5 m when fully open to 2.1 m when fully closed).

E-Level Strategies

General Observations. The goal of limiting peak inside water levels to b. e-tween 2.0 and 2.6 m above NAP for less than one to two tide cycles, was satisfied by several E-level backup strategies. Also, these strategies can be used as primary strategies without prediction. We identified three such strategies:

• Single-stage E-level strategy with E

=

2.25 m and inside water level trigger = 1.65 m.

• Two-stage E-level strategy with E = 2.60 m, inside water level trigger = 1.50 m; but if low slack water ~ 1.0 m, reduce E to 2.25 m.

• Two-stage E-level strategy with E = 2.75 m and other elements identical with the E = 2.60-m strategy above.

(9)

~---.

x

Although both two-stage E-level strategies have very similar inside water levels, closure duration, and head difference performance, they differ substantially in their implications for frequency of barrier closure; the lower the E-level, the more frequent the closures. (The basic E-level strategy with an outside water level trigger as its sole control signal is not promising because it does not provide adequate control of peak inside water levels over .a wide variety of storms.)

From these three promising strategies, we selected one for full evaluation of impacts (reported in the Summary Report). The strategy selected was the two-stage E-level = 2.60 m with inside water level trigger = 1.50 m (but if low slack water

~ 1.0 m, reduce E-level to 2.25 m). The major reason for this choice was that it offered better control over peak inside water level.

Varying E-Level in Single-Stage Strategies. Between E-Ievels of 2.0 and

2.75 m, reducing E-Ievel by 25 cm decreases peak inside water levels by 10 to 20 cm, depending on storm severity.

Varying Inside Water Level Trigger in Single-Stage Strategies. When

this trigger is varied between 1.25 and 1.75 m, reducing the trigger by 25 cm has the following effects:

• Minimum peak inside water levels are reduced by 10 to 20 cm, depending on storm severity. This holds over the entire range of E-Ievel examined. • Maximum peak inside water levels are reduced by 0 to 20 cm, depending

on E-level and storm severity.

Adding the Low Slack Water Trigger: The Two-Stage Strategy. Adding

a low slack water trigger of 1.0 m to the single-stage strategy with an E-level of 2.75 m and an inside water level trigger of 1.5 m has the following effects:

• It does not affect minimum peak inside water levels in storms of the severity experienced in the past (called historical storms), but it causes a 20-cm reduction in the more severe design storms (those for which the barrier is designed).

• Maximum peak inside water levels for the storm set are reduced by 20 to 25 cm, depending on storm severity.

Varying Closing Time. Increasing closing time to 1.5 hours from the nominal

one-hour case does not appear attractive, particularly in design storms, because some peak inside water levels. exceed the upper boundary limited dike watch level. Reducing closing time to a half hour is more attractive and provides equal or slightly better performance than the one-hour counterpart, provided that the inside water level trigger is increased appropriately. The main drawback offaster closing is increased peak head differences while the barrier is closed of up to 80 cm in design storms. This is due to an increased translation wave (the difference between basin and barrier inside water level) caused by the higher closing speed of the gates.

Varying Closing Method. Tapered closing was configured to reduce the

effect of the translation wave on head differences. The one-hour "slow" tapered closing (in which the initial closing speed is twice that of the nominal one-hour linear case) is almost identical with the half-hour linear closing, in terms of peak inside water level and closure duration. The one-hour "fast" taper (having an initial closing speed four times that ofthe nominal one-hour linear case) requires that the inside water level trigger be increased appropriately to achieve accept~ble inside

(10)

xi

water levels. Peak head differences while closing for the slow taper case are be-tween those of the one-hour and half-hour linear cases because of the reduced translation wave. The main advantage of the slow tapered closing over the normal one-hour linear closing is a reduction in peak head difference immediately after closure of 50 cm because of a reduction of translation waves.

Uncertainty in Barrier Effective Aperture. At this time there are residual uncertainties about the ultimate configuration of the storm-surge barrier and the precise flow characteristics through it. This means that the flow contraction coeffi-cient, and hence the effective aperture, of the barrier is uncertain. Accordingly, we explored the effects of varying effective aperture by

±

10 percent from the nominal value of 15,000 sq m. The results showed that peak inside water levels and closure durations are insensitive over this range of variation. Therefore, uncertainty of effective aperture ofthis magnitude should have little effect on performance or the choice of a preferred barrier control strategy.

Attenuator Strategies

Fixed Aperture Strategies. We compared the performance of several fixed aperture strategies (all with backup inside water level triggers that varied with aperture) in several historical storms. These alternative strategies varied from an aperture of 6000 sq m with a trigger of 1.80 m to an aperture of 1500 sq m with a trigger of 2.10 m. We concluded that some measure of control over peak inside water levels and closure duration is possible with fixed aperture strategies when the fixed JJ.A is less than 10,000 sq m. For example, with the smallest fixed aperture (1500 sq m), closure durations of about two tide cycles are common with peak inside water levels very gradually rising from about Ij2 m below NAP to about 1 to 1 liz

m above NAP.

However, using a larger fixed aperture of 10,000 sq m with a two-stage E-level backup incorporating a 1.8-m inside water level trigger produces better results:

• It provides tighter control over peak inside water levels in the most severe storms. In no case was the limited dike watch level of NAP

+

2.6 m exceeded, although it was closely approached.

• It manages to keep minimum peak inside water levels very close to NAP

+

2.0 m for both historical storms and less severe design storms. • Head differences while closing and while closed are comparable to those

for a two-stage E-Ievel strategy.

• However, every closure in design and historical storms involves the back-up closing mode, which makes this strategy less representative of the attenuator concept than the variable aperture strategy discussed next. Variable Aperture Strategies. Current capabilities for prediction of water levels outside the barrier are adequate for use with variable aperture strategies. These strategies provide a good measure of control over peak inside water levels and closure durations over a wide variety of storms. However, adding an ongoing backup strategy (a two-stage E-Ievel strategy with a variable inside water level trigger that depends on residual aperture) improves performance as follows:

(11)

---~---=

~~---

---

-xii

• It provides tighter control over peak inside water levels in severe storms, and reduces them from 2.70 to 2.50 m above NAP in the most severe design storms.

• In less severe design storms and in historical storms, performance is simi-lar with and without the ongoing backup strategy.

• It increases maximum head difference while closing (in the backup mode) when the outside water level is near its peak, but these head differences are still less than the two-stage E-level strategy alone.

• It has little effect on maximum head difference while closed.

Selection of an Attenuator Strategy for Full Evaluation. Al~hough both the fixed p.A strategy of 10,000 sq m with backup and the variable p.A (Attenuator A) strategy with ongoing backup provide adequate control of inside water levels over reasonable closure durations, we have selected only one of these promising strategies-Attenuator A with backup-for full evaluation of its impacts as report-ed in the Summary Report. We chose Attenuator A with backup because it is more representative of the attenuator concept (since fewer backup closures are re-quired), and it gives a more gradual rise in inside water level compared with the fixed p.A strategy of 10,000 sq m.

(12)

ACKNOWLEDGMENTS

A policy analysis such as BARCON owes much to the advice and assistance of innumerable individuals because of the diversity of topics considered, the de-pendence on other research and studies for essential information, and the differ-ences in language and location. It is impossible to acknowledge everyone who contributed in some way. We mention several names here-and the other volumes of the series mention more-but there are undoubtedly others who deserve our gratitude.

Acknowledging assistance from these many individuals and institutions does not imply that they are responsible for, or even necessarily agree with, our findings. If there are shortcomings in this report, the responsibility rests solely with the authors.

We are indebted to several high officials for the constructive questions and assista.nce they provided in several meetings and briefings. They include H. Engel,

Head of the Delta Service;l O. Boom, Chief Engineer-Director of the Directorate for Zeeland; J. F. Agema, head of the Delta Service Hydraulics Division; M. J.

Loschacoff, head of the Delta Service Hydraulic Structures Division; and A. C. de Gaay, head of the Delta Service Environment and Regional Planning Division. Two ad hoc Dutch project groups provided invaluable assistance and served as major contacts. Members of the BARCON Project Group included:

J. Voogt, Chairman, Delta Service J. P. AI, Delta Service

H. D. Rakhorst, Delta Service (replaced by G. van Houweninge) F. J. Remery, Directorate of Bridges

A. Roos, Delta Service H. L. F. Saeijs, Delta Service

H. N. J. Smits, Delta Service (replaced by K. J. Vriesman) K. van der Spek, Delta Service

J. A. van Hiele, Directorate for Zeeland (replaced by J. Rus) Members of the Project Group Oosterschelde Dikes (P.O.D.) included:

A. Roos, Chairman, Delta Service J. W. Boehmer, Delta Service

H. Burger, Netherlands Dike Research Centre (replaced by A.

Penning)

P. G. J. Davis, Delta Service

J. W. Honders, Delft Soil Mechanics Laboratory P. van der Veer, Data Processing Division of RWS P. C. van Goor, Directorate for Zeeland

M. H. Wilderom, Directorate for Water Management and Hydraulic Research (replaced by D. van Dam)

1 These organizational affiliations are those at the time of the study. Several individuals have since

moved to other parts of the RWS or other organizations.

(13)

xiv

Three officials of the Delta Service played crucial roles both at the outset of the study, by helping to define its aims and structure the overall research, and during the study, by offering critical comments on our interim findings. They are T. Goe-mans, head of the Systems Management Division; J. Voogt of the Hydraulics Division and Chairman of the BARCON Project Group; and H. N. J. Smits of the Systems Management Division, who was also responsible for administrative mat-ters.

A. Roos of the Delta Service Hydraulics Division deserves special thanks for several contributions. He was the principal liaison between Rand and the Delta Service for almost the entire duration of the project. He played an indispensable role in formulating and analyzing many of the barrier control strategies, in coor-dinating data-gathering research in the Netherlands, and in providing liaison with other studies and organizations. In addition, as a coauthor of Vol. I ofthe BARCON series, the Summary Report, he was responsible for the section on dike safety.

Rand was assisted in each major area of the BARCON project by a number of Dutch colleagues. F. Spaargaren and G. van Houweninge of the Delta Service Hydraulic Division supplied extensive background information for the overall de-sign of the barrier and to establish the relation between the control strategies and the boundary conditions for design. F. J. Remery and D. P. van Wijk of the Director-ate of Bridges provided important documentation for the control system of the barrier and related reliability problems. H. L. F. Saeijs, J. P. AI, and J. Visser of the Delta Service Environmental Division supplied many data and constructive comments on ecological matters, including salt marshes, intertidal areas, detritus, and the ecological balance. In addition, J. P. AI, as a coauthor of Vol. I, helped prepare the section on ecological impacts.

We thank W. J. Dek and T. Walhout of the Delta Service Water Management Division for compiling the basic data necessary to estimate water management and shipping impacts.

Three people were instrumental in providing us with information for the pre-diction of storm surges. In discussions with H. Timmerman of the Royal Dutch Meteorological Institute (KNMI), he introduced us to his North Sea set-up model as well as other weather models, and gave us valuable advice on the accuracy of predicting surges. A. van Urk ofthe Storm Flood Warning Service (SVSD) provided crucial data on observed and predicted water levels and familiarized us with the role of the SVSD. A. W. Donker of KNMI assisted us with information on the prediction of water levels in -the Oosterschelde and the possibilities for long-term forecasting of storms and their associated surges.

We are grateful to P. H. van der Weele and P. C. van Goor ofthe Oosterschelde Dike Reinforcement Division ofthe RWS Directorate for Zeeland for their informa-tion on dike safety-the quality of the existing dikes and the current dike improve-ment program.

D. Kooman of the Delta Service Hydraulics Division supplied helpful informa-tion on the characteristics of storm surges, hydraulic condiinforma-tions during closing, and possible barrier closing criteria.

We also acknowledge the assistance of many Rand colleagues. J. J. Leendertse, who was born and educated in the Netherlands, gave us valuable counsel on hydro-logical problems, the history of the Oosterschelde debate, and the organization and operation of the Netherlands government. B. F. Goeller, head of the previous POLANO project, was the administrative director ofthe Rand team. We are

(14)

grate-xv

ful to him for providing broad general direction to the project and constructive comments during the preparation of interim and final study briefings, subsequently presented in the Netherlands. L. E. Catlett, although not a coauthor of this volume, performed much of the work on strategy development reported here and served as leader of the Rand BAR CON team.

R. L. Petruschell originally suggested the need for the BARCON study, serving as its first project leader and developing its initial structure. In this early phase, A. F. Abrahamse, D. L. Jaquette, T. F. Kirkwood, L. H. Wegner, and L. W. Miller (a Rand consultant) made important contributions to the study. We thank D. J. Leinweber for his research on the reliability of the barrier control system and his briefings to the RWS. J. H. Bigelow, drawing on his extensive background in the POLANO study, counseled us on ecological subjects, in particular, detritus import.

R. R. Rapp, author of App. B in Vol. III, contributed to our findings on surge forecasting.

Weare grateful to M. B. Berman and J. H. Rosen, Rand reviewers, for their constructive comments that led to many improvements in substance and presenta-tion of this volume.

E. T. Gernert edited this report for style, and served as managing editor for the BARCON series. Among the many artists who were involved in the project, we are especially grateful to J. H. Sloan.

M. P. Dobson, ably assisted by J. P. Marshall and M. Redfield, provided expert

(15)

FIGURES

1.1. The storm-surge barrier across the mouth of the Oosterschelde .. 2

1.2. The three sections of the storm-surge barrier ... 3

1.3. Cutaway perspective of barrier at deepest gate . . . 5

2.1. Water level profiles with open barrier ... 8

2.2. Time sequence of trigger levels for barrier closure ... 9

2.3. Comparison of basic on-off barrier control strategies .. . . 11

2.4. Single-stage E-level strategy . . . 12

2.5. Two-stage E-level strategy. . . 13

2.6. Attenuator A strategy. . . 14

2.7. Attenuator A with ongoing backup strategy. . . 15

3.1. Format for presentation of results: example-a specific control strategy operating in "design storms" ... 17

3.2. Effect of varying E-level in basic E-level strategies: historical storms... 19

3.3. Effect of varying E-level in basic E-level strategies: design storms 19 3.4. Effect of adding an IWL trigger to basic E-level strategy: historical storms (E-level

=

2.75 m, IWL trigger

=

1.50 m) ... 20

3.5. Effect of adding an IWL trigger to basic E-level strategy: design storms (E-level = 2.75 m, IWL trigger = 1.50 m) . . . 20

3.6. Effect of varying IWL trigger in single-stage E-level strategies: historical storms (E-level

=

2.50 m) ... ... ... 21

3.7. Effect of varying IWL trigger in single-stage E-level strategies: design storms (E-level = 2.50 m) ... 21

3.8. Effect of varying IWL trigger in single-stage E-level strategies: historical storms (E-level = 2.25 m) ... ... .. : 22

3.9. Effect of varying IWL trigger in single-stage E-level strategies: design storms (E-level = 2.25 m) ... 22

3.10. Effect of varying IWL trigger in single-stage E-level strategies: historical storms (E-level = 2.00 m) ... 23

3.11. Effect of varying IWL trigger in single-stage E-level strategies: design storms (E-level

=

2.00 m) ... ... 23

3.12. .Effect of varying E-level in single-stage E-level strategies: historical storms (IWL trigger = 1.50 m) ... 25

3.13. Effect of varying E-level in single-stage E-level strategies: design storms (IWL trigger = 1.50 m) . . . 25

3.14. Effect of adding the low slack water (LSW) trigger: historical storms (E-level = 2.75 m, IWL trigger

=

1.50 m) . . . 27

3.15. Effect of adding the low slack water (LSW) trigger: design storms (E-level = 2.75 m, IWL trigger = 1.50 m) . . . 27

3.16. Effect of varying E-level in two-stage E-level strategies: historical storms (IWL trigger

=

1.50 m) ... 29

3.17. Effect of varying E-level in two-stage E-level strategies: design storms (IWL trigger

=

1.50 m) ... ... 29

(17)

3.18. 3.19. 3.20. 3.21. 3.22. 3.23. 3.24. 3.25. 3.26. 3.27. 3.28. 3.29. 3.30. 3.31. 3.32. 4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7. - - - c - -- - -xx

Comparison of promising single-stage and two-stage E-level

strategies: historical storms ... .... .

Comparison of promising single-stage and two-stage E-level

strategies: design storms ... . Effect of reducing closing time: historical storms (single-stage E-level

=

2.75-m strategies) ... ... ... . Effect of reducing closing time: design storms (single-stage E-level = 2.75-m strategies) ... ... ... .

Effect of reducing closing time: historical storms (single-stage E-level = 2.25-m strategies) ... ... . Effect of reducing closing time: design storms (single-stage E-level

= 2.25-m strategies) ... .

Effect of reducing closing time: historical storms (two-stage E-level

= 2.75-m strategies) ... .

Effect of reducing closing time: design storms (two-stage E-level

=

2.75-m strategies) ... ... ... . Effect of varying closing time (two-stage E-level = 2.75-m, IWL

trigger = 1.50-m strategies) .... ... ... . Aperture-time profile for linear and tapered closings ... . Effects of tapered closing: historical storms ... .

Effects of tapered closing: design storms ... . Water levels at barrier during and after closing (1959 design storm resulting in largest head differences) ... ... . Effect of uncertainty in total barrier p..A: historical storms ... . Effect of uncertainty in total barrier p..A: design storms ... .

Performance of fixed p..A attenuator strategies: historical storms

(P-level = 2.75 m) ... ... .

Performance of fixed p..A

=

10,000-sq-m attenuator strategy with

backup ... . Variation in inside water level for attenuator strategies (storm of February 1, 1953) ... ... . Ongoing E-level backup rule for Attenuator A strategy ... .

Inside water levels for the Attenuator A strategy with and

without ongoing backup (part of December 23, 1954, storm) ... .

Comparison of Attenuator A strategies with and without ongoing backup: historical storms .. .... ... ... . Comparison of Attenuator A strategies with and without ongoing backup: design storms ... ... ... .

30 30 33 33 34 34 35 35 36 37 38 38 40 42 42 46 47 49 51 52 53 53

(18)

TABLES

2.1. Categorization of Barrier Control Strategies. . . 10 3.1. Peak Inside Water Levels with Different E-Level and IWL

Triggers: Single-Stage E-Level Strategies ... 26 3.2. Effect of E-Level on Number of Historical Storms in Which No

Closure Occurs ... 26 3.3. Number of Storms in Which LSW Trigg~r Is Exceeded ... 28

3.4. Peak Inside Water Levels with 1.5-Hour Barrier Closing Times .. 32 3.5. Effects of Closing Method and Time on Maximum Head

Differences ... . . . 39 4.1. Peak Basin IWL for Fixed Aperture Strategy ... 47 4.2. Use of Backup IWL Trigger for Fixed IlA = 10,000 Strategy .... 48 4.3. Comparison of Head Differences between Attenuator A and

E-Level Closing Strategies ... 52

(19)

GLOSSARY

Alarmpeil. See E-Ievel.

Aperture (A): The total geometric area through which water may flow through the barrier. The aperture may vary from near zero when all gates are closed to about 18,000 sq m when all gates are fully open.

Attenuator strategy: A strategy that allows the barrier to be operated in a partially closed state to achieve desired effects, such as letting the basin gradually fill during a storm.

Attenuator A strategy: An attenuator strategy designed to reduce the aperture at LSW to achieve a target IWL at the next high water. This aperture is calculat-ed using a simple algorithm that depends on LSW and the next prcalculat-edictcalculat-ed high water level.

Backup strategy: An E-Ievel strategy that serves to back up any strategy using a predicted water level that is subject to error.

BARCON: A research effort between Rand and the Rijkswaterstaat. The study purpose is to perform research related to the policy analysis of alternative strategies for the Oosterschelde storm-surge barrier.

Barrier control strategy: A strategy that consists of (1) actions required to govern the time and rate of storm-surge barrier gate closing and opening; (2) the rules behind the decisions for these actions; and (3) the gathering and processing of information needed for decisionmaking.

Basic E-Ievel strategy: A strategy basing the closing decision on only observed outside water levels.

Design storm set: Twelve variations of each of two extreme storm surges that are so severe that they could be expected to occur only once in several thousand years.

Detritus: Dead organic matter suspended in water.

E-Ievel: An emergency threshold trigger level based on observed outside water levels (known as alarmpeil in Dutch).

E-Ievel strategy: A strategy that permits closing the barrier fully when the ob-served OWL exceeds E-Ievel.

Exceedance frequency: For a given coastal location, the number of times per year that the water level exceeds a certain value.

Extended dike watch level: A water level above which the SVSD alerts provincial water boards, which then fully staff emergency control rooms and man local control posts. The extended dike watch level is exceeded on the average once in five to ten years. This is NAP

+

3.10 m at Burghsluis (just inside the storm-surge barrier location) and higher farther inside the Oosterschelde. Gate: The movable part of the storm-surge barrier used to adjust the barrier

aperture. The current storm-surge barrier design calls for 63 gates of 40 m in width and ranging in height from 5.5 to 11.5 m.

Grenspeil: The once-in-two-year water level at a given location in the Netherlands. At Burghsluis, near the storm-surge barrier location, it is NAP

+ 2.75

m. Head difference: The difference between outside and inside water levels at the

barrier.

(20)

..

xxiv

Historical storm set: A series of 44 historically recorded storms (between 1920 and 1970) that exceeded grenspeil somewhere along the Netherlands coast. HSW: High slack water. HSW occurs at that moment when there is no flow through

the barrier, as the water flow reverses from into the basin to out of the basin.

It corresponds to a maximum mean basin IWL.

Hybrid strategy: A barrier control strategy containing elements of the on-off and attenuator strategies.

Hydraulic jump: A phenomenon associated with certain flow conditions through the barrier in which the IWL at the barrier drops below its expected level, thus increasing the head difference across the barrier or other parts of the barrier foundation.

IMPLIC: A general computer simulation model of hydraulic flow over a two-dimensional network adapted to simulate the Oosterschelde. The model was developed by the Rijkswaterstaat.

Inside translation wave: The difference between the IWL at the inside of the barrier and the mean basin IWL. (See Translation waves.)

IWL: Inside water level. This level ordinarily varies from place to place in the basin.

KNMI: Royal Dutch Meteorological Institute.

Leakage: Water that passes under or around parts of the barrier, which is not totally impermeable.

Limited dike watch level: A water level above which the SVSD alerts provincial water boards, which then partially staff emergency control rooms and prepare for further developments. The limited dike watch level is exceeded on the average once a year. This is NAP

+

2.60 m at Burghsluis (just inside the storm-surge barrier location) and higher farther inside the Oosterschelde.

LSW: Low slack water. LSW occurs at that moment when there is no flow through the barrier, as the water flow reverses from out of the basin to into the basin.

It corresponds to a minimum mean basin IWL.

LSW strategy: The strategy to fully close the barrier at LSW on the basis of a predicted exceedance of the trigger OWL at the next high tide.

Mean basin IWL: A measure of the total water in the Oosterschelde basin. This is the water level that would exist throughout the basin if the water were static or stagnant; it is approximately equal to the level at Wemeldinge, near the center of the basin.

NAP (Normaal Amsterdams Peil): Essentially the mean or reference sea level in the Netherlands.

On-off strategies: Strategies that enable the barrier to be normally in either a fully open or fully closed state, with only a short period (an hour or so) of gate movement in between these states.

Outside translation wave: The difference between the OWL at the barrier and at the boundary to the Oosterschelde in the North Sea. (See Translation waves.) OWL: Outside water level. This level can be measured either at the barrier or at

the boundary in the North Sea.

Piping: Occurrence of a sand-transporting spring in a dike caused by a large and long-lasting head difference at the dike.

P-level: A threshold trigger level based on predicted outside water level (known

(21)

xxv

POLANO (Policy Analysis of the Oosterschelde): A previous joint research effort

between Rand and the Rijkswaterstaat to evaluate alternative methods of preventing flooding in areas surrounding the Oosterschelde.

Primary strategy: The strategy expected to be used predominantly in controlling the barrier. If uncertainties exist about the performance of this strategy, a backup strategy may also be employed.

Reductor strategy. See Attenuator strategy. RWS: Rijkswaterstaat.

Scorecord: A matrix display device that presents various impacts for each of several policy alternatives; for BARCON, it displays control strategies. Set-up: The difference between the observed water level and the predicted

as-tronomical tide that is caused by meteorological and other phenomena. SIMPLIC: A simple mathematical computer simulation model ofthe Oosterschelde

basin.

Single-stage E-level strategy: The strategy that initiates barrier closing when both

OWL (E-Ievel) and IWL triggers are exceeded.

'Slack water: The point at which inside and outside water levels at the barrier are equal, and there is momentarily no flow through the barrier.

Sluitpeil. See P-Ievel.

Storm surge: The large set-up that occurs under storm conditions.

Storm-surge barrier (SSB): A flow-through dam containing many large movable gates at the mouth of the Oosterschelde.

SVSD: Storm Flood Warning Service.

TIWL strategy (target IWL strategy): A strategy that initiates full closure of the barrier to achieve an intermediate target IWL following a predicted exceed-ance of the trigger OWL at the next high tide.

Translation waves: Long waves that arise from the dynamic effects of changing water flow conditions through the channels into the Oosterschelde. (See Inside and Outside translation wave.)

Trigger level: A water level, either inside or outside the barrier, which if exceeded

causes activation of some part of a barrier control strategy.

Two-stage E-Ievel strategy: A strategy, similar to the single-stage E-Ievel strategy, that reduces the OWL trigger for the next high tide when an unusually high

LSW is observed.

Wave overtopping: Large waves associated with high OWLs that actually break over the closed barrier with some consequent increase in IWL.

/1A: Effective aperture-the actual geometric opening A multiplied by a flow

(22)

Chapter 1 INTRODUCTION

This report has two purposes. The first is to show why we selected two specific strategy representations from two of three promising strategy categories (E-level on-off strategies and attenuator strategies). The second is to explore the effects on performance of varying specific elements of the control strategies.

In policy analysis the analyst must demonstrate the consequences (or impacts) (1) of selecting different classes of policies, (2) of varying important elements within a policy, and (3) of important uncertainties in major assumptions. To do less than this would both deny the decisionmaker valuable information and introduce bias, perhaps unintentionally, into the analysis. In the present report we describe the sensitivity analysis conducted to screen out less promising strategies and to explore the effects of varying specific elements of certain promising barrier control strate -gies.

1.1. FOCUS OF THE SENSITIVITY ANALYSIS

The focus ofthe sensitivity analysis is on a limited set of performance measures for assessing different control strategies-basin inside water levels, their durations, and head differences across the barrier while closing and while closed. Two of these measures-inside water level and duration while closed-are surrogate measures for certain ultimate impacts mentioned below and discussed more fully in the

Summary Report, such as dike safety impacts, ecological and commercial fishing impacts, and water management and shipping impacts in the Oosterschelde basin. The third measure-head difference-assesses more directly the ultimate impacts on barrier loads. Occasionally, we use a fourth measure, frequency of closure. Therefore, this report should be read in conjunction with the Summary Report so that the broader implications of selecting a preferred strategy may be understood. The sensitivity analysis presented here is intended to be illustrative, not ex-haustive. For example, it is sufficient to vary a particular policy element-say, closing time-within one policy class to gauge its effect. One need not do it for all barrier control strategies considered, since similar effects obtain in the other classes.

1.2. THE STORM-SURGE BARRIER: A BRIEF DESCRIPTION

The storm-surge barrier will be built along a curving trajectory across the mouth ofthe Oosterschelde from Schouwen on the north to Noord Beveland on the south (see Fig. 1.1). The total distance along the trajectory is approximately 9 km. But the barrier itself, to be built in three sections across three gaps, will be about 2.8 km in length. The southern section across the Roompot Gap is 1440 m long and connects Noord Beveland with the southern work island. (The pillars for the barrier

(23)

•

2

Schouwen

Noord Beveland

Fig. 1.1-The storm-surge barrier across the mouth of the Oosterschelde

will be constructed on this work island.) The middle section across the Schaar van Roggenplaat Gap is 720 m in length and connects the two work islands. The north-ern section across the Hammen Gap is 675 m in length and connects the northnorth-ern work island with Schouwen. There is bottom protection extending out on both sides of the barrier 650 m on each side in the Hammen and Roompot sections and 550 m in the Schaar section.

The current barrier design calls for 66 pillars with 63 openings between them,

to be closed off by gates, with 32 gates in the Roompot section, 16 in the Schaar section, and 15 in the Hammen section (see Fig. 1.2). Each opening is 40 m long. The upper level is at NAP

+

1 m and the lower level varies between 4.5 m and 10.5 m below NAP. Thus, the gates that close off the openings vary in height from 5.5 m for the shallowest to 11.5 m for the deepest.

In the BARCON study we have assumed a nominal effective aperture of 15,000 sq m when the barrier is fully open. The Dutch Parliament mandated a minimum effective aperture of 14,000 sq m, In order to guarantee this effective aperture, the designers of the barrier have had to plan on a gross, or geometric, aperture of 18,000 sq m, obtained ,by adding 1500 sq m to account for a flow contraction co-efficient of 0.9, 1000 'sq m to account for a few gates that might be closed for maintenance, and 1500 sq m to aC,count for possible inaccuracies in mathemati-cal and physimathemati-cal smathemati-cale models, which calculated the tidal condition in the basin

(24)

Noo,d

I

Beveland f...~--- 1965m 1--- - - - -1440 m ---.---~_I -I [-40m Roampot Gap F%~;>>l Dam Work island

Schaar van Roggenplaat Gap

Fig. 1.2-The three sections of the storm-surge barrier

(25)

4

with the barrier in place. The nominal 15,000 sq m assumed seems a reasonable approximation for the operating condition of the barrier.

Figure 1.3 shows a cutaway perspective of the barrier at the deepest gate. The massive reinforced concrete pillars, some 50 by 25 m at their base, with heights up to 45 m, rise to 15 m above NAP. The pillars are set on a prepared bottom, and a sill is constructed between and around them. Upper and lower concrete beams are placed between the pillars. The steel gates move vertically in a slot in the pillar.

The current design calls for the gates to be moved by massive hydraulic cylinders, although an alternative all-mechanical design is under consideration. There will also be a road over the barrier some 12 m above NAP.

1.3. EFFECTS OF THE BARRIER IN NORMAL AND STORMY WEATHER

The presence of the open barrier in nonstormy weather and the closed barrier in stormy weather can have a number of effects, called impacts, on the Ooster

-schelde environs. Such impacts occur in the areas of dike safety, barrier loads, ecology, and water management and shipping. The water level versus time profile inside the Oosterschelde basin causes these impacts. Currently, during normal weather and without a barrier, the water level inside the basin rises and falls with the tide. With the installation of the open barrier this tidal amplitude will be reduced, because of the reduction in the aperture (from about 80,000 to 15,000 sq m) at the mouth of the basin. The water level versus time profile inside the basin during storms will depend on which barrier control strategy is adopted. If the selected strategy implies infrequent closure (i.e., only in very severe storms), high inside water levels (IWLs) for short durations will be experienced more frequently in less severe storms because the barrier is left open. With a full closure of the barrier, relatively stagnant IWLs will be experienced for longer periods, the level being determined by when the barrier is closed and the severity and duration of the storm surge. With a partial closure of the barrier, the IWL will be more variable than with a full closure.

Dikes can fail in a number of ways. In general, dike safety conditions are worst with high quasi-stagnant water levels for a long duration. Dike safety conditions are better when the IWL is permitted to rise gradually up the dike face during the course of a storm or when it remains at a low quasi-stagnant level for a short time. Different coq.trol strategies result in different IWL versus time profiles and hence, different dike safety impacts.

- Barrier loads originate from the head differences across the barrier-that is, the difference between water level immediately outside and immediately inside the barrier. Even though the barrier is designed to accommodate the most severe loads imposed by any barrier control strategy, it is useful to compare the strategies because they result in different loads and different factors of safety. Those that close the barrier fully near low outside water level (OWL) before a storm surge will result in low head differences while closing but high head differences while closed.

Those that wait to close fully until the OWL is relatively high will result in high head differences while closing but lower head differences while closed. Those that

partially close near low water result in low head differences while closing, and the controlled leakage relieves the head difference at the storm peak.

(26)

5 ,

.

~ --' ,_..J-':- ~/ ~ ~.

r--(~~-

)

'K

-,

<~6~ 2S: ~: .~~

____

r---.)

'

--:

~'---

--

-~~~-

.---._--

~

,

-VA

---::=:=---- ..

--

.::

-

-...-::-:: -

---ri!: Pillar L - - - -40m - - - , _ r

-NAP+15m ---NAP+12m Road ...

----.--~

-

---.

Sill

(27)

6

There are a number of possible ecological impacts. Because of the reduction in tidal amplitude with the open barrier, salt marsh areas will shift downward, creat-ing new salt marshes. To prevent erosion of the salt marshes by continuous wave attacks during storms, mean stagnant basin IWLs should be above NAP

+

2.0 m or below NAP. Intertidal fiats, which lie between mean high water and mean low water, contain almost all of the biomass of the basin and are important feeding grounds for birds, fish, and shrimp. To avoid unnatural damage to the biomass by barrier operation, stagnant water levels below NAP

+

0.2 m should be avoided. The best situation for ecological impacts when the barrier is closed is when all intertidal and salt marsh areas are submerged below the level of the damaging wave attack. Generally, this occurs with water levels above NAP

+

2.0 m, with even higher levels approaching NAP

+

2.4 m being preferred. An intermediate situation exists when a gradual rise of water level occurs over these areas during a storm. The worst situation is when a fixed or quasi-stagnant level below NAP

+

2.0 m exists throughout the storm duration for all storms.

Water management and shipping impacts include polder pumping (i.e., pump-ing water out oflow-Iypump-ing polders into the Oosterschelde basin), harbor operations, and operation ofthe sluices to lakes around the basin. In all cases, there is improve -ment over the present situation without the barrier. But barrier control strategies with higher IWLs create more problems, albeit small, than those with lower IWLs. Polder pumping stations must pump against a higher head, generally at a some-what reduced capacity. Harbor operations will be disrupted at water levels above NAP

+

2.0 m, and there is less flexibility in managing the sluices to the Veere and Grevelingen lakes.

1.4. CONSTRUCTING ALTERNATIVE BARRIER CONTROL STRATEGIES

We constructed several alternative control strategies, each aimed at minimiz -ing different adverse effects or impacts of the barrier. Thus, each control strategy aims for a different balance of impacts.

One major alternative, called the "two-stage E-Ievel strategy," emphasizes ecology and simplicity of barrier operation, while attempting to maintain adequate impacts on dike safety, water management and shipping, and barrier loads. It

aims to maintain a high quasi-stagnant IWL above NAP

+

2.0 m and below an upper limit of about NAP

+ 2

.6 m for relatively short periods over a wide range of storm severities. These conditions are best for ecology, as noted above.

A second, more complex, major alternative, called the "attenuator strategy with ongoing backup," emphasizes the best balance of impacts without seeking to improve on one impact category at the expense of the others. It aims to gradually vary the IWL during a storm, reaching a 'maximum level between NAP

+

2.0 m and NAP

+

2.6 m for a very short period over the same range of storm severities. As we noted above, this results in an intermediate situation for both ecology and dike safety impacts.

A third major alternative, called the "target IWL strategy," was tailored to emphasize dike safety and water management and shipping impacts. Its aim was to maintain a low, quasi-stagnant IWL between NAP

+

0.2 m and NAP

+

0.6 m

(28)

7

over the same range of storm severities. Originally, this range of stagnant water levels was thought to provide a safe ecological "window," but subsequent work indicated that major damage to new salt marsh areas would occur. Also, questions arose about the desirability of quasi-stagnant IWLs on the dikes, albeit low, for the durations involved in typical storms at these water levels.

Thus, these three alternative strategies exemplify very different approaches to weighting the relative importance of each impact category. And each strategy has different operational elements and differs from the others in its simplicity or com-plexity and its reliability of operation, as mentioned above. In the present report we discuss the sensitivity analysis conducted for the strategy classes exemplified by the first two alternatives.

1.5. ORGANIZATION OF THIS REPORT

Chapter 2 discusses barrier control strategies in general, categorizes the bar-rier control strategies considered, and describes each class.

Chapter 3 presents the sensitivity analysis conducted of what we call "E-level on-off strategies." It includes the effects on performance of variation in outside water level trigger, inside water level trigger, closing time, closing method, and total effective aperture of the barrier.

Chapter 4 presents the sensitivity analysis of "attenuator strategies." It treats three classes of strategies with appropriate variations of policy elements within the

(29)

Chapter 2 CATEGORIZATION AND DESCRIPTION OF BARRIER

CONTROL STRATEGIES

2.1. INTRODUCTION AND CATEGORIZATION

We have shown in the Summary Report that, with an open barrier, the mean basin IWL closely follows the OWL in the North Sea. That is, the peak IWL is slightly less than the peak OWL, but nearly proportional to it, occurring 1 to 11/2 hours later.! This is illustrated schematically in Fig. 2.1. To keep the IWL below specified limit levels, which may be desired for safety or other reasons, the barrier must be closed partially or completely when the predicted or observed OWL ex-ceeds some trigger level. We have used the OWL as the primary control signal for

several reasons. For strategies that rely on observed water levels, the fact that the OWL slightly exceeds and leads the IWL gives us a conservative prediction ofIWL compared with measuring IWL directly. And for strategies that rely on predicted water levels, in the past only OWL has been measured and predicted.

Trigger OWL

-Limit IWL

-Outside water level

(OWL)

~~/~--~~--~~----T-im-e---~~--~~-~~~---" Mean aSIn _IWLb '

Fig. 2.1-Water level profiles with open barrier

Thus, a barrier control strategy is designed to control the peak basin IWL and uses the exceedance of a trigger OWL (predicted or observed) as its primary control

signal. In general, a barrier control strategy includes (1) the actions that govern the time and rate of gate closing and opening, (2) the rules behind the decisions for these actions, and (3) the gathering and processing of the information needed for

such decisionmaking.

There are a number of ways of categorizing barrier control strategies. One way, for example, is by whether the barrier is closed fully. In this case, we can distin-guish three categories:

I See R-2444/ 1, Chap. 2. The variation ofIWL with time is computed using the Rand SIMPLIC model,

a fast and inexpensive computer model designed to explore alternative barrier control strategies in different storms. SIMPLIC matches the results within a few centimeters of the Rijkwaterstaat's IMPLIC model, a more detailed and complex mathematical model.

(30)

9

1. On-off strategies, in which the barrier is normally in either an open or closed state with a short transient period in between.

2. Attenuator (or reductor) strategies, in which the barrier is ordinarily operated in a partially closed state to achieve desired effects, such as permitting the basin to fill gradually during a storm. In extreme storms,

however, the barrier may be fully closed.

3. Hybrid strategies, which are mixes of the first two categories.

Another way of categorizing strategies is by distinguishing a primary from a backup strategy. A primary strategy can use either the predicted or observed trigger OWL as its primary control signal. If the primary strategy uses predicted

water levels, which are inherently uncertain, its performance will be uncertain. Thus, it needs a backup strategy that uses only the observed OWL as its trigger. If the primary strategy uses observed water levels as its trigger, it needs no backup

strategy.

For primary strategies that use prediction, there is generally a time sequence

of trigger levels for barrier closure, as illustrated in Fig. 2.2. Suppose the primary strategy is to close the barrier if the predicted OWL exceeds some trigger level, which we call the P-level (P for predicted), or the closing level, or sluitpeil in Dutch. During time period 1, a prediction of peak OWL is made. If it does not exceed the P-level, the operator would wait (time point 2) to see whether the backup (or emergency) strategy is to be implemented. If the observed OWL exceeds a trigger value called E-level (E for emergency), or the alarm level, or alarmpeil in Dutch,

Outside water level

1. Primary strategy:

Close barrier if

predicted OWL> P-Ievel

2_ If primary not used, consider backup

Backup (emergency) strategy:

Close barrier if

observed OWL> E-Ievel

Time

(31)

10

the backup strategy is implemented (time point 3). p- and E-levels need not be the same.2

In this report we elect to categorize strategies by the primary-backup and prediction-no prediction distinctions; Table 2.1 lists all of the closing strategies considered in this study. But in the following two chapters we focus only on two strategy categories (E-level on-off and attenuator). All strategies use the same opening rule: Open at high slack water (HSW), when the IWL equals the OWL.3

Table 2.1

CATEGORIZATION OF BARRIER CONTROL STRATEGIES

On-off strategies

Primary Strategy with Prediction

(Close barrier if predicted OWL> P-Ievel)

• Close at low slack water (LSW) • Close to achieve target IWL

Attenuator strategies: partial close at LSW to reduced barrier aperture • Fixed size (Attenuator C)

• Variable to achieve target max IWL at next high tide (Attenuator A) • Variable to achieve target max IWL over entire storm (Attenuator B) Hybrid strategies: attenuator strategies with on-off strategy as ongoing backup

• Attenuator A with ongoing backup

- After partial close, backup remains in on-line operation • Attenuator C with IWL trigger

Primary Strategy without Prediction or Backup Strategy

(Close barrier if observed OWL> E-Ievel) On-off strategies only

• Close on observed OWL > E-level (called "basic E-level strategy") • Close on. observed OWL > E-level and IWL > trigger IWL (called

"single-stage E-level strategy")

• Single-stage E-level and LSW trigger (called "two-stage E-Ievel strategy")

- If specified LSW level exceeded, reduce E-level for next high tide

2.2. ON-OFF STRATEGIES

Figure 2.3 compares three of the on-off strategies. Two-the low slack water (LSW) and target IWL strategies-use prediction. The third-close on observed OWL ~ trigger OWL

=

E-level-does not use prediction and thus can be a primary or a backup strategy; we call it the "basic E-level strategy."

2 The implications of choosing different combinations of P- and E-Ievels are discussed in Vol. I. Basically, a specific choice for each sets the maximum IWL with the open barrier, the average "neces-sary" closure frequency, the fraction of necessary closures that occur with the primary and the backup strategies, and the frequency of "unnecessary" closures. See Chap. 3 of that volume.

3 Other opening rules, such as remaining closed if the next predicted OWL exceeds P-Ievel, were

examined as part of the study. But except for special requirements, they do not perform any better than the simple rule described above.

(32)

trigger

OWL 11

Open

Fig. 2.3-Comparison of basic on-off barrier control strategies

The LSW strategy uses the closing rule: Close at LSW before a predicted exceedance of the trigger OWL (the P-level). It is designed to achieve a low IWL. Because it requires closure at LSW, the prediction requirements (i.e., the "look-ahead" prediction time) are the most severe of the three strategies. Figure 2.3 also shows that the head difference (i.e., the difference between outside and inside water levels) while closing is lowest and near zero because closure begins at LSW.4 But the head difference while closed is highest because final inside water is lowest.

The target IWL strategy uses the closing rule: Close to achieve a specified intermediate (we used NAP

+

20 cm to NAP

+

60 cm) target IWL when there is a predicted exceedance ofthe trigger P-level. Because the target IWL in this strate-gy is in between those of the LSW and basic E-level rules, the required prediction time is less than for the LSW strategy because one waits to begin closure some time after occurrence of LSW. Head difference while closed is lower than for the LSW closure, and some appreciable head difference occurs while closing. 5

The basic E-level strategy is simplest and uses the closing rule: Close when the observed OWL exceeds the trigger E-level. Because the final IWL is highest of the three strategies, head difference while closed is lowest and head difference while closing is highest5 _{for this strategy.}

Of the on-off strategies we considered that do not use prediction, the basic E-level strategy is the simplest because it uses a single trigger (observed exceed-ance of trigger OWL) as its control signal. As discussed below, this strategy has been successively refined to achieve tighter control of final IWL over a wide variety of storms that differ in peak OWL and in OWL rise rates.

The first refinement is to add to the basic E-level strategy a second trigger that monitors IWL; we call this the "single-stage E-level strategy." Both trigger levels

• Traditionally, sluices in the Netherlands operate at slack water to avoid head differences while closing and opening.

S The ranking in head difference while closed between the basic E-level and target IWL strategies