Suzhen Feng

Suzhen Feng

Auto-generated profile, based on publicly available information

Suzhen Feng

College of Automation and Electronic Engineering Qingdao University of Science and Technology Qingdao China

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
13 June 2022 in Journal of Flood Risk Management
Verified
Journal Article
Open access
ZYZetai Yang
Zetai Yang
Hao ZhengHao Zheng
Hao Zheng
SFSuzhen Feng
Suzhen Feng
CCCheng Chen
Cheng Chen
JWJinwen Wang
Jinwen Wang
This paper presents a multireservoir flood control model that incorporates operating rules to alleviate the risk that results from traditional flood control models. The model is accurately reformulated into one that can be solved with the mixed integer linear programming (MILP) and approximated with a two‐stage linear programming (TSLP) to speed up the solution by excluding all the binary variables. A scroll decision‐making strategy is proposed by assuming only a few days of future inflows being predicted with certainty. With 10 historical and 30 designed floods in different magnitudes, the models and methods are applied to the Yangtze River basin, which includes 11 cascade reservoirs. The results on average show the MILP reduces the rule violation by 17%, while the TSLP is more than 150 times faster in CPU time. Excluding one of the reservoirs from the simulation reveals that the Three Gorges (TG) contributes the most in flood detention, accounting for 76% on average for once‐in‐a‐century floods. Interestingly, with the scroll decision‐making strategy implemented, the change from 3 to 7 days of certain forecasting has the most significant reduction in both the rule violation and flood detention, indicating that the forecasting accuracy of more recent inflows is more valuable for flood control.

19 December 2025 in Water Resources Research
Verified
Journal Article
Open access
SFSuzhen Feng
Suzhen Feng
Hao ZhengHao Zheng
Hao Zheng
DGDunwei Gong
Dunwei Gong
JSJing Sun
Jing Sun
JWJinwen Wang
Jinwen Wang
Global climate change and growing water demand exacerbate the imbalances in reservoir resource allocation, necessitating advanced frameworks that move beyond static valuation methods. Traditional valuation methods, constrained by static or homogenized assumptions, fail to capture the spatiotemporal heterogeneity and dynamic trade‐offs inherent in cascade reservoir operations. To address this gap, this study develops a novel marginal value framework based on shadow pricing to assess the marginal values of multidimensional resources—water, storage capacity, and turbine capacity—in weekly hydropower scheduling. The key methodological contribution is overcoming the duality gap challenge in mixed‐integer linear programming (MILP) models by proposing two practical methods: Method‐I fixes integer variables to derive dual multipliers via linear programming, while Method‐II computes shadow prices through perturbation analysis. Validated on 26 cascaded reservoirs in Yunnan, China, both methods yield consistent results, with Method‐I demonstrating superior computational efficiency. Key findings reveal that: (a) irrigation water values in dry seasons and upstream regions exceed wet seasons and downstream by 1.11–7.34 times; (b) Nuozhadu Reservoir's storage capacity shadow price peaks in week 44, signaling flood control‐power generation trade‐offs; (c) Wunonglong and Dachaoshan exhibit the highest marginal turbine capacity values for spillage reduction; and (d) reserve capacity costs surge by 32%–45% in weeks 36–37. This work bridges the fields of resource economics and hydraulic engineering, providing actionable insights for dynamic water pricing, infrastructure investment prioritization, and seasonal ancillary service markets.

17 February 2025 in Journal of Flood Risk Management
Verified
Journal Article
Open access
ZYZetai Yang
Zetai Yang
SFSuzhen Feng
Suzhen Feng
KFKaixiang Fu
Kaixiang Fu
JWJinwen Wang
Jinwen Wang
The concomitant vibration of flood discharge, which would cause structure damages to hydraulic infrastructure and thus incurs threats to nearby communities, has rarely been addressed yet cries for an effective solution in discharge scheduling of sluice gates. This work improves on the traditional practice (Model‐I) that mainly aims to restrain start‐up and shutdown actions of spillway gates with a new model (Model‐II) that includes a flexible vibration damping rule, in which the sluice gates are grouped in priority to be sequentially committed, and in the same group, a reference gate is prioritised to enforce a uniform discharge from the active outlets and the gates are paired to ensure a symmetrical opening. The case studies in the Xiangjiaba Dam (XD) demonstrate the excellent adaptability of the model to gate opening patterns concluded with field experiments and site monitoring, and comparing the two models reveals that Model‐II can enforce preferable operational rules to deliver safer discharge scheduling and potentially to reduce risks from the concomitant vibration of hydraulic facilities and the turbulent flow field around the dam during flood discharging, though leading to a much higher frequency of starting up and shutting down of sluice gates than the traditional Model‐I.

2 June 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
LYLili Yu
Lili Yu
JWJinwen Wang
Jinwen Wang
SLShuangquan Liu
Shuangquan Liu
TMTeng Ma
Teng Ma
SFSuzhen Feng
Suzhen Feng
QMQianqian Ma
Qianqian Ma
This paper compared the performance of cyclic coordinate method (CCM), shuffled complex evolution (SCE-UA), and an evolution strategy-based approach in calibrating the conceptual rainfall-runoff (CRR) models. This evolution strategy method is a new algorithm proposed in this work, called the boundary adjustment evolution strategy (BA-ES). Four widely used hydrological models and four hydrological series with different year-types were selected. Each algorithm, model, and data combination constituted an experiment, and each experiment calibrated 10 thousand sets of optimized parameters. Statistical comparative analysis was conducted on the performance and its influencing factors of algorithms. The results show that CCM has highest calibration accuracy, strongest adaptability, and is the simplest and easiest to use; BA-ES has high calibration accuracy, fastest speed, simple principle, and is easy to use; SCE-UA is inferior to CCM and BA-ES in calibration accuracy and BA-ES in speed, and its intrinsic mechanism is complex and difficult to master, although it excels in stability of calibration results. The performance of algorithms is directly proportional to the models’ ability to fit hydrological laws of basin, and has no significant relationship with the complexity or dimensionality of models. The year-types of hydrological series affects the performance of algorithms by influencing the performance of models. The algorithms performed better in year-types that the models were easier to fit, and vice versa

23 September 2026 in Research Square
Verified
Preprint
Open access
PDF file
JWJinwen Wang
Jinwen Wang
SFSuzhen Feng
Suzhen Feng
HZHao Zheng
Hao Zheng
CCCheng Chen
Cheng Chen
DGDunwei Gong
Dunwei Gong
JSJing Sun
Jing Sun
Water can provide multiple services within the same managed system, but the economic value of a unit of water depends on which service it enables and when it is available. This makes the marginal value of water difficult to quantify when competing uses must be assessed on a common economic scale. Existing approaches generally value individual water services within separate sectoral frameworks, limiting direct comparison of the marginal contribution of the same water stock across uses. Moreover, when these services are represented as hierarchically weighted objectives in an optimization model, redundant differences in objective-weight magnitudes can distort shadow prices without changing the underlying optimal solution. Here we develop a compact-weight commensuration framework that removes this redundant scaling while preserving the priority structure and optimal operating solution of the original model. The resulting shadow prices provide a common economic metric for the marginal value of water across competing services and system states. Applied to the Xiaowan cascaded-reservoir system, the framework substantially reduces orders-of-magnitude disparities in conventional shadow prices without altering the optimal operating state, and reveals distinct seasonal patterns in the value of inflow, storage, energy and capacity. The results show that reservoir water is better understood not as a fixed-volume commodity, but as a portfolio of competing services whose marginal value changes with scarcity, timing and system requirements. This common valuation framework provides a quantitative basis for comparing alternative water uses and informing integrated reservoir allocation and cross-sectoral water-service markets.