Elsevier

Applied Thermal Engineering

Volume 70, Issue 2, 22 September 2014, Pages 1180-1188
Applied Thermal Engineering

Flexibility to seasonal demand variations in pulp mill steam production: The effect of steam savings leading to off-design heat loads

https://doi.org/10.1016/j.applthermaleng.2014.04.059Get rights and content

Highlights

  • A multi-period model for optimization of pulp mill steam production is proposed.

  • The model accounts for boiler minimum load limits and variations in steam demand.

  • Steam savings combined with seasonal variations reduce the flexibility of the utility system.

  • Results of the multi-period and a conventional annual-average model are compared.

  • The importance of using the multi-period approach is demonstrated.

Abstract

This paper focuses on the steam production in a chemical pulp mill that is retrofitted to reduce its process heating demand. A multi-period optimization model for design decisions is proposed that takes into account operational limits of the boilers and variations in heat demand. Large variations in combination with the retrofit cause off-design loads that affect the flexibility of the steam system. The minimum boiler load limits will be a greater constraint on operation when the average load of the boilers is moved closer to the minimum for longer periods of time. As shown in this paper, a conventional approach that considers fixed annual averages of process parameters therefore risks leading to sub-optimal solutions because of neglecting the variations in heat demand and inaccurately modeling the operational limits. The multi-period approach suggested in this paper considers operational flexibility associated with different designs. A case study based on a Kraft pulp mill with a recovery boiler and a bark boiler shows the benefit of this approach. Four scenarios for heat savings and lignin prices are analyzed. Numerical results are presented that compares the solution of the multi-period model with that of a conventional annual-average approach. Differences in designs, energy balances and economic performance are demonstrated.

Introduction

When making decisions about process retrofits for energy savings at an industrial plant, potential operational flexibility towards variations in, for example, heat loads and energy prices should be considered. When process variations cause the load of certain process equipment to approach their minimum and/or maximum operational limits valuable operational flexibility might be lost. The risk of being constrained by such operational limits might increase in a retrofit situation leading to deviations from the original design conditions. Retrofit energy savings will, for example, lead to a reduction in heat load of the steam production units at the plant, causing their average load to approach their minimum load limits.

This paper focuses on variations in process parameters. Variations in energy prices or, for example, carbon prices can also be essential to consider if processes that are flexible in operation with regard to such changes are considered (see e.g. Ref. [1]). Siitonen and Ahtila [2] studied the effect of operational flexibility towards fluctuating carbon prices for energy savings in a pulp and paper mill and showed that its economic value can be significant. Nemet et al. [3] optimized the design of a heat exchanger network over its full lifetime by considering future utility price variations in a multi-period approach.

This study analyses a retrofit project in an existing pulp mill. The purpose of the retrofit project is to reduce the heat demand of the plant. However, in order to assess the value of the steam savings, it is necessary to determine how the steam production is most profitably adjusted in response to the savings.

The pulping process is continuous and typically designed for maximization of quality and throughput of one core product. The operational objective is to maintain the production as close to the design capacity as possible. Consequently, pulp mill energy systems are traditionally modeled using annual averages representing values very close to design conditions. However, recent and expected future changes in wood, pulp and energy market conditions motivate a shift towards producing a larger variety of products including traditional pulp mill energy by-products such as electricity and heat, and emerging lignocellulosic biorefinery products such as different kinds of materials and chemicals (see e.g. Ref. [4]). This transition from the traditional pulp mill to a biorefinery will connect the pulp mills to an increased number of external markets. Also for the traditional by-products of heat and electricity, an increased implementation and production rate can be expected when, for example, energy prices rise. Consequently, mills are likely to become affected by more sources of variations. In combination with the opportunities connected to the diversified product portfolio of a pulp mill biorefinery, this should provide a higher value to technological options that provide flexibility in process operation than what would be seen for the typical pulp mill today.

Methodologies for the design optimization of utility systems with varying demands need to simultaneously consider both design and operational decisions. Several such methodologies have been published in literature. Maia and Qassim [5] used a simulated annealing algorithm to solve the synthesis problem with time-varying demands. However, most published methods rely on a multi-period, mixed-integer linear programming (MILP) formulation. Hui and Natori [6] suggested a model for the optimization of the utility system operation including design decisions by considering both existing and new power generation equipment. Iyer and Grossmann [7] formulated a MILP model for the multi-period synthesis and operational planning of the utility system and proposed a bi-level decomposition algorithm for effective solution of the problem. Marechal and Kalitventzeff [8] used a genetic algorithm to identify the minimum number of operating periods needed to describe the yearly demand variations with sufficient detail and then optimized the synthesis and operation of the utility system using a multi-period MILP model. More recently, the focus has been increasingly directed towards improved modeling of energy equipment performance. Varbanov et al. [9] proposed improved models for steam and gas turbines in part-load operation. Shang and Kokossis [10] considered the performance of turbines and boilers to depend on size, load and operating conditions in their approach to synthesis of utility systems with varying demands, in which they rely on thermodynamic targeting models to reduce the problem to a reasonably sized MILP formulation. Aguilar et al. [11] also considered part-load operations and varying energy demands in their generic modeling framework for utility systems, in which they obtain linearity by starting from the development of linear models for boilers and turbines. Recent advances also include the modeling of variations in steam header properties, either as pre-determined parameters [11] or as variables to be optimized [12]. Common for the cited studies are their general applicability for optimization of complex networks of a wide range of heat and power production units.

In contrast, the present work suggests a simplified, but nonetheless multi-period approach for the specific application to a chemical pulp mill retrofit. A MILP model is suggested for the optimization of design and operating decisions in the steam production system at an existing pulp mill in response to a process heat savings retrofit. The model is deliberately kept simple with regard to, for example, part-load efficiency, linearized investment cost functions and pre-determined steam header properties. The intent is to help enable its integration with more complex, strategic decision-making models that cover not only decisions related to the utility system, but also decisions about the level of energy savings and decisions about integration of new technology and processes at the plant. It should, for example, be possible to integrate it with a model for strategic decision-making under uncertainty [13], possibly also considering multiple objectives [14]. Nonetheless, a multi-period modeling approach has been chosen in order to account for variations in process heat demand. Explicit modeling of operational constraints of the boilers has also been included.

The utility system studied here is different from earlier studies also in its application to a chemical pulp mill. The main steam producer in the pulp mill is the recovery boiler, which does not only fill the purpose of utility production, but also the recovery of process chemicals. It is therefore an important part of the actual process, not only the utility system, and its operational flexibility is more strongly constrained than of a conventional boiler. The recovery boiler does therefore not straightforwardly fit into generic boiler models. Furthermore, this study includes the possibility of investing in lignin separation (see e.g. Ref. [15]), an emerging technology for the pulping industry, for which operating performance data is not yet readily available. Extracting lignin would provide the pulp mill with an opportunity to generate new by-products, but it has also been shown to be important for debottlenecking the recovery boiler in case of a pulp production increase [16], [17]. This paper shows that it could also provide a great opportunity for indirectly increasing the flexibility of the pulp mill utility production.

The aim of this paper is to illustrate the importance of modeling the process variations and operating load limits instead of taking the simplifications one step too far by modeling a single-period, fixed-value problem. The results demonstrate the potentially large errors in unit sizes, energy balances and economic results that can arise if a problem is inadequately simplified to average values. The novelty of the paper is thereby the application of a multi-period modeling approach to strategic decisions in non-conventional technology in an industrial plant (i.e. the pulp mill) that traditionally is studied using annual averages only.

Section snippets

Studied system – a Scandinavian pulp mill delivering heat to a district heating network

An overview of the utility system of a typical pulp mill is shown in Fig. 1. The main steam producer in a chemical pulp mill is the recovery boiler, which purpose is twofold: Recovery of energy and chemicals from the black liquor, which consists of the wood by-products that are not used for the pulp production and the used cooking chemicals. The load of the recovery boiler is determined by the demand for chemical recovery set by the pulp production rate.

Variations in steam demand are therefore

Optimization model

The MILP model is used to identify how the steam production at the pulp mill should best be adapted to meet a reduction in process steam demand. A design decision (investment in lignin extraction) is optimized, considering the effect of varying operating conditions.

Optimal steam production after steam savings retrofit

Fig. 4 illustrates the energy situation at the mill after HP steam savings of 15 kg/s and optimally adjusted boiler operation, including investment in a lignin extraction plant. With a low lignin value (Fig. 4a) the optimal solution implies an investment in lignin extraction despite the fact that bark savings are prioritized in operation. The capacity for lignin extraction is used to reduce the load on the recovery boiler and to reduce the steam excess that is vented to atmosphere after the

Discussion

The pronounced effect of the variations in this case study is partly explained by the magnitude of the variations in relation to the average load, that is, the district heating network is small in relation to the excess heat deliveries from the pulp mill. In a larger district heating grid, the industrial excess heat would constitute a constant base load instead of, as in this case, the dominant heat production source covering almost all the variations in district heating demand.

In this study,

Conclusions

When assessing the economic value of an industrial heat-saving retrofit projects, in the presence of significant heat-load variations, these should be explicitly modeled in order to accurately value flexibility to such variations. This paper shows that failing to do so can lead to errors above 50% in equipment load changes and economic results. If investments can be made that will enable more efficient ways of adjusting the operation of the mill's energy system to the varying conditions, then

Acknowledgements

The work has been carried out with financial support from Chalmers Energy Initiative and from the Swedish Energy Agency (37308-1). The author would like to thank Professor Thore Berntsson for valuable comments on the work and for reviewing the manuscript.

Cited by (6)

  • Understanding steam costs for energy conservation projects

    2016, Applied Energy
    Citation Excerpt :

    A near-optimal solution was also analysed [11] to achieve economic and environmental optimization [12]. The optimization under uncertainty has been studied, including multi-period synthesis [13] and optimization [14], flexibility analysis to seasonal demand variations [15] and market fluctuation [16], etc. While it is possible to calculate the saving as a heat load and a flowrate of steam, it is not so straightforward to relate this to the actual economic savings.

  • Optimal Multi-Period Investment Analysis for Flexible Pulp Mill Utility Systems

    2015, Computer Aided Chemical Engineering
    Citation Excerpt :

    The steam header data for high-pressure (HP), medium-pressure (MP) and low-pressure (LP) steam are pre-determined. Potential costs or difficulties associated with daily variations in the operation of the studied processes and steam production units are neglected (see discussion in Svensson, 2014). For the boilers, part-load operation is represented by the use of marginal efficiencies that are estimated to be valid as long as the boiler load is kept above the minimum load limits set in the model.

  • A steam utility network model for the evaluation of heat integration retrofits - A case study of an oil refinery

    2017, Journal of Sustainable Development of Energy, Water and Environment Systems
View full text