Introduction

Industrial sustainable development is becoming more critical as concerns about global climate change and pollution increase. This is especially true in developing countries, where unreasonable industrial structures are severely hindering low-carbon sustainable development. In 1987, the United Nations Brundtland Commission defined sustainable development as meeting the needs of the present without compromising the ability of future generations to meet their own needs (UNBC 1987). In industrializing countries, the share of industrial production in their GNP is rapidly increasing (Peter et al. 2006). Therefore, it is not surprising that this accelerated industrialization, which requires a dramatic transformation of the industrial sector, has given rise to some serious environmental problems (JEC 2003).

For most developing countries, modernization is considered a primary aim and, therefore, rapid economic growth is necessary to meet these goals, but this increased growth has resulted in an equal growth in carbon emissions, mainly because energy technologies are significantly underdeveloped. However, protecting the global environment is the duty of all people, it is impractical to require developing countries to reduce carbon emissions by sacrificing economic development (Zhang 2009; Ma et al. 2012). Therefore, to find a practical method for developing countries to take part in protecting the global ecological environment while continuing to develop their economies, it is important that they develop their own specific ecological economics (Shi 2002).

China, as one of the economically most expanding countries in the world, has dramatically improved its living standards (Jin 2011; Liu et al. 2010). China’s economy has enjoyed average growth rates close to 9 % over the last quarter of a century (World Bank 2001). However, as a result of this extremely rapid economic growth and the anthropocentric approach to the human-nature relationship, the seriousness of environmental problems has escalated. Consequently, environmental problems are threatening China’s sustainable future. In 1995 the proportion of heavy industry was 52.7 % which rose to 71.1 % in 2008 (Zhang and Xue 2011). The World Bank estimated that economic damage caused by pollution in China costs around 54 billion RMB annually, which is close to 8 % of domestic GDP (World Bank 1997).

China cannot continue with an economic development model that causes significant adverse effects on the environment (Costanza and Ruth 1998). In the immediate future, China needs to better integrate environmental and social costs into its economic development strategy and modify development strategies to reduce carbon emissions (Kenzheguzin and Yessekina 2004).

Since the regional industrial structure is constantly changing, it is necessary to find a dynamic simulation method to measure these changes and offer solutions for the future. System dynamics (SD) approaches as a modeling tool provides a flexible way of building simulation models from causal loops or stock and flow diagrams. Therefore, to explore the potential long-term ecological, economic, institutional, and social interactions in sustainable industrial structural development, the SD approach is the main methodology used in this article.

This article proposes a sustainable industrial restructuring development-oriented model using a new system dynamics multi-objective programming (SD-MOP) model which seeks to balance environmental protection and economic growth. A case study of the South West Chinese city of Leshan is presented to illustrate the difficulties of accommodating the economic and ecological development needs. The remainder of this article is organized as follows. “Literature review” section is a literature review looking at related research. “Problem background” section analyses the problem background and a new model is designed in “Sustainable planning modeling” section . In the modeling process, fuzzy random factors are applied to ensure a more accurate calculation. In “A case study” section, a case study is presented and sustainable development scenarios conducted based on the proposed model. “Simulation results” section  presents the empirical results, and “Discussion and policy proposals” section presents discussion and corresponding policy recommendations. “Conclusion” section presents the conclusion.

Literature review

In this section, a brief review on industrial structure analysis, system dynamic analysis, and fuzzy random analysis is given.

Industrial structure analysis

Industrial structure theories have been proposed for years, but most of these studies have been conducted in developed rather than developing countries. Clark (1987) summarized the changing industrial structures and testified to their importance to economic development. Chenery (1961) standardized open industrial structure theory and put forward a development theory. Lewis (1954) proposed a binary structure transformation theory to explain the economic issues in developing countries, and showed that developing countries could make use of their labor resources to accelerate economic development. Cabezas et al. explored the 12-compartment model for different scenarios in which an industrial process was added to the framework (Cabezas et al. 2005a, b), but pointed out that this model was only applicable to dynamic systems in a cyclic steady state.

A wide range of studies have presented forecasts of industry optimization actions using statistical and econometric approaches (Zhang et al. 2012; Zhou et al. 2012; Heston and Rouwenhorst 2007; Tomasz 2011). However, many of these had limitations. First, while each model emphasized some aspects of economic reality, many other areas were omitted. Moreover, most of these models were unable to offer results for more than one scenario. Second, only a few theoretical models have been constructed to take account of the dynamic tradeoffs between economic growth, energy consumption, pollution, and human efforts in protecting the environment within a compact framework. Therefore, it is necessary to construct dynamic models which include time-dependent motion for the key variables.

System dynamics analysis

The system dynamics approach has provided a theoretical and practical foundation for modeling complex systems in a learning environment, so it can be viewed as an effective approach through which different processes can be examined and different scenarios can be tested from a systems perspective (Costanza et al. 1993; van den Bergh and Nijkamp 1994; Barton et al. 2000). The purpose of the system dynamics modeling approach is to obtain an understanding of, and insights into, system relationships and search for alternative polices to improve the situation (Shi and Roderic 2005).

System dynamics was established on the basis of feedback relationship systems. It is a computer-aided approach for analyzing and solving complex problems with a focus on policy analysis. This method, established in the mid-1950s by MIT Professor Forester (Barton et al. 2004), is based on a summarization of operational research and a synthesis of system theories including cybernetics, information feedback theory, decision-making theory, information theory, system dynamic simulations, and computer science (Xu and Li 2009).

The SD approach has also been applied to a number of studies related to the environment; environmental impact assessment analysis, solid waste management analysis of GHG emissions, environmental planning and management, etc. (Anand et al. 2006; Guneralp and Barlas 2003; Naill et al. 1992; Vennix 1996; Dale 2003).

Problem background

In the industrial structural system there is a complicated interdependence between the many variables, with a long-run bidirectional causal relationship. The SD model in the above studies has mostly been used as a simulation tool. However, there are two problems which are still unsolved in the traditional SD model.

First, much of the research omits the long-term objectives needed to achieve a long-run simulation and optimization. Most previous studies have focused on micro-industrial processing programming (Shadiya et al. 2012) which neglects the complete industrial structural system optimization. Second, few previous studies consider multi-objective characteristics of the industrial management system. The planner need to seeks to satisfy more than one goals in real decision process because many factors such as economic environmental and society factors may significantly affect industrial system in their models. As a response, we used a multi-objective programming method to optimize the operation. Further, there are many uncertainties when setting the objectives in a real situation. Using one kind of variable to describe uncertainty is very important, so it is necessary to construct dynamic models which also consider the fuzzy random variables. Thus, it can be assumed that these uncertain parameters are fuzzy random numbers and are optimized under uncertainty using a combined chance-constrained programming approach.

The choice of Chinese dual world heritage (DWH) city concentration data for our analyses has several advantages. Firstly, China is the world’s largest developing country. A better understanding of industrial restrictions in China has significant policy implications. Secondly, following the intensification of urbanization and industrialization, the contradictions between urban construction and environmental protection are becoming more acute in DWH cities. The rationalization of industrial structures is crucial to their economic development, cultural heritage protection, and social progress. Thirdly, for those traditional industrial cities that wish to follow the Chinese sustainable development strategy, the lessons obtained can also be beneficial.

China has four DWH regions recognized by the United Nations: Mount Tai in Shandong, Yellow Mountain in Anhui, Wuyi Mountain in Fujian, and the Sichuan Leshan Emeishan scenic area (see Fig. 1). The DWH regions in China are shown in Fig. 1. Of these four areas, only Leshan is in an urban area experiencing significant urban growth, so a study in this city which focusses on sustainable economic development is important. Thus, Leshan has been chosen as our case study.

Fig. 1
figure 1

The double world heritage regions in China

Full size image

Sustainable planning modeling

As mentioned earlier, in order to predict regional industrial economic development, a SD-MOP method is used as the main approach.

System dynamics model

First, we define the parameters used to describe and analyze the system. The industrial structure SD model is composed of four sectors; population, capital, resources, and pollution; which represent the societal, the economic, the energy, and the environmental systems, respectively.

In the societal model sector, the industrial sustainability level is determined by the employment increment, the total population, the birth rate, death rate, and average lifespan. There is close correlation between the industrial structure and the employment structure. Scholars such as Fisher (1939) and Clark (1940) have stated that the labor force moves from primary to secondary, and then to the tertiary as the industrial structure develops. For example, China has experienced a declining labor force growth rate mainly due to its population policy but has benefited from a better quality labor force in response to an increasing demand for more skilled and specialized workers in recent years. Both this labor force shift and quality change could contribute to industrial upgrading, but an unbalanced human resource allocation could lead to an unbalanced industrial structure.

In the economic sector, continuing capital investment and consumption affects per capita living standards, influencing the social sector, and pollution emissions which forms negative feedback loop subsystems. At the same time, environmental capital spending influences the environmental subsystem leading to a positive feedback loop.

In the environmental model sector, besides the environment inside the subsystem of the negative reverse feedback loop mentioned above, because of the effect of pollution on the health and wellbeing of the population (measured through births and deaths), and the natural environment, a negative feedback loop is formed. The fuzzy parameters used in this sector (e.g., environmental pollution per unit area) are described in “Multi-objective programming” section.

In the energy sector, energy consumption is mainly influenced by population growth, the gross domestic product (GDP) incremental rate, and energy-saving factors, as well as the existing resources subsystem inside the negative feedback loop. Low industrial optimization because of a reduction in resource consumption results in a positive feedback loop. At the same time, a reduced standard of living also reduces investment which affects the capital subsystem forming a negative feedback loop. The fuzzy parameters used in this sector are also described in “Multi-objective programming” section.

The purpose of describing these sectors is to clarify issues in the model, and to give a clear picture of the key issues and how they interact (Shi and Roderic 2005). There are both conflicts and unity in sustained economic growth, resulting in increased resource consumption and pollutant emissions. Economic growth is often accompanied by energy mining and consumption growth, both of which damage the ecological environment and increase pollutants. Conversely, decreasing energy consumption and pollutant emissions without major changes in the engineering levels means that energy mining and consumption are also reduced, which consequently restricts the speed of economic growth. Achieving the goal of pollution emission reduction, while at the same time maintaining economic growth, is the main purpose of sustainable economic development.

As mentioned earlier, after the analysis of the relationships of the model sectors, we construct a flow diagram to quantitatively simulate the system (see Fig. 2). The level variables are shown as rectangular boxes which represent the accumulated flows to that level. The source and sink of the structure are represented by a cloud. The cloud symbol indicates infinity and marks the boundary of the model. Total GDP, energy consumption, pollution emission and total population are the variables focused on and are also shown as rectangular boxes.

Fig. 2
figure 2

Stock flow chart of system dynamics model

Full size image

Multi-objective programming

Using sensitivity analysis, the parameters exerting the greatest influence on output are screened out and a MOP method is used to determine the best parameters. The variables related to the model are shown in Table 1.

Table 1 Main variables related to the model
Full size table

MOP is the process of simultaneously optimizing two or more conflicting objectives subject to certain constraints. It has been applied in various fields and has been generalized to uncertain cases (Xu and Zhou 2011; Bamufleh et al. 2013). In our system, maximizing the total GDP, minimizing pollution emissions, and energy consumption is an example of a MOP problem. In this article, the above SD-MOP integrated framework is adopted for the assessment of economic benefit and the mitigation of pollution emissions at a regional level.

Fuzzy random extensions

In real-world problems, uncertainties expressed as discrete interval numbers, fuzzy sets, or probability distributions may exist in many system components and affect decision processes. These uncertainties may arise because of the randomness in various natural processes, or because of errors in the modeling parameters. In these cases, fuzzy random variables are an effective tool when dealing with insufficient historical data and are able to make the problem more tractable. To treat the fuzzy random coefficients, a minimum target value under chance programming is chosen to describe the total energy consumption and pollution emissions.

The main advantage of the chance-constrained programming technique is that it is relatively easy to formulate and the problem size of the resultant deterministic equivalent does not increase even for a large number of uncertain parameters. However, chance-constrained programming requires feasible solutions with at least some probabilities specified on the constraints with uncertain parameters (Kim and Diwekar 2002). Thus, these constraints are controlled within a certain numerical range prescribed by the decision makers.

In the programming, the optimization problems are defined as follows:

$$min\bar{f_{1}}$$

subject to \(Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{e}}_{i}x_{i}\leq\bar{f}_{1}\}}\geq\alpha_{1}}\}\geq\beta_{1}\)

$$min\bar{f_{2}}$$

subject to \(Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{c}}_{i}x_{i}\leq\bar{f}_{2}\}}\geq\alpha_{2}}\}\geq\beta_{2}\)

The two objectives are to minimize the energy consumption and pollution emission. In general, it is difficult to get the absolute minimal value because of the existence of the fuzzy random coefficients \(\tilde{\bar{e_{i}}}\) and \(\tilde{\bar{c_{i}}}.\) By proposed the chance-constrained programming model to deal with the programming, we can get the minimal energy consumption and pollution emission under the probabilistic level α and the possibility level β.

Where \(\bar{f}\) is the objective function, x i is the set of decision variables, Pr is the probability and possibility measure of the given probability space of the uncertain parameter, and \(\alpha\in(0,1)\) is the probability level with which each of the entire constraint set \(\sum\limits^{3}_{i=1}\tilde{\bar{e}}_{i}x_{i}\leq\bar{f}_{1}\) and \(\sum\limits^{3}_{i=1}\tilde{\bar{c}}_{i}x_{i}\leq\bar{f}_{2}\) need to satisfy. \(Pos\{\cdot\}\) denotes the possibility of the event. This means a point x i is feasible if and only if the possibility measure of the set \(Pr\{\sum\limits^{3}_{i=1}\tilde{\bar{e}}_{i}x_{i}\leq\bar{f}_{1}\}\geq\alpha_{1}\) and \(Pr\{\sum\limits^{3}_{i=1}\tilde{\bar{c}}_{i}x_{i}\leq\bar{f}_{2}\}\geq\alpha_{2}\) are at least β. The higher the value of α and β, the more reliable the system is with respect to the uncertain parameter.

Objective function

The MOP method is used to determine the best parameters, and is constructed with energy consumption, pollution emissions, and economic benefit optimization as the objective functions. The specific target functions are as follows.

Energy consumption

Minimizing energy consumption is the best way to deal with the problems mentioned. A minimum target value under chance-constraints is chosen to describe the total energy consumption. Thus, we get the following objective function

$${min\bar{f}}_{1}, $$
(1)

where \(Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{e}}_{i}x_{i}\leq\bar{f}_{1}\}}\geq\alpha_{1}}\}\geq\beta_{1}, \,Pos\) means the fuzzy possibility measures, Pr means the probability measures.

Pollution emissions

As pollution data is uncertain we use fuzzy random coefficients. The minimization of environmental pollution is the goal. The objective function with a fuzzy random goal is as follows

$${min\bar{f}}_{2}, $$
(2)

where \(Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{c}}_{i}x_{i}\leq\bar{f}_{2}\}}\geq\alpha_{2}}\}\geq\beta_{2}, \,Pos\) means the fuzzy possibility measures, Pr means the probability measures.

Total economic benefit

The total economic benefit is the summation of the economic benefit of various industries. The objective function is as given below

$$\max f_{3}=\sum\limits^{3}_{i=1}b_{i}x_{i}, $$
(3)

where b i unit economic output of the i industry.

Constraint conditions

The following are the constraints for the industrial structure.

Capacity constraints

From an analysis of the long-term economic growth data in each country, annual GDP growth moves should be less than 0.15. If the rate of economic growth is over this range, the complete economic system faces the danger of imbalance. The secondary industry sector is a developing country’s core sector, so the constraint is set to regulate the speed of secondary industry sector development.

$$0.85x_{2(j-1)}\leq x_{2j}\leq1.15x_{2(j-1)} $$
(4)
Unemployment rate

Employment rates are an important index affecting social stability. According to China’s policy, urban unemployment should be held below 0.05.

$$L_{i}-\sum\limits^{3}_{i=1}p_{i}x_{i}< 0.05 $$
(5)

L i represents the total population in i year.

Energy intensity

According to Chinese policy, annual energy consumption intensity should decrease by more than 0.05.

$$\frac{\sum \limits^{3}_{i=1}e_{ij}x_{ij}}{x_{j}}\leq \frac{\sum \limits^{3}_{i=1}e_{i(j-1)}x_{i(j-1)}}{x_{j-1}}(1-0.05) $$
(6)
Pollution emissions

The environmental pollution constraint should satisfy the expected value constraint, so annual pollution emissions are expected to decline.

$$\frac{\sum\limits^{3}_{i=1}c_{j}x_{j}}{x_{j}}\leq\frac{\sum\limits^{3}_{i=1}c_{j-1}x_{j-1}}{x_{j-1}} $$
(7)
Total industry proportion

Because the industry data represents the proportion of each industry sector, respectively, together they equal 1.

$$\sum\limits^{3}_{i=1}x_{i}=1 $$
(8)
Non-negativity

The departments’ outputs of the model are non-negative.

$$x_{i}\geq0 $$
(9)

Thus, with the above considerations we derive the fuzzy random chance-constrained model (FRCCM) as follows

$$\left\{ \begin{array}{l} \min\bar{f}_{1}\\ \min\bar{f}_{2}\\ \max\, f_{3}=\sum\limits^{3}_{i=1}b_{i}x_{i}\\ {\text{s.t.}}\left\{\begin{array}{l} Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{e}}_{i}x_{i}\leq\bar{f}_{1}\}}\geq\alpha_{1}}\}\geq\beta_{1}\\ Pos\{{Pr\{{\sum\limits^{3}_{i=1}\tilde{\bar{c}}_{i}x_{i}\leq\bar{f}_{2}\}}\geq\alpha_{2}}\}\geq\beta_{2}\\ 0.85x_{2(j-1)}\leq x_{2j}\leq1.15x_{2(j-1)} \\ L_{i}-\sum\limits^{3}_{i=1}p_{i}x_{i}< 0.05 \\ \frac{\sum \limits^{3}_{i=1}e_{ij}x_{ij}}{x_{j}}\leq \frac{\sum \limits^{3}_{i=1}e_{i(j-1)}x_{i(j-1)}}{x_{j-1}}(1-0.05) \\ \frac{\sum\limits^{3}_{i=1}c_{j}x_{j}}{x_{j}}\leq\frac{\sum\limits^{3}_{i=1}c_{j-1}x_{j-1}}{x_{j-1}} \\ \sum\limits^{3}_{i=1}x_{i}=1 \\ x_{i}\geq0 \end{array}\right. \end{array}\right. $$
(10)

A practical approach in dealing with multi-objective problems is to find a set of solutions instead of finding a single aggregated objective-dependent global optimum. The decision makers articulate their preference for the different objectives once they have Pareto front knowledge. Thus, to find out the Pareto front (the ACDB curve in Fig. 3), the two minimize objects are defined as: \(Min\bar{f_{1}}= Max-\bar{f_{1}},\,Min\bar{f_{2}}=Max-\bar{f_{2}}.\) Thus \(MaxF=(-\bar{f_{1}}, -\bar{f_{2}}, f_{3}).\) The Pareto optimal front is shown in Fig. 3.

Fig. 3
figure 3

Pareto front points for optimal solutions

Full size image

A case study

This case study integrates both quantitative and qualitative methods.

Analysis of data

Data were collected from three sources: published government documents, information from informal/semi-structured interviews, and field observations. Recognizing that many research questions cannot be addressed if analysis is based solely on official data, we conducted field observations and interviews during field trips to Leshan in 2010. The original database is shown in Table 2. The parameter values were inserted into the system dynamics model and a system simulation performed using the simulation software VENSIM.

Table 2 The original database table
Full size table
Table 3 Comparison of carbon intensity of the three scenarios
Full size table

Industrial structural development between developing and developed countries

Modern economic growth is essentially a process of structural transformation, where the economic growth stage corresponds to the industrial structure and changes in the industrial structure are a by-product of the economic growth (Clark 1940; Fisher 1939).

Syrquin (1988) identified three stages of structural transformation in the evolution of economies: the first stage focuses upon primary production, the second stage focuses upon shifts towards the manufacturing sector, and during the third stage the service sector share increases along with exports. For both developed and developing countries, industry structure varies with economic size, as measured by gross domestic product, population size, the total added value, and the total gross output (Thakur 2011). The industry development process is shown in Fig. 4.

Fig. 4
figure 4

Industry structure development process between developed and developing countries

Full size image

In developed countries, industrial restructuring is uniquely different as the country is still undergoing an economic transformation process. It was observed earlier that developed countries already generate the largest share of value added in service activities and the CO2 emissions are stabilizing (Jaunky 2011), but this happened only after the transition through an industrial phase, where manufacturing became the dominant source of employment and income.

In developing countries, economic development relies heavily on the development of the secondary sector, with a resulting increase in labor-intensive and pollution-intensive industries. Labor value added is also high in most agricultural sectors in a developing country because of the relatively low levels of mechanization. In China, for example, following two decades of rapid economic growth and an increasing demand for energy, the citizens are becoming more environmentally aware. However, how a developing country can rapidly and sustainably transform itself and its economy to a more modern society remains a critical policy challenge.

Leshan

Leshan (E103°43′N29°35′) located in southwest Sichuan province, as a dual world heritage area, faces considerable pressures and challenges. Although the county claims to practice ecological planning within its industrial development to ensure its stated goal of sustainability, a gap between the policy rhetoric and the practical implementation exists. In order to adapt to the need for global emissions reduction and cultural heritage protection, Leshan aims by 2020 to reduce carbon intensity by 20 %.

The industrial added value share for the three major industry sectors in the GDP is 11.0:46.5:42.5. Although per capita GDP has increased, the current industrial development trends may result in short-term gains at the cost of long-term environmental sustainability.

Carbon intensity increased from 0.6 in 2001 to 0.71 in 2008. Pollution discharge mainly comes from the secondary industry sector’s high polluting industries such as the textile industry and the foundries. New energy use reached only 0.25 in 2008, while coal consumption growth was far higher than the national average target.

The total employment population changed slowly from 341,400 in 2001 to 390,500 in 2008. By 2008, there were 164,000 primary industry employees, with secondary and tertiary industry employees being 119,400 and 107,100, respectively.

This area uses fossil fuel-based systems and is a chemical industry region. The ecological, economic, and social concerns are couched in a political economic system that endorses the pursuit of short-term economic benefits at the expense of long-term environmental considerations (Shi 2002).

Three different scenarios for sustainable development

In this study, three different scenarios are considered for an in-depth analysis of the inter-relationships between the industrial structure, the economic benefit, and environmental development. The decision-making process is shown in Fig. 5.

Fig. 5
figure 5

Decision-making process schematic diagram

Full size image

First, through communication with the local policy decision makers, a general level of confidence is determined. The decision makers are quite confident about the statistical data, but they are unsure as to whether the current industrial structure is suitable for the long-term economic and environmental development. The problems that need to be solved relate to the determination of effective sustainable development methods and the most appropriate industrial structure for the future.

The data for the present study comes from investigations in Leshan using various data collection methods including observation, interviews, and documentation collection. Further, uncertainty parameters were determined using expert experience because of the lack of data (see fuzzy random extensions part). Then, a mathematical programming model was formulated that took into account the decision maker’s expectations of the objective values and the environmentally based soft constraints.

A scenario analysis was introduced into the decision-making process to ensure the practicality and operability of the planning alternatives. In a real decision-making process, this process should provide several alternatives for the decision maker to achieve an optimal policy. The three scenarios are as follows:

  1. (1)

    Inaction scenario: The three industrial sector proportions are 11.0:46.5:42.5 which remain as at present without readjustment or upgrading. The basic environmental investment is 1 % of GDP which remains unchanged, and which does not include investment in carbon emissions reduction.

  2. (2)

    Industrial restructuring scenario: The three industrial sectors are adjusted according to the ideal proportion predicted by the SD-MOP model with the proportion being 10.7:43.8:45.5.

  3. (3)

    Industrial restructuring and carbon reduction scenario: This is considered the optimal scenario and follows the assumption of Scenario 2 but with higher weights for pollution abatement and energy consumption reduction because of higher environmental investment.

Simulation results

In this section, the potential behavior of the integration of the above three policy options was simulated and the results compared with the reference behavior. The simulated data for the Total employed population (Unit: People), GDP (Unit: Ten thousand Yuan), Energy consumption (Unit: Ton of standard coal), Pollutants (Unit: Ten thousand tons), and Carbon intensity (Unit: Ton/Ten thousand Yuan) in the 9-years (from 2011 to 2020) are shown in Fig. 6. Figure 6a shows the forecast figures according to the three different scenarios.

Fig. 6
figure 6

a The GDP of three scenarios. b The employment figure of three scenarios. c The pollution emissions of three scenarios. d The energy consumption of three scenarios

Full size image

As Fig. 6a shows, the industrial structure optimization is superior in the long run, but the distinction is not obvious. In short, the GDP in Scenario 2 and Scenario 3 are both lower than in Scenario 1. Increasing the environmental investment has a slight effect on the economy. The 2020 GDP is predicted to reach 1.0 × 108 (Ten thousand Yuan) whereas the industrial structure without optimization is predicted to reach only 7.5 × 107 (Ten thousand Yuan).

As can be seen from the Fig. 6b, there is no direct relationship between industrial restructuring and total population change. Yet, with a readjustment in the industrial structure, the employment structure changes dramatically. Under the optimized industrial structure, the number of positions in 2020 reaches 4.5 × 105 (People).

As environmental problems are nominated as improvement goals in the model, the optimized industrial structure has a positive impact on environmental action. The pollution emission reduction in the next 10 years is shown in Fig. 6c. As the industrial restructuring increases the tertiary sector proportion, the amount of pollutants in Scenario 2 is less than in Scenario 1. Therefore, it is easy to see how important industrial restructuring and environmental investment is in the reduction of pollutants.

Figure 6d shows the energy consumption figures. As the proportion of tertiary industry sector increases, the energy consumption in Scenario 2 and Scenario 3 are both lower than in Scenario 1. This shows that the constraint condition in our MOP model has a large positive effect on decreasing the average energy consumption increase.

Different from the other two scenarios, the aim to reduce carbon intensity by 20 % in Leshan is only reached in Scenario 3. The comparison of carbon intensity of the three scenarios can be seen in Table 3. This is a result of the inclusion of sustainability measures and the higher environmental investment.

The main conditions for Scenario 1 can be seen in Fig. 7a. Development on the basis of current industrial structures leads to an increase in carbon intensity to 0.94 (Ton/Ten thousand Yuan) in 2020 (Fig. 7b). With readjustment, however, carbon intensity decreases to 0.39 (Ton/Ten thousand Yuan) in 2020, the employed population increases, and pollution reduces to 40 % less than in Scenario 1. Because of the environmental improvements, carbon intensity is predicted to reduce to 0.37 (Ton/Ten thousand Yuan) in 2020 in Scenario 3 (see Fig. 7c). This scenario balances both economic development and environmental management, so it is considered optimal.

Fig. 7
figure 7

a Simulation result of Scenario 1. b Simulation result of Scenario 2. c Simulation result of Scenario 3

Full size image

Discussion and policy proposals

Discussion

Although industrial restructuring theory has been widely studied from both theoretical and empirical perspectives, the intensive debates in western China have generated more noise than answers.

There are large disparities between the scenarios. In terms of economic development, the GDP in Scenario 2 and 3 are lower than in 1 in the short run as a result of the reduction in the proportion of the secondary industry sector and are higher than in 1 in the long run. There are two reasons to explain this unexpected finding. First, China, as a typical developing country, is still in the middle stage of industrialization. GDP growth relies heavily on the development of the secondary industry, which is generally energy intensive. Second, the impact of environmental regulation on GDP growth was negative in the short run but the changes were not significant or even have no effect on economic growth after the economic level reaches a certain point. This echoes previous empirical findings about the short-run and long-run causal relationship on industrial structure change, pollution emissions, economic activity, energy efficiency, and financial development (Hisham 2013; Saidur 2012).

In terms of environmental investment, we find that sustainably oriented industrial restructuring takes longer to affect carbon intensity while increasing environmental investment do not need. Moreover, increasing environmental investment seems to have a slight negative effect on GDP growth, but a positive impact on decreasing carbon intensity. This implies that while appropriate adjustments in industrial structure in the long run can help reduce pollution, increasing environmental investment can also play an active role. Thus, it is better to put more investment into the environment than into industrial restructuring if an area wants to change its environmental pollution in the short term.

In terms of employment, since each industry has a different number of jobs, the optimization of the industrial structure increases employment. This is associated with the introduction of machines in the production process and consequently the elimination of human labor (Schettkat and Yocarini 2003). That is to say, with industrial structure optimization, employment problems can be mitigated. We can see that, from the point of view of labor absorption, the tertiary industry has an important role for developing countries’ economic growth, especially in those developing countries with a rich labor force endowment and an inexpensive labor force pool. Thus, for DWH regions, an improvement in the tertiary industry development is ideal.

Policy proposals

From the above analysis, we propose the following policy changes.

(1) Increase the number of related policies to encourage environmental investment.

Policies should encourage the development of an environmental protection industry and there should be severe penalties for enterprises which have illegal emissions or other actions which lead to environmental degradation. Further, environmental protection investment needs to be increased and the introduction of new energy technologies needs to be strengthened.

(2) Construct integrated chain of tourism industry.

Tourism is an important driving force for economic growth, so it should also be an important force in dealing with the associated environmental problems. It is important that tourists stay for longer to boost consumption. Thus, the local government should exploit tourist resources better to ensure more tourist possibilities.

Conclusion

This article proposes a new SD-MOP model for the sustainable planning of urban industrial restructuring. Three different scenarios which consider the environment, the economy, and society are compared. The particularities of a DWH area in terms of the economic development and environmental protection are revealed. The simulation results show that only Scenario 3 can achieve the aim of decreasing carbon intensity by 20 %. Sustainably oriented industrial restructuring takes longer to affect carbon intensity while increasing environmental investment do not need. Moreover, increasing environmental investment seems to have a slight negative effect on GDP growth. Further, with industrial structure optimization, employment problems can also be mitigated. The proposed model also can be helpful for the sustainable industrial restructuring in traditional industrial cities of developing countries.