low price new pyrrhotite system sand production line sell at a loss in brazzaville

subscribe to read | financial times

Keep abreast of significant corporate, financial and political developments around the world. Stay informed and spot emerging risks and opportunities with independent global reporting, expert commentary and analysis you can trust.

analysis of an application possibility of geopolymer materials as thermal backfill for underground power cable system | springerlink

The circular economy is a closed cycle that allows one to reuse the industrial waste, as well as minimize the energy and resources losses during the production process. This paper presents an innovative idea of the application of a geopolymer cable backfill for underground power cable system installation. The closed cycle, in this case, is formulated as follows: the primary resource is the waste from the combustion of fossil fuels, i.e., fly ash that is utilized to form the geopolymer matrix. The geopolymer then is used as thermal backfill in underground power cable systems. Utilization of combustion by-products in the form of a geopolymer is a highly profitable solution since landfill waste disposal, in this case, generates considerable costs for the electrical energy producers. In typical applications, geopolymers are used as insulators. By adding individual components, the thermal conductivity of 2.0W/(m K), higher than of typical thermal backfills (Fluidized Thermal Backfill), which value is close to 1.5W/(m K), is reached. What is very important, geopolymers can absorb water better than typical sandcement mixtures. As a result, a high thermal conductivity with the temperature increase is maintained. The application of geopolymers as thermal backfills has the potential to improve the flexibility of underground power cable systems, as well as to minimize the material costs of installation. The case study is presented to show the economic benefits of using the combustion by-products as a geopolymer thermal backfill. The finite element method model of an underground power cable system is developed, and optimization of backfill dimensions is provided to minimize the material costs using the geopolymer thermal backfill and to maximize the underground power cable system performance. The main result of this paper is that the application of geopolymers leads to a decrease in underground power cable system costs, compared to traditional backfill (sandcement mixture). The reason is the higher value of thermal conductivity, which allows selecting a cable with a smaller cross-sectional area. Also, the environmental benefits of geopolymer application for cable bedding are discussed.

For high voltage (HV) and extra high voltage (EHV) underground transmission lines, the main challenge, apart from the relatively high installation costs, is its limitation in energy transmission. The overall performance depends mainly on the thermal properties of the ground in which it is situated. The burial technique for the power cable installation, as well as the HV power cable cross section, is shown in Fig.1.

The decrease in the cable conductor area reduces the allowable current transmitted (Shabani and Vahidi 2019). For a high cross-sectional area, larger currents are transferred. The heat generated increases the operating temperature of the cable itself, especially in summer, which can result in cable temperature exceeding 90C. This limit value is, in fact, the melting temperature of the XLPE cable insulation. A prolonged increase in the cable operating temperature leads to the melting of the polyethylene insulation and the transmission line failure. The repair time for HV underground transmission lines is about ten times longer than for overhead lines. During the cable line outage time, each hour of electricity supply disruption entails substantial financial losses to the electricity provider. From the reasons mentioned above, the failure rate of HV and EHV underground transmission lines must be reduced to a minimum (Nemati et al. 2019).

The arrangement of the Underground Power Cable System (UPCS) in the ground is associated with a considerable and unpredictable variability of environmental parameters, determined by the thermal conductivity of the surrounding soil. When designing HV and EHV UPCS, it is practically abandoned to lay the cable lines directly in the ground. To enhance heat dissipation from the cable core, ensuring moisture retention in a region close to the cable line, a specially designed thermal backfill is used. The thermal backfill needs to provide stable thermal conditions for the cable line operation regardless of the type of surrounding soil. Thermal backfill application during UPCS design and installation brings extensive benefits to the energy system. First of all, cable systems ampacity is raised around 1015% by ensuring favorable heat transfer conditions. Also, the cable conductor cross-sectional area is reduced during the design process since expected ampacity will be reached when using a smaller cable. In both cases, significant reductions in construction and operating costs can be achieved.

In Europe, a mixture of sand and Portland cement, in a 12:1 ratio, has been used for years as thermal backfill. Despite its advantages, i.e., low unit price, availability on the market, and adequate mechanical properties, it does not ensure the thermal stability of the cable line operation in the long-term period. After a few months of the thermal load caused by the line operation, its thermal conductivity drops significantly (Ahmad et al. 2019). What is more, the manufacturing of Portland cement caused a considerable environmental burden and energy efficiency issues (Habert 2013).

The Portland cement is produced at a temperature of 14001500C (Huntzinger and Eatmon 2009). During its production, large amounts of CO2 and NOx are emitted into the atmosphere, as reported by (Li et al. 2016). One of the main factors stimulating the development of inorganic polymer technology is the possibility of a real alternative to Portland cement. It is estimated that the synthesis of Portland cement is twice as energy-intensive as the production of geopolymers and causes 4 to 8 times more carbon dioxide emission (Provis and van Deventer 2009). According to several authors (Bajpai et al. 2020; Zhao et al. 2020), using geopolymers as a substitute for Portland cement reduces the environmental load significantly.

In general, geopolymer is a product of alkali activation of any aluminosilicate materials. It consists of long chainscopolymers of silicon and aluminum oxides (fly ash components), as well as stabilizing metal cations, usually sodium, potassium, lithium, or calcium. The alkali activation is performed by mixing NaOH, KOH, Na2SiO3, or K2SiO3 with fly ash. The geopolymerization occurs at low temperature, usually under 100C and requires no special preparation of the raw material (fly ash) before the process (Zhuang et al. 2016).

The use of geopolymers is a viable alternative to the waste disposal from the energy, mining, and metallurgical sectors. It is possible to reuse considerable amounts of industrial by-products through geopolymerization, therefore minimizing the environmental impact of such waste (Bajpai et al. 2020). That is because geopolymerization is an effective and economically justified method of immobilization of heavy metals, such as Pb, Cd, Cr, and Zn, contained in postprocessing waste, in the form of fly ash matrix (ach et al. 2018). Also, fly ash is considered a hazardous material since the concentrations of the toxic trace elements in fly ash could be 410 times higher than in coal (Boca Santa et al. 2016). Therefore, geopolymerization may be used for stabilizing and solidifying those hazardous substances that would be otherwise disposed into landfills and pollute the environment (Mohajerani et al. 2019). What may confirm the immobilizing nature of geopolymers and its environmental benefits, it is a fact that geopolymerization may be successfully used as well in disposing of medical waste ash (Tzanakos et al. 2014). The authors were able to reduce the leachate of heavy metals concentrations below the permitted limits significantly.

This paper presents the utilization possibility of fluidized bed fly ash and blast furnace slag as raw material for a geopolymer thermal backfill composition. Fluidized bed combustion (FBC) technology is gaining its popularity nowadays since it is a very attractive, efficient, and environmentally friendly coal-combustion technology. FBC enables the utilization of low-grade solid fuels, as hard coal, lignite, municipal, and industrial waste, or biomass when emitting less NOx and SO2 compared to traditional coal-combustion technology (Ohenoja et al. 2020). The chemical composition and characteristics of coal ash generated from FBC boilers significantly differ from those of pulverized boilers, as the coal is fired at a relatively lower temperature, and a large amount of limestone is used for desulfurization. Because of a high variation in the produced fly ash quality caused by changes in the fuel composition, FBC fly ash may not be treated as a substitute to the pulverized boiler fly ash. Therefore, FBC fly ash is not used for a Portland cement composition (Zhang et al. 2012). As having low popularity as raw material, fluidized bed fly ash and blast furnace slag are mainly disposed.

The main objective of this paper is to prove that the application of a geopolymer as thermal backfill instead of traditional used sandcement backfills for cable systems is economically feasible. This study presents an economic analysis of a possible implementation of geopolymer materials for the thermal backfill of UPCS. This idea is presented for the first time. The major part of a geopolymer is a matrix that can be composed of the combustion by-products. The combustion by-product that is assumed to be utilized by using the geopolymer is fluidized bed fly-ash. Therefore, there is no need to pay the costs of those by-products utilization. The analysis is performed for three different prices of 1m3 of thermal backfill, to show the potential benefits of using geopolymers in this kind of application. In the study, the finite element method (FEM) model of UPCS is used, described in detail in the previous paper (Oco et al. 2016). In the article (Oco et al. 2015a), the application FEM for modeling of cable systems situated in a multilayered soil is presented. The optimization procedure, based on PSO (Oco et al. 2015b) and Jaya (Oco et al. 2018) algorithms, is also used to determine the cable backfill area and optimize the UPCS system costs.

Fuel combustion produces combustion by-products such as fluidized bed fly ash and blast furnace slag (Hirschi and Chugh 2019). Those materials can be used to form a geopolymer matrix. The use of fly-ash or slag for the production of geopolymers is a pro-ecological approach, as it allows for practical utilization of industrial and energy sector by-products, as fluidized bed fly ash and blast furnace slag. The mass content of combustion by-products can reach up to 50% of the thermal backfill weight. The thermal backfill material could contain, for example, silica-aluminate inorganic binders, sand, and additives that enhance the overall thermal conductivity. The obtained composite can be used to develop a thermal backfill material for the UPCS. The primary benefit of such an approach is the use of combustion by-products directly in power cable systems, which causes a limited need to utilize such a waste. Various engineering groups need to be involved in such products, for example, energy engineering, environmental engineering, material engineering, thermal engineering, and electrical engineering. The following aspects should be considered when developing thermal backfill materials, according to (Radhakrishna 1982):

Cable manufacturers provide values of power cable ampacity when assuming specific environmental conditions. It gives the possibility to compare similar cable performance and to select a particular model of cable for a given investment case-study. It is a common practice to determine the current ratings based on IEC 60287 using the CYMCAP software (CYME 2020). The current ratings are for the following UPCS working conditions:

The value of soil thermal conductivity is critical for the cable core temperature and the cable ampacity. When assuming a constant value of the soil thermal conductivity, at the level of 1.0W/(m K), according to the IEC 60287, 2014, standard, the cable line designers are forced to use a high safety margin for the calculation results. As a result, a broader cross section of the cable conductor is selected than obtained from calculations. The purpose is to ensure reliable operation of the transmission line even when the thermal conductivity of the ground drops below the assumed value. As a result, a significant increase in investment costs of the underground cable line construction is expected to avoid cable overheating.

This study presents an economic analysis of the geopolymer application in the underground transmission line design. The UPCS model is based on the test case considered in practice by the cable line designers. The following assumptions are made:

The heat conduction in the UPCS is considered as a two-dimensional problem (Eq.(1)). Based on Eq.(1), the highest temperature in the system Tmax is calculated. For the present calculation, the 400kV power cable is composed of a conductor, insulation, and an outer layer. The heat source term is only used for cable core and insulation domains. The adopted model is described in detail in the authors previous work (Oco et al. 2018) and is used for the analysis of geopolymer-based thermal backfill materials of power cables.

The system of governing differential equations, given in Eq.(1), is subjected to boundary conditions, shown in Fig.4. The ambient temperature Tamb is considered to be equal to 30C (conditions in the summertime when the air temperature is high). The air heat transfer coefficient hamb is assumed as 5.0W/(m2 K). The right and bottom sides of the heat transfer domain are insulated, while the symmetry boundary condition is applied at the symmetry plane of the UPCS domain. The last assumption allows one to reduce the number of finite elements and decrease the computational time.

After discretization, the resulting nonlinear system of differential equations is solved, and the nodal temperatures are determined. The procedures described in IEC 60287, 2014, standard are used to calculate the cable core and insulation heat losses.

In Eqs. (6)(12), l is a distance between cables, and f is an alternating current frequency, equal to 50Hz. The value of Ks and Kp, is equal to 0.37 and 0.8, according to ICE 60287, 2014. The Round Milliken bare bi-directional wires are considered.

where Ccable is the cost of cables in the UPCS in $/km, Cback is a unit cost of cable backfill material with 1 m2 cross-sectional area and 1km length, wcbp is a fraction of combustion by-products in geopolymer thermal backfill (assumed as equal to 0.5), and Cut is the cost of the utilization of 1 t of combustion-by-products (assumed to be equal 100 $/t).

The calculations were performed using the own FEM code written in MATLAB. Figure5 shows the temperature fields obtained for sandcement mix thermal backfill and geopolymer thermal backfill. It is possible to observe that the application of geopolymer thermal backfill reduces the temperature in the region close to cable core, due to the higher thermal conductivity than of the sandcement mixture.

The economic benefits of the application of a geopolymer are observed for different prices of the 1m3 of thermal backfill (Fig.6). The overall costs of the cable system are calculated for the considered price of thermal backfill equal to 100 $, 150 $, and 200 $. The SCM thermal backfill costs are assumed to be equal to 30 $/m3. Figure6c shows the variation of a single power cable cost concerning the cable cross section. Figure6a shows the scenario when the prices of a geopolymer thermal backfill are equal to 100 $. Due to the significant savings of the company on utilizing of the combustion by-products, the overall costs of the cable system Csys are only slightly higher for the same cable conductor cross section Ac, than the costs when the sandcement mixture is applied. What is essential is a fact that when the geopolymer thermal backfill with thermal conductivity of kb=1.4W/(m K) is used, then the temperature of the cable core is less than 90C, the maximum temperature of cable operation, if one applied cables with Ac=1400 mm2. When the sandcement mixture is used as thermal backfill, the cable core temperature is significantly higher than 90C (Fig.6d). The cables with a larger cross-sectional area of 1600 mm2 shall be used. The potential saving arising from the application of a geopolymer thermal backfill can reach over 60,000 $/km of cable line for this case study.

Material costs of the UPCS installation, when the geopolymer price is set as a 100 $/km; b 150$/km; c 200 $/km; d cable conductor maximum temperature; e cost of the single power cable in 103 $/km; f area of thermal backfill applied

In case, when the geopolymer thermal backfill price is set as 150 $/ m3 (Fig.6b), the potential savings are smaller, but still remunerative with over 30,000 $/km of the cable line. In a case when the geopolymer thermal backfill costs are 200 $/m3 (Fig.6c), then the potential economic benefits of replacement of the sandcement mixture with geopolymer are limited to 10,000 $/km of the cable line. Based on the performed calculations, it is evident that applying the dedicated backfill materials like geopolymers reduces the overall costs of the UPCS compared to the traditional sandcement mixture.

As presented in Fig.6, 0.448 m2 cross-sectional area of the geopolymer backfill may be applied for the case-study since the Ab value is calculated for half of the system considered (Fig.4). Therefore, around 448 m3/km of backfill needs to be applied for each km of the designed cable line length. On the contrary, in engineering calculations and during the UPSC design, the rectangular shape of SCM backfill with 1.4m width and 0.4m height is commonly applied. Thus, approximately 522 m3/km of the SCM is needed. Consequently, the presented approach and detailed thermal analysis of UPCS operation result in decreasing in backfill use for about 74 m3/km (more than 14%). That significant reduction in backfill area may be reached by replacing SCM (kb=0.8W/(m K)), with the geopolymer backfill (kb=1.4W/(m K)). When considering backfill prices, SCM for the transmission line installation costs 15,660 $/km and 44,800 $/km for the geopolymer backfill.

As mentioned earlier in the text, the mass content of combustion by-products in a geopolymer matrix can reach up to 50% of the backfill weight. When assuming the dry density of thermal backfill approximately 1.850 t/m3 (Radhakrishna 1982), it gives around 414.4 t/km of combustion by-products utilized per each km of HV and EHV transmission line built. As reported by (American Coal Ash Association 2018), disposal costs of coal by-products depend on multiple factors, as the specific type of combustion waste, location, transportation methods, climate and terrain, regulatory requirements, and potential for its future use. For the most unfavorable conditions, the price may reach 20 to 40$/t of coal combustion waste disposed. When a disposal site is located near the power plant and the material being disposed can be easily handled, the price may decrease to 35 $/t. When taking into account the costs of the combustion waste disposal around 10 $/t, benefits from its utilization in the form of the geopolymer backfill may give about 4144 $/km savings for the transmission line length. On the contrary, Portland cement prices have reached a level of 123.5 $/t, according to (Statista 2020), which gives approximately 9.51 $/t of the SCM.

The detailed carbon footprint analysis for geopolymer and Portland cement concrete is performed by (Turner and Collins 2013). According to (Turner and Collins 2013), equivalent CO2 emission during the Portland cement manufacturing reaches 820kg CO2-e/t. The fly ash and other coal by-products are considered zero or minimal CO2-e contribution. For the fly ash, it is 27kg CO2-e/t. What is important, geopolymer backfill composition also requires the alkali activator (sodium hydroxide) as well as sodium silicate, which along with fly ash and NaOH, creates the geopolymer binder. According to (Turner and Collins 2013), the total equivalent CO2 emission reaches 108.56kg CO2-e/t for a mixture of fly ash, sodium hydroxide, and sodium silicate for the geopolymer concrete composition. As the investigation of the proper geopolymer backfill composition is still in progress, it is impossible to evaluate its environmental footprint properly. After rough estimation environmental costs for the SCM application may reach around 60,974kg CO2-e/km, and for the geopolymer backfill should be lower than 90,048kg CO2-e/km. What should be stressed, the final composition of the geopolymer backfill is not known yet. Therefore, it is impossible to assess its total carbon footprint.

The HV and EHV underground transmission lines are widely used nowadays. To allow the proper operation of the system, not to exceed the maximum allowable conductor temperature equal to 90C is needed. Thermal backfill materials are used to improve the heat transfer from power cables. The commonly used sandcement mixture has low thermal conductivity, which in the dry state is equal to 0.8W/(m K). There is a need to develop a novel and not expensive thermal backfill materials with higher thermal conductivity. The option here could be geopolymers with a matrix based on fluidized boilers fly-ash, which is challenging to utilize. This study shows the potential economic benefits of the application of this kind of fuel combustion by-products as thermal backfill. For a flat-formation of the UPCS, with the maximum current of 1,200 A, the computations were carried out. It was assumed that the costs of fluidized bed combustion by-products utilization are of 100 $/t. A reasonable value of geopolymer thermal conductivity equal to 1.4W/(m K), and three cost levels of 100 $/m3, 150 $/m3, and 200 $/m3 of thermal backfill were assumed.

The computations showed that in a case when the costs of geopolymer thermal backfill are of 100 $/m3 of thermal backfill, the potential saving in cable system design can lead up to 60,000 $/km savings of cable system material cost. The reason is the possible use of a cable with a smaller cross-sectional area, i.e., 1400 mm2 (when geopolymer thermal backfill is used), compared to the one applied for sandcement mix thermal backfill, i.e., 1600 mm2. Therefore, the utilization of fluidized-bed combustion by-products can create potential economic benefits for the energy distribution industry.

Summing up, geopolymers application in backfilling technology for the HV and EHV underground transmission seems to be promising. By using industrial waste and turning it into a full-value product may surely bring many benefits. Nevertheless, extensive and multidisciplinary studies must be performed to investigate the preferable composition of the backfill. Also, the material test should be conducted to check if the backfill does not interfere with the power cable sheath, causing its corrosion, among others.

Ahmad S, Rizvi Z, Khan MA, Ahmad J, Wuttke F (2019) Experimental study of thermal performance of the backfill material around underground power cable under steady and cyclic thermal loading. Mater Today Proceed 17(1):8595

Anders GJ (1997) Rating Of electric power cables: ampacity computations for transmission, distribution, and industrial applications. Institute of Electrical & Electronics Engineers (IEEE) Press, New York.

Habert G (2013) Environmental impact of Portland cement production. In: Pacheco-Torgal F, Jalali S, Labrincha J, John VM (eds) Civil and structural engineering, eco-efficient concrete. Woodhead Publishing, Cambridge, pp 325.

Hirschi JC, Chugh YP (2019) Sustainable coal waste disposal practices, chapter. In: Hirschi JC (ed) Advances in productive, safe, and responsible coal mining. Woodhead Publishing, Cambridge, pp 245269

ach M, Mierzwiski D, Korniejenko K, Mikua J, Hebda M (2018) Geopolymers as a material suitable for immobilization of fly ash from municipal waste incineration plants. J Air Waste Manage Assoc 68(11):11901197

Nemati HM, Anita SantAnna, Sawomir Nowaczyk, Jan Henning Jrgensen, Patrik Hilber (2019) Reliability evaluation of power cables considering the restoration characteristic. Int JElectrical Power Energy Syst 105:622631

Nexans (2011) High voltage underground power cables, 60500 kV high voltage underground power cables XLPE insulated cables. Available online:https://www.nexans.com/Corporate/2013/60-500_kV_High_Voltage_full_BD2.pdf

Oco P, Bittelli M, Cisek P, Kroener E, Pilarczyk M, Taler D, Rao RV, Vallati A (2016) The performance analysis of a new thermal backfill material for underground power cable system. Appl Thermal Eng 108:233250

Oco P, Cisek P, Pilarczyk M, Taler D (2015a) Numerical simulation of heat dissipation processes in underground power cable system situated in thermal backfill and buried in a multilayered soil. Energ Convers Manage 95:352370

Oco P, Cisek P, Rerak M, Taler D, Rao RV, Vallati A, Pilarczyk M (2018) Thermal performance optimization of the underground power cable system by using a modified Jaya algorithm. Int J Thermal Sci 123:162180

Shabani H, Vahidi B (2019) A probabilistic approach for optimal power cable ampacity computation by considering uncertainty of parameters and economic constraints. Int J Electrical Power Energ Syst 106:432443

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Oco, P., Cisek, P. & Matysiak, M. Analysis of an application possibility of geopolymer materials as thermal backfill for underground power cable system. Clean Techn Environ Policy 23, 869878 (2021). https://doi.org/10.1007/s10098-020-01942-8

a review of experimental studies on sand screen selection for unconsolidated sandstone reservoirs | springerlink

Sand production is a problem that affects hydrocarbon production from unconsolidated sandstone reservoirs. Several factors, such as the strength of the reservoir, its lithification and cementation and reduction in pore pressure, may cause sand to be separated from the rock and transported by hydrocarbons to the well. Producing sand commonly causes erosion and corrosion of downhole and surface equipment, leading to production interruptions and sometimes forces operators to shut-in wells. Several different methods of sand control are available to reduce the impact of sand production. The reviewed papers suggest that the most suitable methods for unconsolidated sandstone reservoirs are stand-alone screens and gravel packs. Because of the cost and complexity of gravel packs, stand-alone screens are usually the first choice. These screens have different geometries, and selection of the most suitable screen depends on the particle size distribution of the grains in the formation and other reservoir and production parameters. A screen retention test, run in a laboratory with screen samples and typical sands, is often used to ensure that the screen is suitable for the reservoir. This paper reviews the main causes of sand production, the properties of unconsolidated sandstones that predispose reservoirs to sand production problems and the selection criteria for the most suitable mitigation method. The process of selecting a screen using experimental screen retention tests is reviewed, and the limitations of these tests are also discussed. Some numerical simulations of experimental tests are also reviewed, since this represents a very cost-effective alternative to laboratory experiments.

Sand production affects more than 70% of the oil and gas reservoirs around the world (Khamehchi et al. 2015; Ikporo and Sylvester 2015). It can have a severe effect on well productivity and equipment as it could plug the well and erode equipment which could lead to loss of containment and also settle in surface vessels. Sand production can be controlled and mitigated by installing sand control both downhole and at the surface. The application of sand control in a reservoir could prevent or minimize the sand from being produced. However, installing unsuitable sand control normally comes with risks, such as high skin and a decrease in productivity index (Hodge et al. 2002; Khamehchi et al. 2015; Ikporo and Sylvester 2015; Changyin et al. 2016; Gupta et al. 2016; Toelsie and Prediepkoemar 2013; Matanovic et al. 2012).

In considering sand control methods, one must differentiate between load-bearing solids and fine particles (fines), where it is actually beneficial to produce fines as long as they can move freely through the screen or gravel packs and not plug it. Sand control usually refers to the control of the load-bearing solids that support the overburden. The critical factor in assessing the risk of sand production is the ability to maintain the sand production below an acceptable rate, and at flow rates which will make the well production acceptable. The aim of this paper is to review the experimental studies on various sand screens to mitigate sand production in unconsolidated reservoirs (Ikporo and Sylvester 2015; King 2013; Hodge et al. 2002; Khamehchi et al. 2015).

Sand production generally occurs when the reservoir sandstone cement is weak and fails under in situ stress or imposed stress, where both stresses were changed during hydrocarbon production. The produced oil or gas from such reservoirs can create problems ranging from erosion of the downhole or surface facilities to well stability and later produced sand disposal. Sand production can occur both naturally, in unconsolidated formations, or due to drilling and production activities.

In completely unconsolidated formations, sand production may happen at the start of the fluid flow from the formation due to drag from the fluid or gas turbulence, which detaches sand grains and carries them to the perforation. It also can start when there are changes in the production rate, water breakthrough or changes in the gasliquid ratio (Deghani 2010; Ikporo and Sylvester 2015; Toelsie and Prediepkoemar 2013).

When sand and hydrocarbon are produced at the surface with a given flow rate, it creates downhole reservoir cavitation, and over time the formation may collapse due to lack of support which may result in a complete loss of productivity. The formation collapse leads to a significant pressure drop near the wellbore. When sand production occurs, the sand grains will accumulate behind the casing to create a lower permeability zone, especially for formations with a high clay content or a wide range of sand grain sizes. Sandstones with narrow grain size distributions show lower variations in permeability. The five main factors affecting the sand production are: the degree of consolidation, reduction in pore pressure, production rate, reservoir fluid viscosity and increasing water production throughout the life of the well (Ikporo and Sylvester 2015; Deghani 2010; Khamehchi et al. 2015).

The degree of consolidation defines how strong the individual sand grains have been bound together and how the cementation process has developed. Typically, the compaction and cementation of sandstone is a secondary geological or diagenetic process. Older sediments and particular lithologies tend to be more consolidated. For this reason, most shallow, geologically younger reservoirs are associated with sand production, as they often have weak cementation that binds the sand grains together. Compressive rock strength is a geomechanical property of rock that is related to the degree of consolidation. Unconsolidated formations usually have a compressive strength of less than 1000 psi or about 6.89MPa (Ikporo and Sylvester 2015; Wan and Wang 2004; Ikporo and Sylvester 2015; Toelsie and Prediepkoemar 2013; Penberthy and Shaughnessy 1992; Roberts 2014; Suman et al. 1991).

Part of the weight of the overlying rocks is supported by the pore pressure in the reservoir. Upon producing hydrocarbon, the pore pressure drops and some of the support is removed. This creates an increased amount of stress on the reservoir to the point where the sand grains may break loose from the matrix and create fines that are produced along with fluids (Penberthy and Shaughnessy 1992; Roberts 2014; Suman et al. 1991; Toelsie and Prediepkoemar 2013).

The production of reservoir fluids creates pressure differential and frictional drag forces that could exceed the formation compressive strength when those two forces are combined. This suggests that there is a critical flow rate, a rate when the combined forces are great enough to exceed the formation compressive strength for the sand production to happen. This critical flow rate may be determined by slowly increasing the production rate until sand production is detected. In many cases, the critical flow rates are usually found to be below the acceptable production rate for the well (Khamehchi and Reisi 2015; Khamehchi et al. 2015; Ikporo and Sylvester 2015; Penberthy and Shaughnessy 1992; Roberts 2014; Suman et al. 1991; Toelsie and Prediepkoemar 2013).

The frictional drag force created by the flow of reservoir fluid is directly related to the velocity of the fluid flow and viscosity of the reservoir fluid being produced. High fluid viscosity will apply a greater frictional drag force to the formation sand grains and will cause sand to be produced from many heavy oil reservoirs (Ikporo and Sylvester 2015; Penberthy and Shaughnessy 1992; Roberts 2014; Suman et al. 1991; Matanovic et al. 2012; Toelsie and Prediepkoemar 2013; Changyin et al. 2016).

Sand production may happen when water enters the well. Water production has a severe impact on the stability of the sand arch around the perforation, which may initiate sand production. Water production can affect the relative permeability in water-wet formations. As more water is produced, the relative permeability of oil decreases and this results in an increase in differential pressure to produce oil at the same production rate. This eventually creates a greater shear force across the formation sand grains and leads to instability of sand arch around each perforation and raises the sand production. Table1 summarizes the various causes of sand production into three categories: formation, completion and production issues.

According to Khamehchi and Reisi (2015), the classification of sand production is considered an essential part of predicting the produced sand rates. This classification has been developed based on field observations to allow for a better comparison and interpretation of sand production.

Transient sand production is when the sand concentration is declining with time under constant well production conditions. It is observed during clean-up after perforating or acidizing after bean-up and after water breakthrough.

This happens when the rate of sand produced is high enough to cause the well to suddenly choke and possibly die. It may be due to slugs of sand creating sand bridges of moderate volume in tubing or choke, for example, during or after bean-up and shut-in operations, or when a massive influx of sand fills and obstructs the wellbore.

A wide range of sand control methods are available including a variety of different downhole sand screens and gravel packs. However, installing each type of sand control carries risks; thus, it is important to determine the correct sand control method for a particular formation. Ott (2008) summarized various types of sand control methods including; no control, slotted liner, wire-wrapped screen, prepacked screen, shrouded metal mesh screen, expandable screen, in situ consolidation (resin), oriented and selective perforation, openhole gravel pack, frac pack and screenless frac pack.

The main factors in the selection of suitable sand control methods are cost, efficiencies in retaining sand and life span. Table2 presents the advantages and disadvantages of the available sand control methods.

Among various sand control methods, the screen-only completions are considered the preferred option for the sand control method for unconsolidated formations, as these methods maximize productivity and minimize completion complexity and cost. This is consistent with a new approach by Parlar et al. (2016), which suggests that to select sand control options, one should start with the simplest and most cost-effective sand control, and move to select the complex and expensive options if the simple ones do not meet the design criteria for the project. A stand-alone screen is usually the first choice for completing an openhole completion that is prone to sand production (Hodge et al. 2002).

Stand-alone sand screens are the lowest cost sand control option. They are highly reliable and simple and give long-term productivity performance. SAS is the preferred option for highly deviated or horizontal openhole completions. One of the key parameters for SAS is the sand retaining precision, which determines the success of sand control and whether high production rates can be achieved. The objective of SAS selection is to identify screens that effectively retain sands while maximizing hydrocarbon production, by choosing optimal sand screen aperture and evaluating the limitations during sand retention tests (Wu et al. 2016; Changyin et al. 2016).

Different types of stand-alone screens a and b premium screens with multiple layers; c wire-wrapped screen, d basic screen; e slotted liner and f prepacked screen. Photographs a to d taken and image e and f created by Jami Morteza

The wire-wrapped screen is a carefully wound triangular-shaped wire with a constant gap in between successive turns. The wire is welded to vertical formers placed at 1-cm interval around the internal diameter of the screen. Wire-wrapped screens have an advantage over prepacked screens, as they do not plug easily with drilling mud, and the plugging materials are easily removed from wire-wrapped screens as it tends to get trapped inside prepack. Wire-wrapped also has an advantage over a slotted liner, where the gap between wire-wrapped wires could be made smaller and achieve much greater precision to allow the screen to retain finer grains than the slotted liner (Markested et al. 1996).

The first step in designing the screen is to describe the reservoir sands from samples taken from available cores and logs. These sand samples will then be tested using a laser particle size analysis (LPSA) to determine the grain size distribution, its uniformity, the range of grain sizes with indications of sorting and grain consolidation (Agunloye and Utunedi 2014; Hodge et al. 2002).

Screen permeability is also one of the important parameters in designing a screen, as it is a true indicator of inflow capacity. The standard practice is to perform sand retention tests with real or simulated formation sand. There are numerous design and performance criteria that should be considered when designing a screen. Among these criteria, the sand retention and plugging resistance (retained permeability) are quite important. Over the years, sizing guidelines have been developed to increase the sand control reliability under specific conditions (Hodge et al. 2002; Khamehchi et al. 2015; Parlar et al. 2016).

Wu et al. (2016) present some of the important screen size selection criteria which were determined from empirical correlations (Table3) based on one or two parameters derived from the grain or particle size distribution (PSD), practical experience and laboratory retention tests.

Coberlys (1937) original screen selection criterion does not produce reliable results since it does not consider the sorting or uniformity of sands. Gillespie et al. (2000) and Ballard and Beare (2003, 2006) suggested alternative criteria which tend to perform better than Coberly (1937) by using sand sorting and uniformity coefficient. It is preferable to determine the sand screen opening size by testing a representative sand in a sand retention laboratory or in numerical modeling simulation.

Sand retention tests are commonly used to select the most appropriate screen to be used in sand control. Due to the problems associated with the empirical criteria, the industry-standard practice is to conduct laboratory sand retention tests on different screen coupons to determine their effective screen opening size. All tests measure pressure during the test (or flow rate if pressure is controlled) and the amount of sand produced. The process works with both reservoir sand and simulated sand. Wetting fluid, flow rate and channeling are the major factors affecting sand retention test results. Sand retention tests were useful to compare the retention performance and plugging potential of alternative screens for given formation sand (Agunloye and Utunedi 2014; Chanpura et al. 2011).

There are two types of sand retention tests: slurry and sand-pack or prepack retention test, which will be discussed in the next section. Both of these tests are able to measure the following parameters (Mondal et al. 2011):

Wu et al. (2016) reported that in the sand retention test, the sand has to be deposited onto the screen at a constant drawdown pressure, and not at a constant flow rate. This is to avoid misinterpretation of the rapid increase in the pressure profile attributed to screen plugging from a constant flow rate test. Screen plugging can be determined by measuring the permeability of the sand screen before and after the test was completed. Figure3 shows the apparatus for the sand retention test.

There are no agreed industry standards on how sand retention should be performed or how the results are interpreted. Parameters, such as the screen permeability, amount of sand produced, and pressure drop across the screen, may be obtained from the test.

Slurry sand retention tests use low sand concentrations pumped through the screen to prevent segregation of the formation sand before it reaches the screen. The sand is suspended in a slurry which is a viscous polymer solution and is added to a high-flow-rate brine stream by a displacement pump to dilute the sand concentration flowing onto the screen. Figure4a shows the experimental set up for a slurry test. Slurry tests measure the weight of solids that passed through the screen as well as the rate of pressure buildup across the screen and the amount of sand contacting the screen (Agunloye and Utunedi 2014).

In the sand-pack test, the sand is placed directly on the screen with a confining stress imposed on the sand, so the sand will be in full contact with the screen. A wetting liquid will then flow through the sand pack and the screen. This test measures the weight of sand produced as well as the pressure drop that occurred during the test. Figure4b shows the experimental setup for the sand-pack test (Agunloye and Utunedi 2014; Wu et al. 2016; Mondal et al. 2011; Chanpura et al. 2011).

It may be expensive and time-consuming to conduct many laboratory sand retention test experiments for selecting the effective sand control method. Numerical models and software have been developed using experimental data from laboratory test. The aim of conducting numerical simulations for screen size selection is to avoid repeating laboratory tests in areas where extensive sand retention test data are available. Currently, numerical modeling does not accurately account for interactions between fluids and the particles (Agunloye and Utunedi 2014; Wu et al. 2016; Feng et al. 2012; Mondal et al. 2010; Markested et al. 1996; Constien and Skidmore 2006).

Feng et al. (2012) developed a fully coupled numerical model by combining computational fluid dynamics (CFD) with discrete element method (DEM) code. This technique simulates the sand slurry flow and the sand retention process to determine the effect of parameters such as liquid velocity, screen slot size and particle concentration, or solid volume ratio, on sand screen performance. This approach, where DEM is used to model solid phase and CFD for fluid phase, can provide information on the interaction forces and the movement of individual particles. It could also reveal:

MS method by Mondal et al. (2010, 2011) created a numerical simulation tool to evaluate the performance of sand screens. The simulation is based on a correlation between numbers of particles (Np) of diameter (Dp) produced through a screen of slot opening (W). It allows the user to estimate the mass and the size distribution of the produced solids using the entire particle size distribution of the formation sand. It is applicable to:

Markested et al. (1996) developed a numerical model to simulate plugging and sand production through a single wire-wrapped screen. It was developed to predict critical slot widths and is based on a fractal model for the particle size distribution of reservoir sands.

Constien and Skidmore (2006) developed a method based on laboratory screens testing, which is called the performance curve or mastercurve for individual screen laminates. The mastercurve could be used to predict the screen performance in well with mixtures of particle size distributions. It is constructed by measuring screen performance for produced solids and retained screen permeability versus a ratio of an effective formation size divided by the size of the screen pore opening. The aim is to reduce the amount of possible screen configuration options as well as the number of tests that are needed to make selection decisions.

Gravel pack was developed in the early 1990s in response to an increasing number of failures of the traditional stand-alone sand screen completions. The driving force for applying openhole gravel packs (OHGPs) in deepwater wells was to safeguard long-term productivity of the well by minimizing screen plugging, which may result in productivity decline and creation of localized areas of high-velocity flow (production hot spots) in non-plugged parts of the screen. However, these factors are equally valid for onshore wells that need to be completed with sand control in relatively long, highly deviated reservoir sections (Vliet et al. 2001).

In gravel packing, sand production can be controlled by careful selection of gravel size considering the formation sand size. The main factor that influences the sand production in gravel packed wells is the flow restriction caused by the gravel pack itself. There are three important parameters in designing and investigation of gravel packed wells that influence gravel pack permeability and cost (Khamehchi et al. 2015; Deghani 2010):

The ideal size of gravel pack sand can be determined from LPSA or sieve analysis from core samples, or bailed samples which tend to be large, or produced samples which tend to be small, which are sized to achieve a suitable grain size ratio. In a gravel pack completion, a screen is used with the gravel pack to prevent the gravel from moving. The common type of screen in the gravel pack is wire-wrap screen. Screen openings should not be larger than the smallest gravel diameter. Three basic tools are used in gravel packing operations:

When reservoir sand is not available, commercial sand can be made with a matching particle size distribution to the reservoir sand, using either commercially quarried outcrop sands or ground silica, or a mixture of both. These types of sand are generally well sorted with narrow size distributions. Ground silica is used to represent the fines in reservoir rocks, and as well as any inaccuracies in the representation of the reservoir sand, the simulated sand will comprise silica only. Ballard et al. (2016) state that using simulated sand could be difficult in the sand retention test because the reservoir sand production can be highly variable, and a much slower fluid flow velocity could cause uncertainties.

Parameters such as the pressure or flow rate used in the system are generally controlled and tend to be one or two magnitudes higher than the field parameters. This could exaggerate the screen performance that might not exist.

Chanpura et al. (2011) also pointed out that the sand produced from sand retention tests cannot be directly used to make quantitative predictions of sand production under field conditions unless the test procedures include the maximum laboratory-measured sand production rates and maximum impairment values.

Ballard et al. (2016) conducted sand retention tests on two reservoir sands (B77 and M1) along with their simulated version. M1 sand is better sorted than B77 sand to determine the differences between using reservoir sand and simulated sand on wire-wrapped and metal mesh screen in sand retention tests. The PSDs created for the simulated sands matched the respective reservoir sands. During sand retention tests, both versions of sand gave different retention results, despite having similar grain size distributions. The authors noted:

This may be explained by the particle shape. The simulated sand is made with well-rounded particles composed entirely of silica, but reservoir sands contain a variety of minerals, which affect the grain shape and its properties. The authors pointed out that these observations may not be widely applicable as they considered only two types of sand in their tests, though they suggest that reservoir sand should be used whenever possible (Ballard et al. 2016).

Chen et al. (2016) developed a new test apparatus to offer more accurate screen performance evaluation. They concerned that the current laboratory tests use a small screen disk, due to its convenience and low cost. The current test methods can only test the minimal opening size and pressure drop of local screen material and could not effectively reflect the performance of an integrated screen pipe run in the wellbore.

Instead of using small disks, they used full-size screen samples (Fig.5). In their method, different types of full-size screens were tested using sand samples from target reservoirs. The results were then compared with the results from the cut small disks mainly from the plots of the particle size distribution of screen and pressure drop across the screen sample. They concluded that the small disks sample can only reflect performances of local sand retention material.

This section will review Brunei reservoirs that produce sand. Bruneis major onshore and offshore fields are situated within the Neogene Baram, Champion and Meligan Deltas and Northwest Borneo along the South China Sea. These deltaic fluvio-marine sediments are composed of several sandstone reservoirs vertically stacked with thin layers of laterally continuous shales as cap rocks. Anglo Saxon Petroleum Co. drilled the Belait-2 well which struck the first oil in Brunei in 1914. Since then, many wells have been drilled and several onshore and offshore fields have been developed. These are poorly consolidated sandstone reservoirs, and many of the wells have experienced sand production. To mitigate the impact of sand production, many wells were completed with internal gravel packing as the preferred method of sand control (Fourie et al. 2013; Saeby et al. 2001).

In 1998, Brunei Shell Petroleum (BSP) decided to redevelop a number of reservoir blocks located in the southeastern part of the Champion field. The reservoirs involved were relatively shallow (<1100 mTVD), with shaly, laminated and unconsolidated sandstone. These reservoirs were to be developed by drilling highly deviated wells, and sand control was to be installed in the producing reservoirs. Due to problems during SAS installation on these deviated wells, BSP decided to install openhole gravel pack (Vliet et al. 2001).

Champion West (CW) is located at 7km NNW of the Champion main field offshore Brunei and has been producing since 1975. This field was developed without considering sand production because the reservoirs were located at higher depth known as the sand production cut-off depth with relatively consolidated rock formation. At that time, the expandable sand screen (ESS) was a new and unproven technology. Three wells were completed with ESS and one well (CW-12) with an internal gravel pack (IGP) for comparison. ESS was considered as a good alternative to gravel packing because of the lower cost, the ease of operation, logistic simplicity and completion flexibility. ESS could expand to eliminate the annulus and make gravel packing operations unnecessary in reservoirs with reactive shale, low fracture gradient, or fractures and faults (Saeby et al. 2001; Lau et al. 2004).

Reservoirs in the Champion field are relatively consolidated sandstones with low risk of major sand failure, but due to expected transient failure during high drawdown and depletion all wells have been completed with either IGP and ESS. Based on Table5, the CW-12 well was completed with conventional acid prepacked IGP with PI of 1525m3/d/bar and skin of 15. Meanwhile, the production data of CW-13, 14 and 15 with ESS completion show higher productivity. CW-15 was tested and showed high productivity and low skin (Q=800m3/d, PI=40m3/d/bar, skin=5). CW-13 and 14 ESS completions showed the PI in the range of 2550m3/d/bar. In the initial stages of production, no sand or water has been observed (Saeby et al.2001).

South West Ampa (SWA) field is located 10km offshore Brunei Darussalam. The initial discovery was made in 1963 with SWA-1 which showed a shallow reservoir with API gravity of 40, initial oil viscosity of 0.35cp with relatively high solution gasoil ratio and variable condensate content. SWA field consists of many thin stacked sand layers, with a history of sand production. The shale stability issues and multiple sand layers made the sand control completions challenging. Several openhole gravel packs were installed in the field and were unsuccessful due to the collapse of shale once the openhole was displaced to brine. Consequently, ESS was chosen as the sand control method to replace gravel packing and was first installed in SWA-290 well (Lau et al. 2004).

Results show that ESS was successfully run and expanded 543m of 4 ESS strings and expanded into a 6 horizontal reservoir section. Prior to running the ESS, the mud was further conditioned over 325 mesh screens to reduce the particle size in the mud for running and expanding the screens. Similar 4 ESS was successfully run in Champion Field, CP-306 (Lau et al. 2004).

The Egret field is located 43km offshore in 60m water depth. The field consists of stacked sandstone formations in mainly gas and some oil-rim reservoirs. The first production was from three gas wells in 2003. Two of the three gas wells, EG-1A and EG-1B, needed sand control, and a combination of hydraulic fracturing and ESS was selected. The combined technologies were a key factor in achieving the desired product performance. After the sand control was installed, the tendency for the screen to get plugged was minimized and the sand production was successfully controlled with a low skin factor. The fines production from the Egret field was also successfully mitigated, with no production decline observed (Abdul-Rahman et al. 2006).

Zeidan et al. (2018) investigated a reservoir called Lower Fars from Umm Niqa field in north Kuwait. This field has been successfully completed with vertical cased SAS despite being in a challenging environment with unconsolidated, sub-hydrostatic-sand and a highly sour and moderately corrosive environment. The reservoir was originally planned to be completed with a gravel pack because of the PSD result of the sand showing high Uc of about 7.5% and 11% of fine sands. This plan was then challenged by the author, and it was decided to complete the reservoir using Halliburtons PoroMax SAS (vertical cased SAS). The decision was made based on a thorough analysis of the formation sand, the design of the screen as well as the completion fluid during SAS installation.

Daramola and Alinnor (2018) present remedial sand control for a low-permeability sandstone reservoir. The field was an oil field located offshore Nigeria. Between 2014 and 2016, four wells failed due to multiple sanding events, unstable production rates and platform trips. Further investigations of the bottom-hole pressure data showed that three wells, A1, A2 and A3, failed due to high pressure drawdown, leading to screen breakage, and sand bridge in tubing. A4 well failed either due to the tubing restriction, screen breakage, sand bridge in tubing or scale formation buildup. The asset team decided to design and install frac-pack completion to improve sand control. Frac pack was selected because it has better durability for wells with high pressure drawdown.

Ojeh-Oziegbe et al. (2019) describe the successful installation of a single-trip stand-alone screens completion (STC-SAS) in Bonga deepwater reservoir. Bonga field is located in the southwest of Warri, on the continental slope of the Niger Delta. Due to the declining oil price, the main reason for using STC-SAS was that it could reduce the rig completions time by about 50% compared to the conventional multiple trip SAS completions, hence saving operation costs.

Openhole gravel packs (OHGPs) with a predrilled liner have been successfully installed in 4 wells in the Raven field as reported in Tahirov et al. (2019). Raven field is located in Egypt and is high-pressure high-temperature (HPHT) gas field with a reservoir pressure of over 10,700 psi and reservoir temperature around 141C (285F). The reservoir contains stacked channel formation and requires sand control completion to sustain long-term gas production. All 4 wells showed good performance results with very low skin numbers (+1.5 to +5).

Mahakam River delta is located in the East Kalimantan Province of Borneo, Indonesia. It consists of a large gas field (Tunu) and an oil filed (Handil), where the primary targets located in the shallow, unconsolidated reservoir sands. The gross interval that requires sand control was believed to be more than 1000m long. To save rig time to complete the wells with long multilayer intervals, multizone single-trip gravel pack (MZ-STGP) completion was selected to maximize oil and gas recovery. To date, more than 650 zones have been successfully installed with MZ-STGP (Muryanto et al. (2018).

Sand production is controlled by four major factors: reservoir rock properties (lithological, chemistry and mechanical), fluid properties (fluid phases and chemistry, water invasion), pressure regime (production strategy) and secondary interventions such as water or chemical flooding. A better understanding of the impact of these factors for a given reservoir can significantly improve the effectiveness of sand production mitigation strategies.

Various methods of sand control are available including slotted liner, wire-wrapped screen, prepacked screen, shrouded metal mesh screen, expandable screen, in situ consolidation (resin), oriented and selective perforation, openhole and cased-hole gravel pack, frac pack and screenless frac pack. Each method has associated risks such as installation difficulty, cost, level of fines production, impact on well productivity and longevity. It is essential to determine the reservoir and well parameters such as rock strength, grain size distribution, lithological variations, well type and completion, surface facilities tolerance before selecting the sand control method.

Stand-alone sand screens are of low cost, reliable and simple, with relatively good long-term productivity particularly for highly deviated or horizontal openhole completions. If stand-alone screens are to be used, standard experimental screen retention tests will usually be run to determine the most appropriate screen for a given set of conditions; typically for a well in particular location in a reservoir with a known fluid composition and pressure. The limitation of such tests is that they usually do not test the behavior of the screen over the full range of operating parameters. A rigorous screen selection procedure, based on reservoirs grain size distribution, is essential to choose the best screen for a given reservoir.

Openhole and cased-hole gravel packs are effective alternatives to stand-alone screens. These packs minimize the sand production with the selection of appropriate sized gravel for the produced formation sand. Gravel packs are generally designed for long-term productivity of the well and are expensive and require larger-diameter holes to install.

Screen permeability and the associated sand retention and plugging resistance are indicators of inflow capacity which are important parameters in screen selection that can be determined from screen retention tests. The standard practice is to perform these tests with either real or simulated formation sand. There are two main types of sand retention test: slurry and sand-pack or prepack retention test. Both of these can measure the mass of sand produced as a function of time, the pressure developed around the screen and the particle size distribution of the produced sand.

The recent development of numerical simulation techniques may provide an approach to solving the general problem of screen selection for different grain shapes and size distributions, different fluids and different pressures, once these simulations have been calibrated with experimental tests. They may also suggest improvements to experimental techniques to allow the investigation of the behavior of stand-alone screens, expandable screens and gravel packs over a range of reservoir parameters encountered during the life of the field.

The unconsolidated sands from Bruneis deltaic and fluvio-deltaic reservoir sandstones are vertically stacked with thin layers of laterally continuous shales as cap rocks. These reservoirs have encountered significant sand production issues, so most offshore and onshore wells have been completed with stand-alone screens, expandable sand screens or openhole gravel packs. The expandable and stand-alone screens have had limited success in particular areas. Openhole gravel packs have been outperforming the screens, maintaining permeability across the screen, retaining sand effectively and showing a long-term resistance to plugging.

Abdul-Rahman S, Lim D, Lim J, Ong K (2006) Innovative use of expandable sand screens combined with propped hydraulic fracturing technology in two wells with intelligent completions in egret field, Brunei. In: SPE Asia Pacific oil & gas conference and exhibition. Society of petroleum engineers

Ballard T, Beare S, Wigg N (2016) Sand retention testing: reservoir sand or simulated sandDoes it matter? In: SPE international conference & exhibition on formation damage control. Society of petroleum engineers

Daramola B, Alinnor C (2018) Optimising sand control and production strategies in a low permeability sandstone oil field. In: SPE international conference and exhibition on formation damage control. Society of petroleum engineers

Fourie B, Marpaung B, Jansen R, Wong A, Mok D, Karlsey N (2013) First installations of the 9-5/8-in. enhanced single trip multi-zone sand control technology in Offshore Brunei. In: IPTC 2013: international petroleum technology conference

Gupta A, Kamat D, Zulkapli M, Borhan N, Kobbeltvedt A, Hammersmark J, Sam A (2016) An alternate sand handling technology for efficient sand management: pilot and way forward. In: Offshore technology conference Asia. Offshore Technology Conference

Hodge RM, Burton RC, Constien V, Skidmore V (2002) An evaluation method for screen-only and gravel-pack completions. In: International symposium and exhibition on formation damage control. Society of petroleum engineers

Mondal S, Sharma MM, Chanpura R, Parlar M, Ayoub JA (2010) Numerical simulations of screen performance in standalone screen applications for sand control. In: SPE annual technical conference and exhibition. Society of petroleum engineers

Mondal S, Sharma M, Hodge R, Chanpura R, Parlar M, Ayoub J (2011) A new method for the design and selection of premium/woven sand screens. In: SPE annual technical conference and exhibition. Society of petroleum engineers

Muryanto B, Fransiskus W, Wijaya R, Styward B, Ji Y, Albertson E, Widyastuti A (2018) Applications of a multizone single-trip gravel-pack system in developing a shallow-gas field. In: Offshore technology conference Asia. Offshore Technology Conference

Ojeh-Oziegbe O, Olatunji I, Alawode O, Walker J, Murdoch E, Patel D, Aye Y (2019) Successful installation of the first deep water single trip stand-alone screens in the industry saves rig time on bonga project. In: Offshore Technology Conference. Offshore Technology Conference

Saeby J, Lange F, Aitken S, Aldaz W (2001) The use of expandable sand-control technology as a step change for multiple-zone smart well completiona case study. In: SPE Asia Pacific oil and gas conference and exhibition. Society of petroleum engineers

Toelsie S, Prediepkoemar G (2013) Sand control in shallow unconsolidated sandstone oil reservoirs at staatsolie NV suriname. In: SPE European formation damage conference & exhbition. Society of petroleum engineers

Wu B, Choi S, Feng Y, Denke R, Barton T, Wong C, Madon B (2016) Evaluating sand screen performance using improved sand retention test and numerical modelling. In: Offshore technology conference Asia. Offshore Technology Conference

Zeidan A, Al-Bader H, Pandey C, Al-Ibrahim A, Ayyavoo M, Al-Ateeq A, Bosilca D (2018) Successful installation of stand alone screen in challenging environment in Umm Niqa Field. In: SPE international heavy oil conference and exhibition. Society of petroleum engineers

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Ahad, N.A., Jami, M. & Tyson, S. A review of experimental studies on sand screen selection for unconsolidated sandstone reservoirs. J Petrol Explor Prod Technol 10, 16751688 (2020). https://doi.org/10.1007/s13202-019-00826-y

advances in improved/enhanced oil recovery technologies for tight and shale reservoirs - sciencedirect

This paper presents a comprehensive review of the technical progress as well as updated knowledge and understandings of IOR/EOR technologies for tight oil reservoirs. Critical and in-depth assessment of various IOR/EOR methods is made upon the best practice and lessons learned, mainly, in the North America. In the past few years, many traditional and new IOR/EOR methods have been tested in laboratory and piloted in field to investigate their potential in improving oil recovery from unconventional plays, including water injection, miscible and immiscible gas injection, water-alternating-gas injection, chemical flooding, and nanotechnology. Feasibility concerns and technical challenges, such as low injectivity, formation damage, and low sweep efficiency arising from extremely low permeability and high heterogeneity in fractured tight oil reservoirs, are raised for directly adopting traditional IOR/EOR methods. IOR/EOR mechanisms in tight oil reservoirs mainly involve gas and oil flows in nanometer pores, gas dissolution and diffusion through low permeability matrix, oil swelling, wettability alteration, IFT reduction, and fracture-matrix interaction, thus thorough understanding of flow and transport mechanisms in multi-scale pores and fractures is indispensable for developing effective IOR/EOR technologies. To optimize the selection of specific gas species or chemical formulas, it is necessary to conduct preliminary assessment of practicability and viability with both experimental studies and numerical simulations for operation upscaling and production prediction before field implementation.

remanufacturing - an overview | sciencedirect topics

Remanufacturing is performed in a factory setting with supporting tool and test sets that are equivalent to those used in current production, with instructions contained in floor controlled process documentation.

Remanufacturing technology is an effective way to reduce waste and environmental pollution. This area is a developing new research field, and a growing and developing advanced manufacturing technology, offering an extension to the whole life cycle of many manufacturing processes. Remanufacturing technology provides an important technical support to industrial sustainable development, bringing great benefits to the development of the national economy, and becoming a new point of economic growth [15].

Laser remanufacturing uses laser beams as a heat source to renew and improve failed metal components, whilst simultaneously improving their performance. Based on laser surface modification technology, specially developed alloy materials with high erosion and corrosion resistance (as required) are added to the metal components, and optimum processing parameters are chosen.

The three major component reuse options are not equal but rather exist on a hierarchy with remanufacture at the top, followed by reconditioning and then repair. Remanufacturing is a process of returning a used product to at least original performance specification from the customers perspective and giving the resultant product a warranty that is at least equal to that of a newly manufactured equivalent (Ijomah, 2002; Ijomah et al., 2004). Currently, remanufacturing is profitable typically for large complex mechanical and electromechanical products with highly stable product and process technology (Ijomah, 2002; Ijomah et al., 2007a), materials and components that are costly to manufacture or may become costly in the future. The value of reusing these products components relative to the cost of disassembly makes manual disassembly worthwhile, which enables profitable remanufacture of these products.

Remanufactured products have warranties equal to that of new alternatives whilst repaired and reconditioned ones have inferior guarantees. Typically, with reconditioning the warranty applies to all major wearing parts, while for repair it applies only to the component that has been repaired.

Remanufactured products lose their identity while repaired and reconditioned products retain theirs because in remanufacturing all product components are assessed, and those that cannot be brought back at least to original performance specification are replaced with new components.

Table8.1 defines and differentiates repair, reconditioning and remanufacturing. Figure8.1 shows the three processes on a hierarchy based on the work content that they typically require, the performance that should be obtained from them, and the value of the warranty that they normally carry.

The key advantage of remanufacturing over reconditioning and repair is that it permits an organisation to combine the key order winners of low price and product quality, especially as remanufacturing also includes increasing the performance and quality of the used product beyond that of its original standards when new. This ability of remanufacturing to deliver high quality is especially important to A class manufacturers and customers who value the reputation of their service and brand name above low product cost. Xerox is a key example of successful remanufacture because its copiers typically undergo seven life cycles. This means that seven revenue streams are generated from the manufacture of a single product, and materials are diverted from landfill or recycling at least six times (Gray and Charter, 2006).

The disadvantage of remanufacture to the lesser product recovery processes is that it is generally more expensive because of the greater resource and work content involved. Thus there are many products where remanufacturing would be cost prohibitive given current remanufacturing technology and knowledge base. Domestic appliances remanufacturing for example, would not be viable as a profitable business. This is because the cost of processing items such as fridges and cookers for recycling continues to decrease and according to AMDEA (2008) would be less than 5.00 in 2009, whilst the value obtained at the treatment plant continues to increase. Also, the value of steel doubled between 2002 and 2006 (AMDEA, 2008), thus increasing the profitability of recycling relatively low price goods with good metallic content. Interviews by the authors of major domestic appliance manufacturers such as Lec Refrigeration and Merloni indicate that remanufacturing of domestic appliances is cost prohibitive at least within the EU. The main reason here is the cost of manual labour involved in remanufacturing as well as additional costs such as that for testing to safety standards. Such tests are expensive to run and their costs in new manufacture can be limited by running in batches; however, with remanufacturing the test must be undertaken individually.

The inventory management objective is controlling the external ingredient orders and the internal ingredient recovery process, for supporting a needed service level and minimizing the variable and fixed cost. There are two folds following the returns flows effects. One side is that overhauling an old product is cheaper than producing the new one. The other side increasing the uncertainty, which can lead to higher safety stock levels, cause to reliable planning becomes more difficult. The process of recovery, in reuse system might vanish with the returned products which enter to the usable inventory directly.

The other systems input parameters which require to be described externally for system are the predicting of future returns and an appropriate economic valuation of the returned items. Three essential aspects differs from those in traditional inventory control systems. First, as the return flow results the level of inventory between new item replenishments, is no longer necessarily reducing but can rise also. Second, for satisfying demands impose, external orders and recovery have to be coordinated. Third, by differentiate between products yet to be repaired and usable the situation explained at the above naturally causes to a two-echelon inventory system. Therefore, in this context, surveys on adequate strategies for echelon stock control, such as PULL against PUSH policies are relevant.

Using the remanufacturing operation, a usable stock could be improved into a newly manufactured stock. Figure 13.5 demonstrates an inventory system with remanufacturing. The model has the following assumptions:

The objective is to minimize total setup and holding costs. Two alternatives are considered: (1) a joint setup for manufacturing and remanufacturing when the same production line is used for both processes, and (2) separate setups for manufacturing and remanufacturing.

Remanufacturing and upcycling of materials obeys make-made reuse strategy. Likely, leftover textiles such as discarded linens are used as vintage fabrics to produce aprons, banners, raincoats by a steam laundry. Traditional craft skills are to reuse discarded materials in a distinct design for exclusive product looks. The unique trait of companies in remanufacturing loops is resulting in a premium sales price with the unique designs, developing business models will support materials from both within and between industries. Thus, the remanufactured textile gains attraction in market, representing the principles of circularity system. These are reassured with high quality of product having no chemical contamination, clean and fitting (www.suschem.org). Large volume textile recycling system is operated by collecting waste from collection bins at common places to reprocess. Those reused in the domestic market can be resold and half of them can be exported. Some of them can be used for threads, fibers, and synthetic felt and for energy recovery.

The charity organizations make profit out of secondhand clothing through sales. The textile collection programs will teach how to use garments responsibly and its value to customers. They offer discounts to customers as returning their old clothes. This collection system improves in-store traffic and customer loyalty. Besides, the used garments are sorted to attain significant economic return. Cascaded use is important in circular business model, that secondhand clothes undergo subsequent loop of remanufacturing and end up in the outermost loop of recycling.

Carbon Fibre Remanufacturing (CFR) develops new products for industrial, commercial and retail product applications using reclaimed carbon fibre. The company developed an alternative to landfilling of excess postindustrial carbon fibre materials. It was estimated that 30% of carbon fibre from the composites industry becomes excess or part of the waste stream in the manufacturing process. The company developed carbon fibre fire-retardant fabric, insulation, friction/brakes, filtration, thermal barriers and laminate compounds using reclaimed carbon fibre waste from composite industries. Remanufactured carbon fibres retain 100% of their virgin properties. They can be cut to specified lengths; be incorporated into nonwoven rolled cloth, veil, mat/felt product; and compounded to specifications for bulk and sheet-moulded product applications (Van de Velde and Kiekens, 2002). CFR is in various stages of moving excess postindustrial carbon fibres into various carbon fibre material forms as well as working with an increasing number of fabric and composite firms in the development of final whole goods (Anon, 2015). Reclaimed carbon fibres are optimized when produced in various broad goods, including:

Applying strategies to increase remanufacturing can be highly beneficial if the hardware business is to be combined with consumables or software. Examples of this are inkjet printers requiring brand-specific ink cartridges and game consoles requiring software. In such cases most of the profit is made through the consumables and/or software. It pays therefore to maximize the 'fleet in the market'. Even if there is a financial deficit in the remanufacturing, the total business proposition can be very positive. Such a strategy can be enhanced by actively promoting trade-in/trade-up schemes for first users; in this way the number of products with an age lower than the length of the technology cycle can be substantially increased. In a proprietary study (Rose et al, 2002), the authors demonstrated that such a combination of strategies can be environmentally very beneficial and economically attractive. Implementation failed, however, because the company concerned decided to only sell new products.

Active engagement in remanufacturing and reuse can also be through subcontracting third parties which specialize in the field. When the flow of goods is sufficiently controlled, the risk of damage to the brand image and of price erosion of new market launches is very low. However, if there is no such control and discarded goods are exported to second or third world countries before remanufacturing and reuse, there is much more potential for negative fall outs. This happened for instance when the borders of Eastern Europe opened in 1990. This is relevant today as well: although the export of e-waste is forbidden by the Basel Convention, export for reuse is allowed in most cases. It is estimated in the Netherlands that some 15% of the total WEEE disappears out of the country through (il)legal exports. It is suspected that most of this will end up in the informal recycling sector or will be improperly reused; in practice very few of it will end up in state-of-the-art remanufacturing.

In spite of all this, there is considerable potential for reuse and remanufacturing of products in the developed world. This can be concluded by comparing the figures for the units sold, and the number in the market. For TVs this points to a period of 8 years for first use. The average life at discarding is, however, 12 years. This suggests a large amount of secondary use. Other indications are the observation that the amount of discarded products coming back through trade is relatively low. Apparently in this sector there is a business in post-consumer goods. Also trading of goods returned at municipal recycling centres occurs in some countries in Europe, municipalities even have official contracts with dealers in secondary goods (or have export contracts). All of this takes place outside channels used by producers; in the opinion of the author there would be an environmental potential and added value if producers would engage themselves more actively in secondary markets. This potential is bigger than anticipated on numbers alone. Many users feel negatively about a brand when products have to be discarded completely because of something which is perceived by them (or really is) as a minor defect. Also high examination and repair costs raise a lot of anger. Even if a product really has to be discarded there is still the idea that it has value. Producers who are capable of dealing well with such issues generate a lot of goodwill and as a consequence, a lot of brand loyalty.

Add-ons to these secondary consumer markets could be given by including returns from business to business commercial activities and repaired production rejects. Even in such a case reuse and remanufacturing activities will stay relatively limited. The WEEE Directive therefore rightly focuses on material recycling.

The item reuse, remanufacturing and restoration, requesting less assets and vitality are monetary too than regular reusing of materials as poor quality crude materials. The time the incentive in the assets spends/lives inside the inward circles ought to be amplified. Materials should first be recuperated for reuse, restoration furthermore, repair, at that point for remanufacturing and later for crude material usage, which has been the fundamental concentration in conventional reusing. As indicated by circular economy, burning for vitality ought to be the second to last choice while landfill transfer is the last choice (Korhonen etal., 2018). Along these lines, the item esteem chain and life cycle hold the most noteworthy conceivable esteem and quality to the extent that this would be possible and is additionally as vitality productive. Once a crude material is separated, refined, and delivered with the standard costs, it bodes well to utilize the esteem created to the extent that this would be possible, i.e., keep the item work/administration and utilize an incentive in financial flow to the extent that this would be possible.

Retreading of nanocomposite tyres belongs to the remanufacturing category and fits step 2 in the resource cascade of Table11.1. Remanufactured (or refurbished) cell phones, in which nanomaterials may be applied, have a thriving market (Geyer and Blass, 2010). In the case of three-way catalytic converters, processes for in situ regeneration by removing poisonous contaminants (Ca, Mg, P, Pb, S, Zn) have been reported (Christou et al., 2007; Subramanian et al., 2011), which fit step 2, remanufacturing, in Table11.2.

For a typical micro-injection-molded part, the runner system and other wasted parts can amount to 95 to 98% of the total mass, due to the minimum volume of a shot possible on available hardware that was not specifically made for micro-injection molding. Such parts can be seen in Fig. 25-4. The effects of recycling of polymer, both with and without fiber reinforcement, have been studied in [14]. Even if the mechanical characteristics of a part decrease with the number of injection cycles, it is possible to design a piece with a mechanical limit that allows a number of recycling cycles. Stress/strain curves in relation to the number of successive injections or recycled polymer are available in the literature.

After a certain amount of injections, it is shown that there is a loss in mechanical properties (mainly due to fiber degradation), but these characteristics may still be in the range of use for some other design. Moreover, by starting with mechanical properties greater than necessary it is possible to stay within expected limits after a number of injections. Of course, a careful choice of material is needed in order not to select a very unenvironmentally friendly material, hence losing on the one hand what is won on the other.

A DFE way of solving this problem could be the use of carbon nano-tube reinforced polymers. Indeed, nano-tube polymer composites not only behave differently in use, but also in manufacturing processing [15]. Of course, such composites will also require LCA and health-hazard studies. The recycling of carbon nano-tubes is usually targeted at recycling them without alteration or cost-related issues. Environmentally-benign ways of recycling polymer-based composites with nano-tubes are investigated in [16,17]

In addition to considerations of reuse, remanufacturing, materials recycling and energy recovery for fuel cell hardware, LCA can be used to quantify and interpret a broader set of impacts of fuel cell systems from extraction of the raw materials for fabrication, through the fabrication process and system operation, to ultimate disposal of the equipment. The LCA protocol has been standardized by the ISO (19972003) to include four research phases. The first phase, goal and scope definition, describes the reasons for carrying out the study, the intended audience, geographic and temporal considerations, the system function and boundaries, data categories, comparative (or reference) systems, impact assessment and interpretation methods, and plans for critical review. Next, in the inventory model, the life cycle is subdivided into a set of unit processes such that each encompasses the activities of a single operation or a group of operations. The inventory model quantifies material and energy use and waste by each unit process such that processes are linked to one another by economic flows (flows within the economic system such as the production of biomass or the use of fuel or steel) and consume and produce environmental flows (flows out of and into the environment such as the consumption of energy, land or iron ore, or carbon dioxide emissions). Thirdly, impact assessment estimates the contribution of environmental flows to environmental benefits (such as habitat protection) and impacts (such as global warming). Finally, the interpretation step identifies sensitive parameters as well as quantifies uncertainties in results.

Because the details of hardware design are currently primarily held in the private sector and because fuel cells have yet to move from pilot or demonstration projects to wide-scale production, fuel cell LCAs have typically focused on a single prototype with just one or two assumed power densities dictating the size of the stack (i.e. the same set of materials are assumed to have two possible power densities) or a single prototype created by alternative fabrication processes (i.e. all processes are assumed to fabricate stacks of equal performance). A preferred approach builds LCA models around computational models (electrochemical and thermodynamic) of the stack and balance of plant. Such models estimate system use of fuel and generation of electricity (their efficiency) and hardware needs (as material mass and including hardware replacement) over a specified system life and can be used to compare fuel cell systems to each other (e.g. design variants of PEMFCs can be compared; PEMFC systems can be compared to PAFCs) and to systems based on other generation technologies (e.g. fuel cells can be compared to combustion technologies).

In such LCA models, it is critical to consider the origin of the fuel, and the performance of both the stack and the balance of plant over the system life for a given set of fuel cell materials. With respect to the origin of the fuel, an opportunity exists to use fuel cell materials for appropriate fuel constituents to maximize efficiency, noting at the same time that there is significant variation in the life cycle energy and emissions in fuel production (Table 5.2) and that materials and energy use for on-site reforming of liquid petroleum and natural gases is important to consider. When considering stack and balance of plant performance, fuel cell materials, fuel utilization by the stack and stack durability (including materials degradation) can be very important to life cycle results.

Source: based on data for 2007 from the US Department of Energy Argonne National Laboratorys Greenhouse Gases, Regulated Emissions and Energy Use in Transportation (GREET) Model version 2.7 available at http://greet.anl.gov/

At this point we would like to leave the field of low temperature PEMFCs for an excursion to solid oxide fuel cells (SOFCs) and consider an example of an LCA for SOFC modules (with each module combining many stacks within a steel pressure vessel) that are mass-produced for central power production. It can be assumed that the length of the stack flow field can be related to not only the mass of stack materials (e.g. the amount of stainless steel or ceramics used) but also to the stack efficiency, which subsequently dictates in part the number of modules needed to achieve the desired power. Including consideration of the balance of plant, the length of the flow field also dictates the amount of fuel available for combustion off the stacks (also related to the chemistry of balance-of-plant emissions) and therefore the energy that might also be obtained by the overall system. Finally, it is assumed that the SOFC plant is to be compared to construction of a natural gas power plant expected to last for 40 years. If the degradation of fuel cell materials or the design configuration dictates that each SOFC module be replaced every five years, then the full life cycle (materials production through manufacturing through collection and recovery) of the SOFC models must be computed for eight system replacements as compared to the construction of a single natural gas plant. As a result, hardware design including recovery scenarios might become critical to life cycle improvements.

Table 5.3 presents related LCA results. In the table, life cycle energy consumption, plant steel consumption, life cycle air emissions, generation of recyclables, solid waste generation, resource consumption and materials production water consumption are compared for a baseline SOFC system with stacks using a YSZ electrolyte and 70% externally steam reformed natural gas + 30% internally steam reformed natural gas system at 90% fuel utilization and design variants. Specifically, changes due to variation of the fuel utilization, the use of hydrogen fuel, the use of a SSZ electrolyte in place of YSZ, variations on the manufacturing sequence and an aggressive SOFC remanufacturing scenario are included. The remanufacturing scenario investigated here assumes all materials except the ceramics can be reused over the 40 year period. Further research is needed to provide better estimates of the materials and energy flows associated with SOFC remanufacturing. In Table 5.3, a + indicates that the design or manufacturing alternative improves upon the baseline, a indicates that the design or manufacturing alternative deteriorates from the baseline, and shaded cells indicate apparent optima or combinations of design alternatives that appear to optimize a single flow.

SOFC stack remanufacturing will reduce the contribution to all impacts from the baseline more than any other design alternative investigated because life cycle impacts are expected to track with plant steel consumption not fuel consumption. If remanufacturing is not an option, the use of recycled steel improves the life cycle profile.

Whereas the use of the manufacturing sequence multi-fired cast anode and deposited elect and cathode at an estimated stack life of six years reduces the contribution to all impacts from the system which casts two out of the three components, the use of co-fired sequences at an estimated stack life of four years are expected to increase the contribution to all impacts.

Thus, by considering differences in fuel cell materials and configuration, LCA can provide very specific plant and system design guidance. Further understanding opportunities for fuel cell hardware recovery, especially for component remanufacturing and materials recycling, can be important to LCA results. In this way, LCA can contribute important findings as fuel cells move from laboratory and pilot status to wide-scale production.

livestock integration into soybean systems improves long-term system stability and profits without compromising crop yields | scientific reports

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Climate models project greater weather variability over the coming decades. High yielding systems that can maintain stable crop yields under variable environmental scenarios are critical to enhance food security. However, the effect of adding a trophic level (i.e. herbivores) on the long-term stability of agricultural systems is not well understood. We used a 16-year dataset from an integrated soybean-beef cattle experiment to measure the impacts of grazing on the stability of key crop, pasture, animal and whole-system outcomes. Treatments consisted of four grazing intensities (10, 20, 30 and 40cm sward height) on mixed black oat (Avena strigosa) and Italian ryegrass (Lolium multiflorum) pastures and an ungrazed control. Stability of both human-digestible protein production and profitability increased at moderate to light grazing intensities, while over-intensification or absence of grazing decreased system stability. Grazing did not affect subsequent soybean yields but reduced the chance of crop failure and financial loss in unfavorable years. At both lighter and heavier grazing intensities, tradeoffs occurred between the stability of herbage production and animal live weight gains. We show that ecological intensification of specialized soybean systems using livestock integration can increase system stability and profitability, but the probability of winwin outcomes depends on management.

Production of domestic animals and crops have been interconnected since the early days of agriculture and integrated crop-livestock systems (ICLS) remain the cornerstone of smallholding systems and global food security1,2,3. These systems are characterized by cropland grazing and usage of services provided by animals (e.g., nutrient recycling and weed control) to reduce input needs and enhance crop yields4,5. Yet, industrialization and intensification of farming systems has led to increasingly specialized operations and decoupling of crops and livestock in the last decades6,7. These highly productive specialized systems, such as monoculture cropping and feedlots, often rely heavily on external inputs and have high environmental costs, including contamination of water resources8,9 and greenhouse gas emissions10,11.

Biologically simplified agroecosystems are also more vulnerable to extreme weather events12,13,14,15 and projected increases in the frequency and severity of droughts and heavy rainfall events with climate change will challenge current crop production models16,17,18. In addition, world population is projected to increase by 25% and reach 9.8 billion people by 205019. Given the rising demand for food, developing and adopting high yielding sustainable production systems able to maintain crop yields under different weather scenarios is critical to maintain global food security in an increasingly challenging production environment20,21,22.

Re-coupling crops and animals to form more biodiverse agroecosystems is increasingly proposed as a strategy to reconcile high levels of food production with maintenance of fundamental ecosystem services underlying sustainability2. ICLS are designed to harness complementarities and synergies between soil, plants and animals across various spatiotemporal scales (e.g., within farms through seasonal pasture-crop rotations23 or grazing of understory vegetation in perennial cropping systems24). In subtropical Brazil, ICLS are usually implemented in the form of an annual rotation of cash crops followed by grass cover crops grazed by beef cattle23. Out of 14.7 million ha currently cultivated with summer cash crops in the region (soybean, maize and rice), only 2.5 million ha are rotated with winter cash crops and 2.2 million ha with off-season maize25, while the rest often has winter cover crops in no-till systems which could be grazed23. Considering that this is the Brazilian region with the highest levels of ICLS adoption (13% of the total area cultivated with crops is integrated with livestock)26, there is a large unexplored potential for ICLS implementation. Implementing commercial-scale ICLS is a complex challenge27 and concerns over the impact of livestock on subsequent crop yields and the greater managerial intensity and knowledge demanded by ICLS have been listed as the main barriers to adoption23,28,29,30.

Various reports have reviewed the benefits and tradeoffs of crop-livestock integration for soil quality, nutrient cycling, crop and animal production and farm economic performance in a wide range of systems and regions of the world23,28,29,30,31,32. Livestock integration increases land-use efficiency and farm profitability while providing opportunities to bolster ecological mechanisms underlying resilience33. Income diversification can reduce risks from uncontrolled variability in climate and market fluctuations, as annual returns from crop and livestock commodities are often uncorrelated28,34,35. At the field scale, self-regulating processes such as greater nutrient cycling32, higher microbial functional diversity36 and improved soil structure37 and organic matter38 in grazed systems are suggested to increase systems biophysical buffering capacity to less optimal environmental conditions, in ways that still require better understanding14,37. Livestock production within ICLS also takes advantage of crop residues and grasses inedible to humans to produce high-quality food (e.g., beef and milk byproducts), thus reducing market competition for human-edible feed resources39. However, if we aim to use crop-livestock integration as a tool for sustainable intensification, it is imperative to assess its contribution to not only system productivity but also stability over the long-term.

Stability has multiple meanings in ecology and statistics and encompasses concepts like resistance and resilience14,40,41,42. In this study, we considered the concept of stability as related to variability and defined a stable system as one that changes least in response to environmental changes43. Management approaches that promote biodiversity (e.g., organic agriculture and crop rotation diversity) and conservation practices (e.g., permanent soil cover and reduced disturbance) have been shown to enhance yield stability13,15,22,42,44,45,46,47,48,49,50,51,52,53. Besides income diversification, more biodiverse systems can stabilize agroecosystem productivity through cross-scale mechanisms ranging from redundancy and facilitation in plant communities41,53, to creating habitat for natural enemies to promote pest suppression12. Conservation practices, in turn, can improve properties related to soil health and crop yield stability, such as soil organic matter and water retention49,50. However, the effects of increasing system diversity and ecological complexity by adding a trophic level (i.e., grazing animals) on the long-term stability of no-till cropping systems have not yet been studied with the same level of detail.

The primary goal of this study was to evaluate long-term yields and stability of ICLS yields and profitability compared to non-integrated systems under a range of environmental conditions and test their potential as a strategy for sustainable intensification. We hypothesized that increased biodiversity and ecological complexity created by crop-livestock integration in no-till systems improve yields while decreasing vulnerability of system yields and profitability to weather variation. We tested this hypothesis using a 16-year dataset from a long-term, no-till integrated soybean-beef cattle system in southern Brazil and measured the impacts of cover crop grazing at different intensities during the winter period on probability of high and low performance13,15,44, minimum and maximum yield potentials and stability of key crop, pasture, animal and whole-system outcomes using established metrics of stability22,44,45,46,47,48,49,50. Our results provide insight into the long-term stability of subtropical soybean systems performance and the potential of livestock integration to build up sustainability and resilience in agriculture.

The experiment was established at Espinilho Farm, in the municipality of So Miguel das Misses, Rio Grande do Sul State, southern Brazil (28 56 14 S, 54 20 52 W, 465m above sea level) in 2001. The region has a warm, humid subtropical climate (Cfa, Kppen classification system) with an average annual temperature of 18.6C and average annual precipitation of 1898mm54. Temperature and precipitation during the experimental period analyzed here (20012016) were collected by a weather station located at the experimental site (Supplementary Fig. S1). Missing weather data points were estimated using linear regression with values from the nearest meteorological station as predictor (National Institute of Meteorology, Cruz Alta, 78km from the study site 28 36 12 S, 53 40 25 W, 427m a.s.l.). The soil in the experimental site is an Oxisol (Rhodic Hapludox)55, with clayey texture (540, 270 and 190gkg1 of clay, silt and sand, respectively) and a deep, well drained profile.

The area has been managed as no-till soybean [Glycine max (L.) Merr.] cropland since 1993. In 2001, 22 hectares of land began to be managed as an integrated soybean-beef cattle annual rotation with a mixture of black oat (Avena strigosa Schreb.) and Italian ryegrass (Lolium multiflorum Lam.) pastures grazed during the winter between soybean crops. Soybean was direct-seeded after the animals were removed from the experimental area, typically in November, and harvested after 14211days. After soybean harvest (AprilMay), experimental plots were drill-seeded with black oats into the volunteer ryegrass sward from the previous winter, immediately followed by broadcast seeding of ryegrass to ensure successful establishment for both species in all treatments.

The experiment was established as a randomized complete block design with three replicates. Treatments consisted of four grazing intensities (intense, moderate, moderate-light and light) defined by contrasting sward heights under continuous stocking (10, 20, 30 and 40cm, respectively) and an ungrazed control with the same pasture species used as winter cover crops. Plot areas were 0.1ha for the ungrazed treatment and ranged from 0.8 to 3.6ha for grazed treatments. Plots differed in area to reduce the number of animals required to maintain the target treatment heights, especially for shorter swards.

Fertilization rates and soybean cultivars changed according to recommendations over the years but were the same for all plots, including the ungrazed treatment. An average of 160kgha1 urea (46% N) was applied yearly, split into two equal winter applications during the stocking period: (1) when pasture reached V3V4 growth stage (i.e., plants with 34 fully expanded leaves on the main stem) and (2) just before animals entered the experimental plots, approximately 1month after the first application. From 2001 to 2011, P and K (on average, 60kgha1 P2O5 and 70kgha1 K2O) were applied at soybean sowing. From 2012 to 2016, P and K (on average, 45kgha1 P2O5 and 60kgha1 K2O) were applied at pasture sowing to take advantage of the improved nutrient recycling provided by the grazing animals for primary production. The exact amount of fertilized applied each year was based on standard recommendations56 and soil analysis.

Grazing usually started in JuneJuly, when average sward height reached 244cm (or 1485379kgha1 of dry matter) and lasted 12416days. To ensure that treatments remained close to their nominal targets (Supplementary Fig. S2), sward height was measured at 100 random points per plot every 15days with a sward stick57. Three tester animals remained permanently in the plots over the stocking period and put-and-take animals were added or removed to adjust sward heights58. Average stocking rates used to maintain target sward heights throughout the stocking period were 376, 651, 948 and 1331kg of live weight ha1 for light to intense grazing. Experimental animals were crossbred AngusHerefordNelore steers with initial body weight of 21023kg and 12months of age on average. Steers were weighed at the beginning and at the end of the stocking period after 12h of fasting.

We assessed five key indicators of crop, pasture, animal and whole-system performance: (1) soybean grain yield; (2) total herbage production; (3) animal live weight gain; (4) human-digestible protein (HDP) production; and (5) profitability. Year in all analyses refers to the year when soybeans were sown. To avoid bias, specific years were removed from the analysis when data for one or more treatments were missing for a variable. Years 2001, 2003 and 2008 were excluded from the analyses of soybean yield, protein production and income. Years 2001, 2004, 2005, 2006, 2007 and 2012 were excluded from the analyses of herbage yield. Year 2012 was excluded from the analysis of animal production, protein production and income.

Soybean yield (kg of grains ha1) was determined at full grain maturity at 13% moisture content. Total herbage production (kg of dry matter ha1) was calculated as the sum of pasture herbage mass on the first grazing day and the daily herbage accumulation rates over the whole stocking period. Daily herbage accumulation rates were estimated every 28days using grazing exclusion cages59, following a standard protocol described by Nunes et al.60. Steer live weight gain (kg of live weight ha1) was calculated as the product of number of animals per hectare, average daily gain (kg of live weight steer1day1) of the tester animals and number of grazing days of the stocking period.

We adopted human-digestible protein (HDP) as a metric to account for added production from the livestock component when comparing integrated to non-integrated systems. Livestock contributes to supplying human protein demand as much or more than crop production39,61 and do so by converting proteins from non-edible (grass) into edible forms. HDP is not intended as a comprehensive nutritional analysis; rather, it is an unbiased indicator of whole-system food production62. Total HDP production (kg ha1) was calculated as the sum of protein from human-edible sources (i.e., animal and crop components of the system) multiplied by protein digestibility of the products (beef and soybeans)61. We estimated the protein content of a 350kg live weight steer at the end of the stocking period as 19% of its body weight, based on National Research Councils equations63. Soybean protein content was assumed to be 35% for a grain moisture content of 13%64.

We used gross profit (USD ha1) as a metric of profitability, calculated as: (1) the difference between operational costs of soybean and cover crops and revenues from grain sales in the specialized system (Supplementary Table S1); and (2) the difference between costs of soybean, cover crops and livestock operations (Supplementary Table S2), opportunity cost of capital invested in beef cattle [calculated as the product of average stocking rate, cattle price and saving account interest rate equivalent to the average number of grazing days (~2% interest rate) according to the Central Bank of Brazil, Supplementary Table S2]65 and revenues from animal and grain sales in the integrated crop-livestock systems. Revenues were calculated using yearly market sale prices of beef cattle and soybean grains on November and April, respectively (Supplementary Table S3)66,67. Historic nominal prices were transformed into real values using the General Market Price Index (IGP-M) from the Getlio Vargas Foundation (FGV), Brazil, using 2016 as the base year for the analysis68 and converted from Brazilian Reals (BRL) to U.S. Dollars (USD) using the exchange rates of the respective months69. We used steer live weight gain to calculate income from livestock, assuming similar price per unit mass of beef at purchase and sale. Annual costs of soybean production were obtained from Brazils National Supply Company for the study region70. Soybean costs were considered the same for all treatments, given similar crop management across experimental units. Annual costs of cover crop establishment, animal medicines and mineral supplementation for the period of 20022011 were obtained from the economic analysis done by Oliveira et al.33 in the same experimental protocol. For the period of 20122016 costs were estimated using linear regression. Both costs and revenues were detrended prior to the calculation of annual profits.

All statistical analyses were performed in R (version 3.6.1)71. Long-term crop, pasture, animal and whole-system mean yields were analyzed using the lme4 package for mixed linear models72 with treatments as fixed effects and years, blocks and plots within blocks as random effects (y~factor(year)*treatment+(1|block/plot)). Yield trends over the 16years were analyzed using linear mixed-effects models with treatments and years as fixed effects and blocks and plots within blocks as random effects (y~year*treatment+(1|block/plot)). Analysis of variance (ANOVA, Supplementary Tables S4 and S5) was performed and when significant effects were detected, treatment means were compared with Tukey test at 95% confidence level using the emmeans73 and lmerTest74 packages. Residuals of all analyses were visually checked for homogeneity of variance and normality was tested with quantilequantile plots using the R car package75. When the residuals were not homogeneous or the distribution was not normal, data were log or square root transformed as appropriate.

We assessed stability of production (soybean yield, total herbage production, animal live weight gain, human-digestible protein) and profitability using four different metrics of stability: (1) yield range, which is the maximum amplitude between minimum and maximum yield values in a time series44,45; (2) coefficient of variation and (3) standard deviation21,44,45; and (4) Finlay and Wilkinsons stability metric (FW) derived from the linear regression of treatment yield on the mean yield of the location/year, or Environmental Index (EI)44,45,46,47,48,49. Regression of detrended yield on EI, also called adaptability analysis47, can assess stability or treatment-specific effect across a range of environments49. Based on regression of detrended yield on EI, stable systems are those with smaller slope (less sensitive to changes in environment).

Yield range was calculated as the difference between the highest and the lowest yields for each variable over the experimental period. Coefficient of variation, standard deviation and FW regressions were calculated using detrended data. Detrending removed long-term linear trends potentially generated by treatments in order to only consider variability of the residuals around the mean of each treatment due to transient environmental conditions. Data were detrended by removing treatment effects and treatment-specific linear temporal trends using the residuals of the linear model y~year*treatment. The overall average of the response variable was added to the residuals to get intuitively more understandable values (addition of the same constant to all values does not affect relevant statistical results).

Detrended data were analyzed as a function of the Environmental Index (EI) for each year and treatment with the following model: detrended y~EI*treatment. EI was calculated as the average yield of all treatments for each year, so that the highest and lowest EI indicated the year of highest and lowest system performance respectively. FW regression slopes were calculated and compared using simultaneous general linear tests with the R multcomp package76.

Yield range, coefficient of variation and standard deviation were analyzed as a function of treatment and block (y~treatment+block) and when significant differences were detected, treatment means were compared with Tukey test at 95% confidence level (=0.05) using the R agricolae package77.

Treatments were ranked from the lowest (i.e., greatest stability, rank #1) to the highest value (i.e., lowest stability, rank #5) for each stability metric regardless of the statistical significance. The overall stability of each system output was ranked based on mean stability rank for the four stability metrics, such that treatments with higher overall ranks indicated higher stability of yield or profitability.

Minimum and maximum yield potentials were calculated based on predicted responses for the smallest and largest observed EI values for each studied indicator44,49,50. Treatment effects on minimum and maximum yield potentials were tested with Tukey test at 95% confidence level through the equation \( HSD=q{\sqrt{2 SE} }\), where HSD is Tukeys honest significant difference, q is the studentized range statistic obtained using the qtukey function from R stats package71, and SE is the standard error of the mean for the studied variable.

To determine the probability of extreme yield events over the given range of environmental conditions (EI), we modelled probability distributions of each treatments detrended data using the density function in R (Supplementary Code S1, adapted from Gaudin et al.13). Treatment distributions were compared to a randomized distribution created by bootstrapping data and ignoring treatment effects. Downside risk and probabilities of high performances were defined as estimated probabilities of achieving results below the 10th percentile and above the 90th percentile, respectively, for each of the studied indicators. 5000 randomizations were sufficient to stabilize the p values for every system output. Treatment effects on the downside risk or probability of high performance were identified when observed results were significantly different from the randomized distribution at the 95% confidence level beyond the determined percentiles.

Soybean yields were not affected by winter grazing of cover crops, regardless of the grazing intensity (p=0.375, Table1). Total herbage production increased with increasing sward height (p<0.001, Table1) but remained low in the ungrazed treatment. Steers live weight gain per unit area increased with grazing intensity (p<0.001, Table1). Addition of cattle to the system increased total human-digestible protein production by up to 13% (p=0.065, Table1). Profitability in the two highest grazing intensities was 38% greater than in the two lowest ones, and 112% greater than in the ungrazed treatment (p<0.001, Table1).

All variables, except for total herbage production, presented an increasing linear trend over time (Supplementary Fig. S3, Supplementary Table S5). None of the linear trends were significantly affected by treatments, as indicated by the absence of treatment by year interactions (Supplementary Table S5). When year was included as a factor (categorical variable) in the model, there was a significant treatment by year interaction for total herbage production and live weight gain (Supplementary Table S4). However, we were unable to detect a clear pattern in the interactions.

Soybean yield was the most stable when the pasture phase was managed at moderate grazing intensities (G20 and G30) according to the overall stability rank (Table1). Ungrazed (UG) and lightly grazed (G40) treatments were more sensitive to the environmental gradient than more intensively grazed treatments (FW slopes>1, Fig.1a, Table1), indicating lower stability. The ungrazed treatment presented the narrowest yield range but was ranked worst in all the other stability metrics, making it the least favorable to soybean yield stability (Table1). Intense (G10) and light grazing (G40) were similar and intermediate in overall stability.

Yield stability of (a) soybean yield (kg grain ha1), (b) total herbage production (kg dry matter ha1), (c) animal live weight (LW) gain (kg LW ha1), (d) human-digestible protein (HDP) production (kg HDP ha1) and (e) profitability (USD ha1) of soybean systems integrated with different levels of cattle grazing during the winter period. Environmental index (EI) was calculated as the yearly mean detrended yield. Dashed lines are the regression of detrended yields against the EI without treatment effects. G10: intense grazing (10cm sward height); G20: moderate grazing (20cm sward height); G30: moderate-light grazing (30cm sward height); G40: light grazing (40cm sward height); UG: ungrazed cover crop. Smaller slopes indicate greater yield stability.

Conversely, total herbage production was the least stable under moderate grazing intensities, with G30 and G20 ranking fifth and fourth, respectively, in all stability metrics (Table1). Both treatments were more responsive to changes in Environmental Index (Fig.1b). The UG control presented the most stable herbage production over the years, ranking first in FW slopes and yield range, and second in CV and standard deviation, followed by G10 and G40 (Table1).

Increasing grazing intensity reduced the overall stability of live weight gain (Table1). Light grazing (G40) ranked first for all stability metrics for live weight gain and, along with G30, was significantly more stable than G10 and G20 to the environmental gradient (p<0.05, Fig.1c, Table1).

Both human-digestible protein (HDP) production and profitability showed greater stability when pastures were grazed at moderate to light intensities (G30 and G40), while either over-intensification or the absence of grazing decreased system stability (Table1). The UG control had a 26% and 83% higher CV for HDP production and profitability, respectively, than the grazed treatments and was the only stability metric showing statistical significance for profitability (p<0.05, Table1). FW slopes for HDP production followed the same trends as soybean yields, with greater slopes (>1) for the G40 and UG treatments (Fig.1d, Table1). Profitability, in turn, trended together with live weight gains, with lower FW slopes for G30 and G40 and greater slopes for G10 and G20 (Fig.1e, Table1). Differently from live weight gains, however, G20 was less stable than G10 in all stability metrics except for CV (Table1). The combination of prices and live weight gains might have been the reason why G10 and G20 switched positions in the profitability rank.

The absence of grazing (UG) significantly (1) increased the downside risk for soybean yield (=0.01, Fig.2a) without significantly impacting the minimum yield potential (Table2); (2) increased risks of obtaining low HDP production (=0.01, Fig.2d); (3) increased risks of financial loss (=0.05, Fig.2e); and (4) had the lowest minimum profitability (215.42USDha1, p<0.05, Table2). Both downside risk and minimum yield potential can be used as proxies of system resistance, since they represent the ability of a system to avoid crop failure or financial loss under stressful environmental conditions44. That said, livestock integration increased system resistance to financial loss by 81% in the lightest grazing intensity (G40) and up to 188% in the highest grazing intensity (G10) compared to the ungrazed control in the harshest environmental conditions (p<0.05, Table2). Despite minimum HDP production not being significantly different among treatments, it was 55% greater in grazed treatments compared to UG and up to 69% greater than UG in the highest grazing intensity.

Effect of grazing intensity on the probability of obtaining high and low (a) soybean yield (kg grain ha1), (b) total herbage production (kg dry matter ha1), (c) animal live weight (LW) gain (kg LW ha1), (d) human-digestible protein (HDP) production (kg HDP ha1) and (e) profitability (USD ha1) of soybean systems integrated with different levels of cattle grazing during the winter period in southern Brazil. Shown are the probabilities of yielding below the 10th percentile (orange bars) or above the 90th percentile (blue bars). Statistically significant treatment effect was identified for higher probability of high/low yields at the 95% (*) or 99% (**) confidence level and for lower probability of high/low yields at the 95% (#) or 99% (##) confidence level. G10: intense grazing (10cm sward height); G20: moderate grazing (20cm sward height); G30: moderate-light grazing (30cm sward height); G40: light grazing (40cm sward height); UG: ungrazed cover crop.

Conversely, UG presented a lower risk of low herbage production (=0.01, Fig.2b) despite no statistical differences in minimum yield potential (Table2). G20 and G30 had higher probability of low herbage production (18 and 19%, respectively), but were not different from the random distribution (=0.05, Fig.2b). G20 probability of low live weight gain was significantly higher (=0.05). G40 presented significantly lower downside risk for live weight gain (=0.01, Fig.2c) but also a significantly lower minimum live weight gain potential (=0.05, Table2).

Treatment effect on the probability of high performance was larger for live weight gains than for the other variables. High (G10) and moderate (G20) grazing intensities significantly increased the chance of obtaining live weight gains above the 90th percentile (=0.01 and =0.05, respectively, Fig.2c). Conversely, moderate-light (G30) and light (G40) grazing intensities reduced the chance of high live weight gains (=0.05 and =0.01, respectively, Fig.2c) and maximum yield potentials relative to G10 and G20 (Table2).

We observed greater maximum profitability potential in G10 and G20 than in the UG control (1866.73 average vs. 1447.02, a 29% increase, p<0.05, Table2), but probability of high performance was not affected (Fig.2e). No changes in probability of high performance were detected for soybean yield, total herbage production and HDP production. Likewise, maximum yield potentials were not statistically different between treatments (Table2), despite the important difference in pasture dry matter production from G10 and UG to moderate to light grazing intensities (G20, G30 and G40) that ranged from 1445.87kg DM ha1 (G20 vs. UG) to 2300.98kg DM ha1 (G30 vs. G10).

The most important findings of our study were: (1) grazing did not impair subsequent soybean yields regardless of grazing intensity, but moderate grazing intensities favored long-term yield stability (Fig.3a); (2) herbage production was more stable over the years but significantly lower under heavy grazing and in the absence of grazing (Fig.3b); (3) live weight gains were generally greater but less stable at higher grazing intensities (Fig.3c); (4) grazing at moderate to light intensities increased the stability of HDP production, while over-intensification and absence of grazing increased system vulnerability to environmental oscillations (Fig.3d); and (5) livestock integration under lighter grazing intensities provided more stable profits over time, but risk of financial loss reduced and overall system profitability increased with grazing intensity (Fig.3e).

Tradeoffs between performance and stability of (a) soybean yield (kg grain ha1), (b) total herbage production (kg dry matter ha1), (c) animal live weight (LW) gain (kg LW ha1), (d) human-digestible protein (HDP) production (kg HDP ha1) and (e) profitability (USD ha1) of soybean systems integrated with different levels of cattle grazing during the winter period in southern Brazil. Values represent standardized ratio to the maximum value for each metric. Yield stability is the average rank of four stability metrics (Table1).

Integrated crop-livestock systems (ICLS) are proposed as one possible strategy towards the sustainable intensification of food systems2,23,29,30. In a context of climate change and increased environmental pressure, stability of agricultural systems performancenot just performance per seneeds to be evaluated to prioritize management strategies with the greatest adaptive gains. Mining data from long-term trials provides opportunities to comprehensively assess the performance and stability of key crop, pasture, animal and whole-system indicators when livestock is integrated into specialized cropping systems.

Grazing did not impair soybean grain yields regardless of grazing intensity, but moderate grazing intensities favored soybean long-term yield stability (Fig.3a). Our analysis of soybean yields supports previous studies showing that grazing is not detrimental to crop productivity37,78,79. The impact of livestock on subsequent crop yields has long been a concern, mainly due to potential soil compaction caused by animal trampling, consumption of cover crop biomass and nutrient export when animals are removed from the system23,78. Results from literature on ICLS have shown everything from decreases in subsequent crop yield80, to no effect37,78,79 and even increases23,30,81. In our systems, grazing is combined with low disturbance (i.e., no-till) which may help mitigate potential negative impacts such as soil compaction. When conservation agricultural practices are used and grazing is well-managed (i.e., assuming no overgrazing or abnormally wet years), effects on soil physical attributes such as increased soil density have been shown to be transient, restricted to soil surface, and of limited impacts on yields78.

Our study provides the first evidence of grazing-induced long-term yield stability in no-till soybean systems where crops and livestock were integrated under moderate grazing intensity (Fig.3a). Furthermore, our risk analysis has shown that the absence of grazing increases the risk of yielding below the 10th percentile in unfavorable years (Fig.2a), despite greater litter amounts covering soil in ungrazed plots60. The underlying processes of increased yield stability with moderate grazing may be associated with increased biological diversity and ecological interactions created by livestock integration. Properties associated with the maintenance of soil functions and crop stability such as soil aggregation32, microbial diversity36,82 and ratios of beneficial over detrimental soil nematodes82 were shown to be improved by moderate grazing in previous studies at this experimental site and may have provided better growing conditions for the soybean crop in stressful years. Moderate grazing intensities enhance root growth, exudation and turnover which, combined with manure deposition, can directly benefit soil aggregation and microbial activity and diversity32. This in turn can lead to greater soil physical stabilization, organic matter accumulation83,84 and nutrient cycling84. These soil health benefits, including more biodiverse soil communities, may be particularly relevant to maintain soil functioning under stress as shown in other systems85,86 and potential core mechanisms underlying crop yield stability.

Total herbage production increased with increasing sward height but remained low in the ungrazed treatment, and was the least stable under moderate grazing intensities, demonstrating a possible trade-off between yield and stability in forage crops (Fig.3b). No grazing (UG) and heavy grazing (G10) treatments were more stable but produced significantly less forage over the years (Table1). Moderate grazing intensities created more responsive forage growth to better environmental conditions and along with light grazing (G40) were able to produce~11 tons DM ha1, while G10 and UG reached less than 10 tons DM ha1 even in the best environment (Table2). Lower stocking rates in G40 and UG favored the maintenance of target sward heights (and consequently leaf area index) in dry years, so that daily herbage accumulation rates in these treatments were less affected by poor environmental conditions. These results support long established plantherbivore models87 and the existence of two stable steady-states between vegetation growth and animal consumption in grazing lands: a low-productivity stable equilibrium at low plant biomass (G10), and a high-productivity stable equilibrium at high plant biomass (somewhere between G40 and UG). Moderate grazing (G20 and G30) provided a mid-range unstable state at which pasture growth is high, but herbage mass and accumulation rates are more easily affected by disturbances (e.g., weather fluctuations, fertilization or grazing itself), thus requiring more frequent adjustments of stocking rate to keep sward heights close to the nominal targets87.

In the absence of grazing, forage yields presented a lower risk of low production in unfavorable years (Fig.2). Keeping a dense layer of residual biomass on the soil surface in no-till systems (during winter as cover crop/pasture and after winter, as straw) improves soil water retention37 and protects soil from erosion88 and weed outbreak89 with potential benefits to crops in rotation. For this reason, crop-livestock integration is seen by many farmers as detrimental to no-till systems. However, prior research at this site showed no direct impacts of greater litter mass on crop yields in the ungrazed system37. On the other hand, the greater herbage production under moderate to light grazing intensities and the reduced probability of low forage yields in the ungrazed system found in our study may help explain the increased soil carbon stocks found by previous authors in areas managed under these approaches compared to intensely grazed areas after a decade of crop-livestock integration at this site38. Previous studies in humid continental climate of the US have shown that agronomic practices able to increase soil water holding capacity and organic matter can buffer yield volatility of rainfed maize49, suggesting that our results may apply to different agroecosystems around the world.

The linear increase of live weight gains per unit area (Table1) with grazing intensity is consistent with previous studies and can be attributed to increased stocking rate required to keep pasture at target sward heights90. Constraints in animal dry matter intake when forage allowances are limiting could result in a quadratic response of live weight gain, with greater gains associated with moderate grazing intensities91,92. The shortest sward height used in our study in fact limits the intake93 and consequently the individual live weight gains90,93, but it was not restrictive enough to show the quadratic pattern when results were expressed on a per area basis because greater stocking rate compensated the decrease in individual performance.

Our analysis showed a clear trade-off between yields and stability of live weight gains (Fig.3c). Live weight gains were generally greater, but less stable at higher grazing intensities. Although pasture growth is less stable and requires more frequent stocking rate adjustments under moderate grazing intensities, more intense stocking rate adjustments are required at the extremities of the grazing intensity gradient. In other words, the closer to a stable state, the stronger the push (i.e., addition or removal of animals) in the opposite direction required to shift states will be87. In our case, this was translated as a strong removal of animals from the plots when swards got too short to allow pasture regrowth in higher grazing intensities, which probably resulted in less stable live weight gains. Besides being less stable, literature also shows that higher grazing intensities lead to greater greenhouse gas emissions, especially methane93. Thus, to sustainably intensify ICLS, a conciliatory stocking rate90,94 able to achieve high animal yields and overall system stability while keeping low environmental footprint should be pursued.

Intensification of ruminant production in the last decades has increased protein production per area of land use, but primarily as a result of increased use of feed concentrates and human-edible nutrients in developed countries10,39. However, addressing the ability of a system to sustainably increase food production must consider the quality of food produced for human nutrition as well as the ability of this system to produce food from human inedible resources39. Grazing at moderate to light intensities increased HDP production and stability, while over-intensification and absence of grazing increased system vulnerability to environmental oscillations (Fig.3d). Ungrazed cover crops represented a risk to food production in unfavorable years (Fig.2d), since low soybean protein yields are not buffered by livestock protein yields as in integrated systems. By comprising protein from both crop and animal components of the system, our HDP analysis can be used as a measure of land-use efficiency61. Despite lacking statistical significance, grazing improved land-use efficiency by up to 13% due to the contribution of grass-based beef, an animal-derived protein of higher quality in human nutrition metrics than plant derived proteins39.

The greater profitability of integrated systems, particularly in heavier grazing intensities (G10 and G20, Fig.3e, Table1), was similar to results from a previous study at this site33 but differs in the magnitude of the results. While we observed profits 38% greater in the two highest grazing intensities (G10 and G20) compared to the two lowest ones (G30 and G40), and 112% greater than in UG treatment, Oliveira et al.33 found 27% and 100% increases (averages of 669, 526 and 334 USD ha1, respectively). This difference might be explained by international meat prices, which raised steadily during the period comprised by their study (20012011) and remained relatively stable at a higher level after that95. Furthermore, two major droughts occurred during their study period and severely affected soybean yields and profitability of the systems. Moreover, soybean yields kept trending upwards from 2011 to 2016 (Supplementary Fig. S3). Another possible explanation is that those authors used the nominal purchase and sale prices practiced by the farmer at every beginning and end of stocking seasons, which might have led to lower profits because purchase prices per kg of yearling steers are usually higher than sale prices of steers at the end of the fattening period. Although one could argue that their method is more realistic than ours, we consider our method more reliable because it disregards potential benefits or disadvantages faced by the farmer when trading the animals over the years.

The decrease in stability of whole-system profits with the over-intensification or the absence of grazing was consistent with HDP production, with G30 and G40 being the most stable treatments (Fig.3e, Table1). However, while stability of HDP production was the lowest in the absence of grazing (UG), profits were less stable in G20 followed by UG and G10 according to the overall rank (Table1). Considering our ranking criterion, this outcome was a result of G10 and G20 ranking 4th and 5th in every stability metric except for CV, in which they ranked 1st and 2nd (Table1). The difference of CV to the other metrics is that it brings information of the variability relative to the mean. Therefore, according to the CV, livestock integration reduced the variability relative to the mean profit compared to UG regardless of grazing intensity, but when it comes to pure variability G10 and G20 were the least stable treatments.

These results countered our expectation that stability of profits would increase with stocking density, since literature suggests crops and livestock markets are uncorrelated, which would work as a buffer against climate and price fluctuations28,34. However, our results arise within the positively correlated beef cattle and soybean market prices in Rio Grande do Sul State during the experimental period (r=0.44, 95% confidence interval=0.300.56, p<0.001, Supplementary Fig. S4a). Brazilian market prices of soybeans and cattle were also highly correlated in the last 2 decades (r=0.88, 95% confidence interval=0.870.88, p<0.001, Supplementary Fig. S4b)96,97. Thus, while extra income from livestock might have buffered profit oscilations in G30 an G40 compared to UG (mainly through increasing system resistance in less optimal environments, Table2), outstanding system performance in optimal environmental conditions (see maximum yield potential, Table2) increased regression slopes and interannual variability in G10 and G20, resulting in the lower stability observed for these treatments (Table1). Furthermore, the significantly higher risk of yielding below the 10th percentile (Fig.2e) and the lower minimum profitability potential in UG (Table2) represent a riskier farm portfolio, while animal production in grazed treatments provide a mean to smoothing farm incomes in poor crop production years (Table2). These findings highlight the importance of using multiple metrics for studies associating system performance and long-term stability.

In conclusion, our data suggest that livestock integration into specialized soybean systems under moderate to light grazing intensities benefits whole system stability to environmental variability and confirm that grazing does not impair subsequent soybean yields in annual soybean-pasture rotations. Instead, it reduces the chance of crop failure in unfavorable years. Moreover, while livestock integration under lighter grazing intensities provides more stable profits over time, economic risk reduction and overall system profitability increase with grazing intensity, showing that probability and nature of winwin outcomes is a matter of management. Our results likely apply to other ICLS designs, but best pasture management remains paramount to achieve benefits and reduce potential tradeoffs. Our study also highlights the importance of long-term experimental protocols to understand complex temporal system responses such as yield stability and improve predictions and adaptation to climate change. Questions remain regarding what mechanisms are driving these results, especially for grazing-induced soybean yield stability, but intensification of ecological processes likely plays a pivotal role.

Peterson, C. A., Bell, L. W., Carvalho, P. C. F. & Gaudin, A. C. M. Resilience of an integrated croplivestock system to climate change: a simulation analysis of cover crop grazing in southern Brazil. Front. Sustain. Food Syst. 4, 604099 (2020).

Peyraud, J. L. & Peeters, A. The role of grassland based production system in the protein security. Grassland Science in Europe - The multiple roles of grassland in the European bioeconomy 21, 2943 (2016).

Lightfoot, C. W. F., Dear, K. B. G. & Mead, R. Intercropping sorghum with cowpea in dryland farming systems in Botswana. II. Comparative stability of alternative cropping systems. Exp. Agric. 23, 435442 (1987).

Li, M., Peterson, C. A., Tautges, N. E., Scow, K. M. & Gaudin, A. C. M. Yields and resilience outcomes of organic, cover crop, and conventional practices in a Mediterranean climate. Sci. Rep. 9, 12283 (2019).

van Zanten, H. H. E., Mollenhorst, H., Klootwijk, C. W., van Middelaar, C. E. & Boer, I. J. M. Global food supply: land use efficiency of livestock systems. Int. J. Life Cycle Assess. 21, 747758 (2016).

Companhia Nacional de Abastecimento (CONAB). Planilhas de custos de produo - Sries histricas. https://www.conab.gov.br/info-agro/custos-de-producao/planilhas-de-custo-de-producao/itemlist/category/414-planilhas-de-custos-de-producao-series-historicas (2019).

Kunrath, T. R., Carvalho, P. C. F., Cadenazzi, M., Bredemeier, C. & Anghinoni, I. Grazing management in an integrated crop-livestock system: soybean development and grain yield. Rev. Cincia Agronmica 46, 645653 (2015).

de Souza Filho, W. et al. Mitigation of enteric methane emissions through pasture management in integrated crop-livestock systems: trade-offs between animal performance and environmental impacts. J. Clean. Prod. 213, 968975 (2019).

The authors thank the Garcia de Garcia family and farm staff at Agropecuria Cerro Coroado for their longstanding support at Espinilho Farm. We also thank Drs. Caitlin Peterson (University of California Davis), Christian Bredemeier (Federal University of Rio Grande do Sul), Jrme Bindelle (University of Lige), Leah Renwick (University of California Davis) and Reuben Mark Sulc (Ohio State University) for their valuable comments on the manuscript. We are grateful to Dr. Timothy Bowles (University of California Berkeley) for providing the R code on which we based our script for the probability analyses, and to Dr. Paulo Waquil (Federal University of Rio Grande do Sul) for helping us with the economic analysis. We acknowledge the anonymous reviewers whose comments improved the final quality of the paper. Last but not least, we thank all the M.Sc., Ph.D. and undergraduate students that worked at the long-term ICLS experiment since its establishment in 2001. This study was financed in part by the Brazilian National Research Council (CNPq) and by the Coordenao de Aperfeioamento de Pessoal de Nvel SuperiorBrasil (CAPES) Finance Code 001 through regular M.Sc. and Ph.D. scholarships and through Project Number 88881.189747/2018-01 of Programa de Doutorado-Sanduche no Exterior (PDSE-CAPES).

P.A.A.N. and A.C.M.G. conceived the original idea of this study. P.C.F.C. conceived the long-term experiment. E.A.L., P.C.F.C. and A.C.M.G. supervised the project. P.A.A.N. wrote the manuscript with input from E.A.L., P.C.F.C., M.L. and A.C.M.G. P.A.A.N. and E.A.L. analyzed the data. P.A.A.N., E.A.L., P.C.F.C. and M.L. contributed to the interpretation of results. P.A.A.N. prepared tables and figures with input from E.A.L., M.L. and A.C.M.G. P.A.A.N., W.S.F., T.R.K. and A.P.M. performed field work over years and organized the long-term dataset. All authors contributed to reviewing and revising the manuscript.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

de Albuquerque Nunes, P.A., Laca, E.A., de Faccio Carvalho, P.C. et al. Livestock integration into soybean systems improves long-term system stability and profits without compromising crop yields. Sci Rep 11, 1649 (2021). https://doi.org/10.1038/s41598-021-81270-z