A framework for identifying performance targets for ...

Sustainable Materials and Technologies 1–2 (2014) 17–25

Contents lists available at ScienceDirect

Sustainable Materials and Technologies

A framework for identifying performance targets for sustainable nanomaterials Robert I. MacCuspie a,⁎, Harvey Hyman b, Chris Yakymyshyn a, Sesha S. Srinivasan b, Jaspreet Dhau a, Christina Drake a a b

College of Engineering, Florida Polytechnic University, 4700 Research Way, Lakeland, FL 33805-8531, USA College of Innovation & Technology, Florida Polytechnic University, 4700 Research Way, Lakeland, FL 33805-8531, USA

a r t i c l e

i n f o

Article history: Received 16 September 2014 Received in revised form 30 October 2014 Accepted 13 November 2014 Available online 27 November 2014

a b s t r a c t An issue in the application of nano-enabled products is how can we evaluate sustainable solutions to current system problems based on performance criteria? This work describes the application of an Input–Process–Output (IPO) model as a framework for a life-cycle analysis approach to identify performance metrics and criteria for evaluating the application of nanomaterials to improve the sustainability of a system. A case study is presented describing a scenario whereby a nano-enabled biocidal paint is considered for a remediation effort to reduce growth of dark molds and bacteria on refrigerated warehouses. The framework is applied to support identification of the energy-consuming steps (such as increased refrigeration energy burden, cleaning and repainting), selection of performance metrics for evaluating consumption, and determination of thresholds to measure sustainability outcomes. © 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction Sustainability is a term that encompasses a broad range of goals. In recent years, a focus has emerged on identifying ways of increasing sustainability [10]. Examples include studies of sustainable materials for buildings in specific cities [2], materials selection for developing sustainable products [20] or automotive applications [7]. However, quantitative metrics of sustainability often depend upon the case being considered and the goals. Examples of goals could include decreasing energy consumption, decreasing water consumption, or generating less hazardous waste, with corresponding quantitative metrics of metered kWh consumption, metered water consumption and gallons of hazardous waste, respectively. The costs and benefits of each of these metrics can vary depending upon the location of the system. For example, in California and Arizona supplies of fresh water are scarcer than in the Northeast. Similarly, the energy generation portfolio varies from state to state. If carbon emissions from electricity generation are a desired quantitative metric, this is related to metered kWh consumption. However, the correlation between those two metrics will be of relatively higher or lower concern depending upon the location within the country. For example, natural gas is used to generate the majority of electricity in Florida, while hydroelectric generation dominates Washington state [33]. The broad range of potential metrics by which to assess sustainability creates a range of decisions for researchers seeking to develop

⁎ Corresponding author. E-mail address: rmaccuspie@flpoly.org (R.I. MacCuspie).

new materials that provide sustainability solutions. Identification of useful and measurable metrics, and meaningful targets for these metrics, is a requirement for defining success. In complex systems many of these metrics can have dependencies upon one another that must be considered. Therefore, this work aims to present a framework for identifying material performance targets and metrics in the context of analyzing system sustainability performance improvements, illustrating this framework with an example of how this applies to nano-enabled sustainable materials. 2. System, lifecycle, process and process modeling For purposes of this work, a system is defined as a collection of people, products, technology and tools organized in a particular way. In this discussion, lifecycle is defined to be a representational model of stages in a process that has a beginning, defined stages in the middle, and an end — a cradle to a grave. Within a lifecycle, stages can proceed in a loop back towards beginning earlier stages, or proceed towards the end. A system can be modeled as a lifecycle, and a lifecycle can be modeled as a process. A process is defined as a collection of activities organized in such a way as to produce a result. The result is tied to a specified goal or objective of the system. The system itself is described in a graphical or narrative model intended to express its characteristics in a way that that is easily understood by others not familiar with the system. Once a goal or objective of the system is established, a method for achieving that goal or objective can be specified. The model describes the methods and applications used to produce the result (goal to be achieved), inputs

http://dx.doi.org/10.1016/j.susmat.2014.11.003 2214-9937/© 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

18

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

required, and the process for transforming inputs to outputs. In this paper a one-way direction in the process is assumed. We apply an IPO (Input–Process–Output) model to classify the stages of the process and explain the system lifecycle described in this framework. The IPO model has particular advantages that make it an attractive tool to apply to a system oriented problem set. First, IPO provides a structured approach for identifying goals and objectives of a system as outputs and how those outputs might be measured to evaluate choices in process methods. Second, the structured approach of IPO supports a gap analysis for selecting what inputs are required to achieve which outputs. In this case, the primary goal of the system is improved sustainability. Using an IPO framework in a case such as this allows decision makers, who may lack necessary technical expertise in specific process and method selections, to view the entire system as a black box and apply a lifecycle approach to assess alternative proposed treatment plans. From the lifecycle analysis point of view, where we are focused on the goal of sustainability as evaluated in terms of costs of system inputs and benefits of system outputs to assess choices of methods in the process. As a support tool, the IPO model (Fig. 1) has been in existence for quite some time, and has been used to describe both theoretical and existent systems primarily in the fields of business process modeling and data analytics. It has been used to explain characteristics in technology systems [15], measure success factors in project management [31], predict user acceptance in information systems [8], and describe team characteristics in new product development [30]. This paper describes how the IPO model can be applied as an evaluative tool for the domain of sustainability in materials and technology. The IPO model represents a system in three stages: input, process and output. Inputs are modeled as consumables and efforts that are introduced to a system at the beginning stage of the lifecycle. Outputs are modeled as the result produced by the system. Process is modeled as the conversion of the inputs to the outputs. To illustrate how this framework can be implemented, a model system case study will be presented. The scenario is as follows. A refrigerated warehouse, painted white, is susceptible to mold growth. When mold grows, the exterior color of the building darkens and thus increases the electricity consumption due to the greater cooling burden. In order to decrease cooling energy consumption, the warehouse building is pressure washed to clean the dark mold from the building. As a potential sustainability solution to reduce the frequency of pressure washing, a biocidal paint could be applied. Therefore, the following questions must be answered: 1) what is the target biocidal performance required to result in a net increase in sustainability, and 2) what metrics should be used to define and measure that increase in sustainability? In the case described in this paper, the IPO model is adapted to support identification of performance targets for sustainability metrics, using lifecycle stages identified in this use case of reducing mold growth through application of biocidal paint. This enables evaluating choices in

materials applied and implementation methods performed to achieve the goal of the system: increased performance in sustainability. Applying the IPO model to assessing the sustainability of nano-enabled products is, to the knowledge of the authors, a novel application of the IPO model approach. Fig. 2 illustrates the model process of a refrigerated warehouse. A warehouse, assumed to be initially a pristine painted white, requires a given energy consumption for cooling (Ec1). A generalized equation is shown in Fig. 2 for the variables to consider when calculating the energy consumption for cooling. Over an initial time, τ1, contamination begins to build up, requiring a greater energy consumption for cooling (Ec2). Eventually, a heavy buildup causes the greatest energy consumption (Ec3) to be reached. At this point, the warehouse can either be pressure washed or repainted. The decision to pressure wash or repaint could be made based on whether the energy consumption (Ec) is below or above a threshold (Eth). Pressure washing would remove some, but not all, of the contamination, and would require an energy consumption for washing (Ewash) and material resources (Mwash). Repainting would return the building to the initial pristine surface, and require energy consumption (Epaint) and material resources (Mpaint). Energy consumption and material resources should consider all aspects of the process of washing or repainting. For example, not only the energy to manufacture the paint and supplies used, but also if hazardous waste is generated the energy of disposing of the waste. In the case of mold growing on the warehouse exterior surface, the darkened color and higher emissivity of the mold compared to the original white surface increases the cooling burden on the warehouse. The darker color of the mold decreases the amount of light reflected compared to the pristine white surface, thereby decreasing the surface albedo value (a) and increasing Ec. Using Fig. 2 for guidance, one can model this use case using three separate processes that can be combined into a larger ecosystem lifecycle model, as summarized in Table 1. The first process is Dark Mold Growth. In this process, we model mold growth factors as the inputs, mold growth rate as the process, and the change in surface albedo over time as the output. Measureable consumables of the input could include factors such as availability of food and water sources, biocidal properties of the painted surface, and surface temperature. While the change in surface albedo is potentially a directly measurable unit of the output (or result produced), a more convenient measure is the kWh of energy consumed for cooling (Ec). The second process is Pressure Washing. In this process, the energy required to perform the washing (Ewash) and the energy required to make the materials consumed in washing (Mwash) are the measurable consumables of input. The process is modeled as pressure washing, and albedo (or Ec) as the measureable unit of output (or result produced). The third process is Repainting. In this process, Epaint and Mpaint, the energy of repainting and making materials consumed in repainting, are modeled as the measureable consumable, repainting as the process, and return of albedo to the pristine value as the measureable unit of output (or result produced).

3. Selection of variables & data analysis Once we complete the modeling of the system process, we next select measured variables (MVs) to represent the system outputs and system inputs. The corresponding outputs and inputs represent the choices in methods for the process being evaluated.

Fig. 1. The Input–Process–Output, or IPO, model.

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

19

Fig. 2. Schematic diagram of energy and material resources associated with managing a refrigerated structure such as a warehouse. Ec = energy consumption for cooling at stage 1 (pristine surface), 2 (contaminated) or 3 (heavy buildup); Eth = energy consumption threshold to determine when to paint exterior; Epaint, Ewash = energy consumption for washing or painting exterior; Mpaint, Mwash = material resources, including waste, associated with painting or washing exterior; τ = elapsed time; ΔT = temperature differential from interior to exterior of building; I = solar insolation; A = exterior surface area; a = surface albedo; r = location on the building.

In this case study, linear regression is the proposed method of system analysis. The advantage of using linear regression is that it offers a simple and easily understood means of comparing choices. We express the MVs as dependent and independent variables in linear equations and narrative statements. There are two types of equations we have used for the system model in this case. The first equation is called “main effects.” This expression describes the main effect of the independent variables (IVs) upon the dependent variable (DV) Sustainability, under the assumption that each IV is operating independently of the other IVs, meaning that the IVs do not have combined effects. Analysis of “main effects” is a good logical place to start, because we first need to determine if each IV is in fact significant, meaning is there a detectable impact produced on the system by this IV? If a particular IV is not supported as significant, then it is dropped from the model, and eliminated from consideration for inclusion in the system process evaluation. The second equation is called “full effects.” This expression describes the main effects of the IVs and also their combined effects. There are multiple orders of effects possible. Meaning, it is possible to create equations to model multiple interactions among the independent variables. In this case, to keep things simple in our explanation, we are limiting our discussion to one-way interactions between variables as an example, meaning each variable is expressed in the full effects equation independently, and as an interactive term with each other variable. Hence the reader will note that there are 51 terms in the “full effects” model listed below. This is displayed to give the reader an insight into the complexity involved when attempting to analyze interactions among variables. Ultimately, a software package such as SAS 9.2 (used by these authors previously), will calculate the significance of variables modeled with multiple orders of interactive effects. Suggested independent variables for a scenario such as the case study offered here, include air temperature, desired internal temperature (for ΔT), solar insolation, exterior surface area of the building, surface albedo, surface temperature, biocidal action of paint, humidity and rainfall (water source for mold growth factors), Ewash, Mwash, Epaint, and Mpaint. The use of the inputs–processes–outputs developed in Table 1 help guide identification of what independent variables might be selected for inclusion. The interactive effects among selected variables (as expressed in the full effects equation) reflect the reasonable assumption that some variables have relationships with, and dependencies upon, each other. For example, the surface area (A) of the building impacts Ec as previously stated, as well as Ewash, Mwash, Epaint, and Mpaint. Additionally, the total energy required for pressure washing will be proportional to the surface area required to be washed; similarly, the materials consumed will increase as surface area increases, measured in gallons of paint and converted to the energy required to make one gallon of paint. However, the more interactive effects one models among variables, the less likely one is to detect a significant effect. Hence, why the most productive analytical prescription is to test the independent effect of each variable, and if determined to be significant, then continue with testing for interactive effects, increasing the orders of effects (interactions). The model can continue to be expanded until additional terms are found to be unsupported and the model is declared complete. In the case of a linear regression model, a test statistic is used as the assessment procedure. In the described scenario, we perform a two-stage analysis: (1) Whether the selected independent variables (IVs) are significant, and (2) whether interaction exists between them. The selected IVs represent the system inputs. These are the measured consumables used to affect the output result. The dependent variable represents the measure

Table 1 List of inputs, processes, and outputs. Descriptor

Input

Process

Output

Dark mold growth

Mold growth factors (surface temp, food +, paint biocidal activity −) Energy and materials Ewash, Mwash Energy and materials Epaint, Mpaint

Mold growth rate

dadt N 0

Pressure washing

a0 N a′ N a3

Repainting

a′ = a0

Pressure washing Repainting

20

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

of Sustainability. In situations where a large set of variables exist, say 20 or so, we recommend employing an automated technique called “step-wise” to screen for significant variables for inclusion — in this work we have already selected variables identified in the previous section assumed to be significant. The models depicted below are the linear equations formed by the independent variables, representing inputs to the system, and the dependent variable, (DV) Sustainability. We have written the equations in both a full model (each independent variable and its interactive effect) and a reduced model (main effects — independent variables alone with no interaction). The resulting equations are expressed as follows: Main effects (reduced) model: DVðsustainabilityÞ ¼ B0 þ B1 X1 þ B2 X2 þ B3 X3 þ B4 X4 þ B5 X5 þ B6 X6 þ B7 X7 þ B8 X8 þ B9 X9 þ B10 X10 þ e: Full model: DVðsustainabilityÞ ¼ B0 þ B1 X1 þ B2 X2 þ B3 X3 þ B4 X4 þ B5 X5 þ B6 X6 þ B7 X7 þ B8 X8 þ B9 X9 þ B10 X10 B11 X1 X2 þ B12 X1 X3 þ B13 X1 X4 þ B14 X1 X5 þ B13 X1 X6 þ B14 X1 X7 þ B15 X1 X8 þ B16 X1 X9 þ B17 X1 X10 þ B18 X2 X3 þ B19 X2 X4 þ B20 X2 X5 þ B21 X2 X6 þ B22 X2 X7 þ B23 X2 X8 þ B24 X2 X9 þ B25 X2 X10 þ B26 X3 X4 þ B26 X3 X5 þ B27 X23 X6 þ B28 X3 X7 þ B29 X3 X8 þ B30 X3 X9 þ B31 X3 X10 þ B32 X4 X5 þ B33 X4 X6 þ B33 X4 X7 þ B34 X4 X8 þ B35 X5 X9 þ B36 X6 X10 þ B37 X5 X6 þ B38 X5 X7 þ B39 X5 X8 þ B40 X5 X9 þ B41 X5 X10 þ B42 X6 X7 þ B43 X6 X8 þ B44 X6 X9 þ B45 X6 X10 þ B46 X7 X8 þ B47 X7 X9 þ B48 X7 X10 þ B49 X8 X9 þ B50 X8 X10 þ B51 X9 X10 þ e where: • • • • • • • • • •

Air temp: X1 = collected hourly Internal Temp: X2 = target cooling temperature kWh: X3 = hourly cooling energy consumption Mold growth rate: X4 = daily average Humidity: X5 = daily average Pressure washing water consumption: X6 = Liters water used, yearly average Mpaint: X7 = kWh used to manufacture consumables used in repainting Epaint: X8 = kWh used repainting Ewash: X9 = kWh used pressure-washing Mwash: X10 = kWh used to manufacture consumables used in pressure-washing.

The “main effects” or reduced model, expresses the effect of each of the 10 independent variables listed above upon the dependent variable sustainability. The full model expresses both the main effect of each of the 10 independent variables and the interactive effect between them. The coefficient (B) represents the magnitude of the effect for each independent variable upon sustainability. This gives us an estimation for how much impact a single unit of a selected variable will have upon sustainability. The error term within each equation is represented by (e). At this point we need to mention that linear regression relies on the assumptions that the data is normal distributed, that the errors are independent and evenly distributed about the mean, and homoscedasticity (variance about the regression line is the same for all X values, also called homogeneity of variance). It is important to note that modeling the variables to reflect more realistic distribution of probabilities can lead to more accurate forecasts and predictions [14,16]. However, for purposes of clarity in communicating this framework, we are proceeding with the above assumptions and recommend the use of probability distributions when practical. The coefficient (B) leads the user to identifying which variables will have the greatest effect on sustainability. Consequently, those independent variables and those interactive effects between independent variables which have higher coefficients will be the performance targets that have the greatest impact on sustainability. The order of magnitude provides a qualitative identification of the performance metrics, and depending upon the error of the analysis can provide semi-quantitative to quantitative performance targets. To simplify demonstration of how to arrive at target performance metrics, assumptions can be made using previously published data to generate the data for our case study described. In a previous report, the mean heated area of a sample of 221 warehouses in Florida was stated to be 30,114 ft2 [22]. Assuming an average warehouse height of 25 ft, and a 2:1 building aspect ratio, the warehouse would be 122.7 ft by 245.4 ft, and have a total surface area (including the roof and 4 walls) of 48,519 ft2. Strategy guidelines to calculate an accurate heating and cooling load were recently reported [5]. As shown in Fig. 3, the cooling loads is the sum of external load and internal load, or stated differently as the sensible load and latent load [1,4]. The cooling load distribution is represented in Fig. 4, which reveals that the majority of load contribution (47%) arises from the enclosure, which comprises of roof, walls, windows and ventilation systems. The equipment like refrigeration units and light fixtures contribute to 41% of the cooling load. Approximately 12% of the load originates from occupancy of humans. In considering all of the detailed information provided above, the contribution of dark mold growth on the exterior painted surfaces (walls and roof) of a refrigerated warehouse are estimated to be approximately 12% of the overall cooling burden on the building. Considering the IPO model and descriptors in Table 1, the consequence of each of these models on Ec is illustrated in Fig. 5 using hypothetical data to compare a normal and biocidal paint. In Fig. 5a, a set of assumptions called Set A considers only the following: • The impact of biocidal paint on growth rate • Pressure washing will not return the surface albedo to the original value. • Ratio of Ec′ to Ec0 is arbitrarily given a boundary condition of 3 before pressure washing occurs.

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

21

Fig. 3. Illustration of the many potential sources of heating and cooling burdens upon a refrigerated warehouse. This work focuses only on the increased burden due to conduction through opaque surfaces, as a consequence of dark mold growth on the exterior of the warehouse. Adapted from [1].

In Fig. 5b, a slightly more sophisticated model is plotted. The added assumptions of Set B include the following: • The mold growth rate will be faster after each round of pressure washing • The rate of increase in mold growth rate for biocidal paint will be half the rate of increase for normal paint • Repainting will occur after 6 rounds of pressure washing. In Fig. 5c, the assumptions are further refined. In this work it is assumed that the surface albedo (a) increases as dark mold growth increases, and will double the walls and roof 12% component of the overall cooling burden. It is therefore important to note that the initial arbitrary boundary condition on Ec′/Ec0 of 3 is grossly overestimated. If one assumes a boundary condition of doubling the burden of only the 12% exterior building envelope contribution, a maximum Ec′/Ec0 ratio of 1.12 would be reached. In reality, due to the unsightly aesthetics of such a situation, it is very unlikely that the 1.12 boundary condition would ever be approached. A more realistic assumption might be tolerating no more than 15% coverage of dark mold

Fig. 4. Cooling load distribution, based on reported data [1].

22

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

a

Assumption Set A 3

Ec' / Ec0

2.5 2 1.5 1 0.5 0 Time Normal Paint

b

Biocidal Paint

Assumption Set B 3

Ec' / Ec0

2.5 2 1.5 1 0.5 0 Time Normal Paint-Faster Growth Each Wash; Repaint after 6 Biocidal Paint- Faster Growth Each Wash; Repaint after 6

c

Assumption Set C 3

Ec' / Ec0

2.5 2 1.5 1 0.5 0 Time Normal Paint

Biocidal Paint

Fig. 5. a. Hypothetical energy of cooling increase and reduction over time, as related to dark mold growth on exterior painted surfaces. Pressure washing events cause a reduction in cooling energy demand, with biocidal paint application causing decreased demand for pressure washing. b. Hypothetical energy of cooling increase, including assumptions that time between pressure washings decreases each round until a repainting threshold is reached. c. Revision of assumptions made for the boundary condition on Ec′/Ec0 before pressure washing occurs. Both panels are the same raw data, scaled differently on the y-axis, to illustrate how both assumptions and presentation of the data are important for drawing meaningful conclusions on the true impact upon sustainability. (c) intends to show the total impact on cooling energy burden and is scaled from 0 to 1.02 to visualize the integrated energy consumption more readily.

growth before pressure washing. This would mean that the boundary condition of Ec′/Ec0 before repainting would become 1.018, rather than the value of 3 assumed in Fig. 5a and b. Therefore, Fig. 5c is presented with these assumptions: • • • • •

The impact of biocidal paint on growth rate Pressure washing will not return the surface albedo to the original value. Esthetics impact repainting rate Only 12% of cooling burden is from building envelope Revised boundary condition of a maximum Ec′/Ec0 ratio of 1.018.

Based on the previously cited input data and assumptions, we have estimated the rough order of magnitude (ROM) of heating and cooling load requirements for an Orlando, Florida based warehouse premises. For a 48,515 ft2 warehouse surface area, the heating load is approximately 515,272.00 Btu/h (using a heating load requirement of ~ 10.62 Btu/h per square foot [5]), which is equivalent to 150 kWh. The cooling load

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

23

is calculated by adding both the sensible load of about 7.5 Btu/h and latent load of 2 Btu/h, for a square foot area [5], and hence we obtained ~ 451,759 Btu/h for a 48,515 ft2 surface area of the warehouse. This in turn is comparable to about 132 kWh energy consumption needed to cool the warehouse. If we assume a 15% increase in cooling burden from the 12% exterior surfaces contribution, then the cooling burden would increase to 134.4 kWh. If dark mold growth is mitigated by biocidal nanoparticle formulated painting on the exterior walls, then this would mitigate the 2.4 kWh energy consumption increase per square foot. Another sustainability metric proposed is the water consumption during pressure washing. This is related to the pressure washing water consumption (X6) by multiplying the water consumption per pressure wash (Wpw) by the annual frequency of pressure washing. Water consumption (Wpw, L) could be estimated as proportional to the exterior surface area of the building (A, ft2), pressure washer flow rate (PWflow, L/min), and pressure washer cleaning rate (PWrate, ft2/min), as shown in Eq. (2). Using an example pressure washer that uses 15.4 L/min, an assumed working rate of cleaning 50 ft2/min, and the model warehouse described would provide an estimated average pressure washing water consumption of approximately 15,000 L. Using the hypothetical data of Fig. 5b, there would be 14 pressure washings for normal paint compared to 6 pressure washings for the biocidal paint. In this hypothetical scenario, the maximum water conservation potential (Wpw max) is to save 120,000 L of water per biocidal paint repainting cycle. WPW ¼

A  PWflow PWrate

ð2Þ

Therefore, using these generalized assumptions as an illustrative hypothetical example, a performance target of 2.4 kWh of energy consumption versus 120,000 L of water consumption can be established, given a biocidal performance target of decreasing the mold growth rate by approximately a factor of 5. 4. Comparison of biocidal paint to normal paint Nanotechnology-based additives could potentially offer biocidal performance. For example, silver nanoparticles have been shown to have antimicrobial properties primarily through the release of silver ions and generation of reactive oxygen species [17,29]. Additionally, the potential exists for photoactive forms of nanoscale TiO2 to be used to generate reactive oxygen species [19,26], as does the option of adding biocidal organic molecules to the paint formulation. When considering whether nanotechnologies will offer improvements to sustainability, comparisons to all available approaches must be considered. This work will focus on adding silver nanoparticles (AgNPs) to the paint. When considering the life cycle of a AgNP paint, the potential for release of AgNPs into the environment must be considered [6,24]. There is a growing body of literature describing the potential cycle of fate and transformations of AgNPs [3,11,12,21]. For example, it has recently been shown that AgNP dissolution can occur under acidic environments [23,28]. These sustainability costs for energy and water due to environmental health and safety considerations can be grouped into the terms, EnanoEHS and WnanoEHS, respectively Paint formulations generally fall into two types: latex and oil-based. In each formulation, there is typically 10–20% extenders and additives, as illustrated in Fig. 6 [13]. For this work, the assumption will be made that one could use a 0.1% loading of silver nanoparticles to achieve the desired biocidal properties, while maintaining other desired properties of the paint. It is also important to consider the surface coatings on the nanoparticles, and make careful measurements on the surfaces of the nanoparticles [25,32], which may impact their ability to maintain desired biocidal properties. The energy and water consumed to manufacture, transport, and include the nanomaterials into the paint formulation will vary depending upon precursor nanomaterial manufacturing process selected, and the final viscosity of the paint; however, the calculation of these totals will ultimately be used to compare against the

sustainability performance target to evaluate if the solution increases overall sustainability, relating to the variable X7, through measurement of the gallons of paint used and multiplying by the marginal energy costs. For example, it is reported in literature synthesis approaches to achieve AgNPs on the order of mg/L, which are then purified and concentrated up to mg/mL [9,27,35]. The resulting waste stream from this concentration and purification must often be treated as hazardous waste, due to the risk of it containing traces of dissolved silver or AgNPs. This hazardous waste stream disposal creates a tremendous energy consumption burden on the process of adding AgNPs to a nano-enabled biocidal paint, with a subsequent impact upon the Mpaint and X7 variables. With paint performance in mind, a look at the paint formulations would indicate that the Oil Based formulations may be better candidates for biocidal enhancement assuming a 0.1% silver nanoparticle loading, based on the larger percentage of extenders and additives to the formulation. This is quite reasonable, given a recent report that a Minimum Inhibitory Concentration (MIC) of less than 3 ppm of colloidal silver nanoparticles was found for 14 bacteria as well as 5 fungal species [18], implying that perhaps adequate biocidal properties could be achieved at far less than 0.1% AgNP loading. This performance target would enable materials researchers to examine if new formulations with biocidal effectiveness would be feasible without sacrificing base paint performance. Based on the simple model proposed for the framework of evaluating nano-enabled biocidal paint for refrigerated warehousing, different complex tradeoffs can begin to be evaluated. For example, if the goal is to decrease overall cooling burden (energy) and/or number of pressure washings (energy and water), then by the assumptions made in this model, it would appear that the use of biocidal paint is a resource saving technology. However, the assumption that the repaint times are common between biocidal and non-biocidal paint should be considered. This in itself does not make the use of biocidal paints moot, in that even if its biocidal activity approached zero more quickly than the repaint

Fig. 6. Typical paint formulations based on data reported for latex and oil-based paint types [13].

24

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

time of normal paint, it is possible that the framework then returns to the non-biocidal paint model. Therefore, the potential for longer times between repaints may exist, making up for the gains in normal cooling and pressure washing energy and water costs when the paint loses its activity. Over the entire lifetime of the building, this would have the potential to lead to less energy and water consumed, as well as less paint waste from the repaintings. When considering this, it is recommended that a study be conducted upon the release rates of active ingredients, and the decrease in the activity of the biocidal coating over time as active ingredients become exhausted. This type of quantitative measurement can refine the numerical relationship between measured variables used in the IPO model. A second point is the tradeoff of warehouse energy and water consumption to hazardous waste creation. While the overall benefits confined to the warehouse scenario may be positive, a look at the production of the AgNPs for incorporation into biocidal paint as a well as the warehouse scenario would give a more holistic view of the total environmental benefit on the use of biocidal paints. It may be that the benefit of biocidal paint in the case of refrigerated warehousing does not outweigh the creation of hazardous waste in the large scale production of AgNPs for the biocidal paints. It is worth noting that the United States Environmental Protection Agency has regulatory authority through the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) for its Office of Pesticide Programs to request a review of all new pesticide products [34]. This authority would include claims of killing mold or bacteria, including antimicrobial claims made by silver nanoparticle products. Regulatory uncertainty existed for whether silver nanoparticles would be treated as a new pesticide product until 2011, when the first application for a nanosilver pesticide product was submitted to EPA. 5. Recommendations In summary, we have proposed the use of the IPO model to develop a framework to support accurate reflection of the realities of a system process under evaluation. A system model such as IPO focuses on process oriented questions and how far-reaching the scope of work will be for a given analysis. The most robust analyses will necessarily attempt to estimate the quantitative relationships between the measurable variables in the system. If inevitably assumptions need to be made, it is important to clearly state those assumptions for later refinement of the models. For example, in this work assumptions about warehouse size, cooling burden from building envelope, and pressure washer water consumption rates all impact tradeoff relationships. Use of linear regression on measurable variables supports the instantiation of the IPO model to evaluate quantitative impacts upon the final desired metrics, such as energy consumption or water consumption. • Identify the question being asked, and focus the scope of work • Develop an IPO model of the systems and processes involved • Identify measurable independent variables that relate to the inputs and outputs • Identify whether relationships and dependencies exist between independent variables • Use linear regression analysis 6. Conclusions A framework for identifying performance targets for sustainability of new nanomaterials is presented. The framework uses an Input– Process–Output model to identify factors that will have the greatest weight on shifting the sustainability of a solution, providing qualitative insight into what the areas of greatest impact could be. Through making assumptions based on the specific use case or through generalized assumptions based on average data available, the determination of specific performance targets becomes a tractable calculation. The identification of material property performance targets was illustrated for

the case presented here of biocidal activity for nanomaterial additives to paint where biocidal performance must improve by a given proportion over the water and energy consumption required to make a nano-enabled biocidal paint formation, based on assumptions or actual data for the specific case under study. This framework can enable both researchers to set goals when developing next-generation materials, and help managers making data-driven decisions. Acknowledgments The authors acknowledge support from Florida Polytechnic University to invest their time in creating this manuscript. References [1] A. Bhatia, HVAC made easy: a guide to heating & cooling load estimation. Retrieved from http://www.pdhonline.org/courses/m196/m196content.pdf, 2012. [2] U.G.Y. Abeysundara, S. Babel, S. Gheewala, A matrix in life cycle perspective for selecting sustainable materials for buildings in Sri Lanka, Build. Environ. 44 (5) (2009) 997–1004, http://dx.doi.org/10.1016/j.buildenv.2008.07.005. [3] Akaighe, MacCuspie, Navarro, Aga, Banerjee, Sohn, Sharma, Humic acid-induced silver nanoparticle formation under environmentally relevant conditions, Environ. Sci. Technol. 45 (9) (2011) 3895–3901, http://dx.doi.org/10.1021/es103946g. [4] ASHRAE, Chapter 18: nonresidential cooling and heating load calculations, ASHRAE Handbook: Fundamentals, ASHRAE, 2009. [5] A. Burdick, Strategy guideline: accurate heating and cooling load calculations, US DOE, 2011. (doi://www.nrel.gov/docs/fy11osti/51603.pdf). [6] Costanza, El Badawy, Tolaymat, Comment on “120 years of nanosilver history: implications for policy makers”, Environ. Sci. Technol. 45 (17) (2011) 7591–7592, http://dx.doi.org/10.1021/es200666n. [7] Cunha, Campos, Cristovão, Vila, Santos, Parajó, Sustainable materials in automotive applications, Plast. Rubber Compos. 35 (6–7) (2006) 233–241, http://dx.doi.org/10. 1179/174328906X146487. [8] F.D. Davis, Perceived usefulness, perceived ease of use, and user acceptance of information technology, MIS Q. 13 (3) (1989) 319–340, http://dx.doi.org/10.2307/ 249008. [9] Fauss, MacCuspie, Oyanedel-Craver, Smith, Swami, Disinfection action of electrostatic versus steric-stabilized silver nanoparticles on E. coli under different water chemistries, Colloids Surf. B: Biointerfaces 113 (2014) 77–84, http://dx.doi.org/10. 1016/j.colsurfb.2013.08.027. [10] J. Fiksel, A framework for sustainable materials management, JOM 58 (8) (2006) 15–22, http://dx.doi.org/10.1007/s11837-006-0047-3. [11] Gorham, MacCuspie, Klein, Fairbrother, Holbrook, UV-induced photochemical transformations of citrate-capped silver nanoparticle suspensions, J. Nanoparticle Res. 14 (10) (2012), http://dx.doi.org/10.1007/s11051-012-1139-3. [12] Gorham, Rohlfing, Lippa, MacCuspie, Hemmati, David Holbrook, Storage wars: how citrate-capped silver nanoparticle suspensions are affected by not-so-trivial decisions, J. Nanoparticle Res. 16 (4) (2014), http://dx.doi.org/10.1007/s11051014-2339-9. [13] T. Greiner, V. Veleva, A. Phipps, Paint product stewardship: a background report for the national dialogue on paint product stewardship, University of Massachusetts/ Lowell, Lowell, MA, 2004. [14] Hristozov, Zabeo, Foran, Isigonis, Critto, Marcomini, Linkov, A weight of evidence approach for hazard screening of engineered nanomaterials, Nanotoxicology 8 (1) (2014) 72–87, http://dx.doi.org/10.3109/17435390.2012.750695. [15] H. Hyman, Systems acquisition, integration and implementation for engineers and IT professionals, First edition ed. Sentia Publishing, Sentia, TX, 2014. [16] Keisler, Collier, Chu, Sinatra, Linkov, Value of information analysis: the state of application, Environ. Syst. Decis. 34 (1) (2013) 3–23, http://dx.doi.org/10.1007/s10669013-9439-4. [17] J.S. Kim, E. Kuk, K.N. Yu, J. Kim, S.J. Park, H.J. Lee, . . ., M. Cho, Antimicrobial effects of silver nanoparticles, Nanomedicine 3 (1) (2007) 95–101, http://dx.doi.org/10.1016/ j.nano.2006.12.001. [18] H. Laroo, Colloidal nano silver-its production method, properties, standards and its bio-efficacy as an inorganic antibiotic, J. Phys. Chem. Biophys. (2013), http://dx. doi.org/10.4172/2161-0398.1000130. [19] Li, Srinivasan, Kislov, Schmidt, Stefanakos, Goswami, Enhancement of TiO2 photocatalytic activity by N-doping using the gas phase impregnation method, MRS Proceedings, 2011, p. 1217, http://dx.doi.org/10.1557/PROC-1217-Y03-35. [20] L.Y. Ljungberg, Materials selection and design for development of sustainable products, Mater. Des. 28 (2) (2007) 466–479, http://dx.doi.org/10.1016/j.matdes. 2005.09.006. [21] R.I. MacCuspie, Chapter 4: characterization of nanomaterials for nanoEHS studies, in: M. Hull, D. Bowman (Eds.), Nanotechnology Environmental Health and Safety, Second Edition: Risks, Regulation, and Management (Micro and Nano Technologies), 2nd editionWilliam Andrew, 2014. [22] M.A. Morales, J.P. Heaney, K.R. Friedman, J.M. Martin, http://www.conserve floridawater.org/publications/JAWWA2011_CIIWaterUse_M%20Morales.pdf J. Am. Water Works Assoc. 103 (6) (2011) 84–96. [23] S.K. Mwilu, A.M. El Badawy, K. Bradham, C. Nelson, D. Thomas, K.G. Scheckel, . . ., K.R. Rogers, Changes in silver nanoparticles exposed to human synthetic stomach fluid:

R.I. MacCuspie et al. / Sustainable Materials and Technologies 1–2 (2014) 17–25

[24] [25]

[26]

[27]

[28]

[29]

effects of particle size and surface chemistry, Sci. Total Environ. 447 (2013) 90–98, http://dx.doi.org/10.1016/j.scitotenv.2012.12.036. Nowack, Krug, Height, 120 years of nanosilver history: implications for policy makers, Environ. Sci. Technol. 45 (4) (2011) 1177–1183, http://dx.doi.org/10.1021/es103316q. Pease, Tsai, Zangmeister, Zachariah, Tarlov, Quantifying the surface coverage of conjugate molecules on functionalized nanoparticles, J. Phys. Chem. C 111 (46) (2007) 17155–17157. V. Renugopalakrishnan, A.M. Kannan, S. Srinivasan, V. Thavasi, S. Ramakrishna, P. Li, . . ., G. Audette, Chapter 5 — nanomaterials for energy conversion applications, in: H.S. Nalwa (Ed.), Nanomaterials for energy storage applications, 2009, pp. 155–178. Roberts, McKinney, Kan, Krajnak, Frazer, Thomas, . . ., Castranova, Pulmonary and cardiovascular responses of rats to inhalation of silver nanoparticles, J. Toxic. Environ. Health A 76 (11) (2013) 651–668, http://dx.doi.org/10.1080/15287394. 2013.792024. K.R. Rogers, K. Bradham, T. Tolaymat, D.J. Thomas, T. Hartmann, L. Ma, A. Williams, Alterations in physical state of silver nanoparticles exposed to synthetic human stomach fluid, Sci. Total Environ. 420 (2012) 334–339, http://dx.doi.org/10.1016/j. scitotenv.2012.01.044. V.K. Sharma, R.A. Yngard, Y. Lin, Silver nanoparticles: green synthesis and their antimicrobial activities, Adv. Colloid Interf. Sci. 145 (1–2) (2009) 83–96, http://dx.doi. org/10.1016/j.cis.2008.09.002.

25

[30] R.M. Stock, How should customers be integrated for effective interorganizational NPD teams? an Input–Process–Output perspective, J. Prod. Innov. Manag. 31 (3) (2014) 535–551, http://dx.doi.org/10.1111/jpim.12112. [31] A. Subiyakto, A.R. Ahlan, Implementation of input–process–output model for measuring information system project success, TELKOMNIKA Indones. J. Electr. Eng. 12 (7) (2014), http://dx.doi.org/10.11591/telkomnika.v12i7.5699. [32] Tsai, DelRio, Keene, Tyner, MacCuspie, Cho, . . ., Hackley, Adsorption and conformation of serum albumin protein on gold nanoparticles investigated using dimensional measurements and in situ spectroscopic methods, Langmuir 27 (6) (2011) 2464–2477, http://dx.doi.org/10.1021/la104124d. [33] US EIA, U.S. energy information administration — EIA — independent statistics and analysisRetrieved from http://www.eia.gov/state/?sid=WA#tabs-42014. [34] US EPA, Federal insecticide, fungicide, and rodenticide act (FIFRA) | agriculture | US EPA, 2014. [35] Zook, Long, Cleveland, Geronimo, MacCuspie, Measuring silver nanoparticle dissolution in complex biological and environmental matrices using UV–visible absorbance, Anal. Bioanal. Chem. 401 (6) (2011) 1993–2002, http://dx.doi.org/10.1007/ s00216-011-5266-y.