(1998) Dairy Manure Management: An Application of Probabilistic ...

2 downloads 0 Views 160KB Size Report
Andrew Johnson,* Amy Purvis Thurow, and Donald Vietor. ABSTRACT. Contemporary agricultural enterprises are having difficulty dealing with changing public ...
Published May, 1998

ENVIRONMENTALISSUES Dairy ManureManagement:An Application of Probabilistic

Risk Assessment

Andrew Johnson,* AmyPurvis Thurow, and Donald Vietor ABSTRACT Contemporary agricultural enterprises are having difficulty dealing with changing public perceptions about the environmental consequences of current management practices. This study was initiated to investigate the role Probabilistic Risk Assessment(PRA)could play in managingrisks and communicatingrisk information to regulators and the public. The utility of PRAfor dairy manureand wastewater managementsystems was evaluated on three Texas dairies beginning in spring 1992. Informationcollected fromdairymen,along with computerized risk analysis, wasused to develop risk estimates for various environmental hazards associated with the dairies. Qualitative and quantitative analysis of individual probability estimates allowedevaluation of the risks posed by various potential nutrient loss pathways. Integration of these probability estimates with simple cost-effectiveness evaluations provided a risk/cost-effectiveness value used in comparing alternative waste management technologies. Results indicate PRA, wedded with cost-effectiveness analysis, may provide a new and effective wayfor the agricultural communityto evaluate, manage, and communicaterisk information.

A

RICULTURE in America has a problem. Recently, public support for agriculture has eroded, in part as a result of perceptions that an increasingly industrialized agriculture shares responsibility for undesirable changes in rural communities, economies, and the environment (Batie, 1988). Nutrient release from concentrated animal-feeding operations is particularly contentious among issues related to environmental change. Specifically, runoff or leaching of N and P can have potentially adverse affects on water quality and human health (Gast et al., 1978; Sharpley et al., 1994). Most farmers consider nutrient concerns as an environmental problem or technological problem. Alternatively, nutrient release into the environment can be framed as a risk problem. The objective of this discussion of environmental issues is to demonstrate how framing manure management as a risk problem has potential to assist farmers in communicatingwith regulators, their neighbors, and the public. Recent changes in public perceptions are troublesome to manyfarmers since, historically, agriculture has enjoyed favored policy treatment, due to the linkages between strong local agricultural economies A. Purvis Thurow, Dep. of Agric. Economics, Texas A&MUniv., College Station, TX 77843; D.MVietor and A.F. Johnson, Dep. of Soil and Crop Sciences, Texas A&M Univ., College Station, TX 77843. Research supported by grants from the Texas Advanced Technology Program (Pr@ 999902-128), USDA-CSRSSpecial Grants Water Quality Research Program (no. 93-34214-9614), and USDA-NRICGP (no. 94-37403-0969). Received 10 Feb. 1997. *Corresponding author ([email protected]).

and economic growth. Over the last decade large agricultural enterprises are more often viewed as "corporate factory entities" rather than as "family farms" deserving special societal protection that farmers have come to expect (Fallert et al., 1994). Along with this change in the way agriculture is viewed has come a dramatic redefinition of farming as "part of the environmental problem rather than as part of the solution" (Batie, 1988). Producers and agricultural scientists now find themselves in the uncomfortable position of justifying their actions to a public they perceive as increasingly hostile and, perhaps more significantly, largely ignorant of the economicand physical realities facing a modern animal enterprise (Gerlin, 1994). An example of the current interaction between farmers and rural communities is epitomized by one farmer’s recent attempts to secure a Clean Water Act-mandated National Pollution Discharge Elimination System permit to operate a concentrated animal-feeding operation. In the spring of 1993 a California dairy farmer applied for a permit to operate a 2020 cow dairy in Erath County, Texas. His proposal met or exceeded all regulations governing protection of surface waters from discharges of livestock wastes. Nonetheless, in December 1993 the Texas Natural Resources Conservation Commission (TNRCC)denied him a permit to operate, citing concerns that manurefrom his dairy might still contaminate the nearby Paluxy River recharge zone. The local newspaper reported that "according to published reports, lagoons and runoff control structures incorporated into the application were three to four times larger than [TNRCC]standards; clay pond liners also exceed requirements" (Holan, 1993). Favorable recommendations by TNRCC’sown consulting engineers were ignored, and the permit request was denied. The executive director of the Texas Association of Dairymenarticulated the local dairy community’s viewpoint: "given these facts, it appears the TNRCC commissioners may have set a precedent that, despite clear and convincing evidence that an applicant’s environmental protection measures go above and beyond what is traditionally required, they may nonetheless deny a permit if confronted by local opposition and conjecture about the specific site" (Brister, 1993). This Texas dairy producers’ experiences in negotiating a permit to site a new concentrated animal feeding operation demonstrated that simply stating that a technology is safe, and parading out experts to confirm that assertion, is not a highly advancednor effective method Abbreviations: PRA, Probabilistic Risk Assessment; TNRCC,Texas Natural Resources Conservation Commission.

Published in J. Environ. Qual. 27:481-487 (1998).

481

482

J. ENVIRON. QUAL.,VOL.27, MAY-JUNE 1998

of risk communication. Fischoff (1995) synthesized progression of stages in risk communication starting with the elementary idea that "all we have to do is get the numbers right" and advancing through five more stages to the current state of the art, described as a summation of the previous stages plus the realization that "all we have to do is make them partners." Animal agriculture and associated nutrient managementis in the early stages of developingthis spirit of collaboration and trust-building. Initiation of risk managementand communication may be a strategy for repairing and building public support for agriculture. Furthermore, an openness to the art and science of communication and risk managementcould help the agricultural community and allied scientists to avoid reinventing 20 yr’s worth of knowledge about risk communication already available from the experiences of other industries. OBJECTIVES The intent of this EnvironmentalIssues paper is to demonstrate howan application of PRAwill help operators of concentrated animal feeding operations to managerisk and communicaterisk information to the public. Probabilistic Risk Assessmentis not a new methodologybut its application to agricultural environmentalrisk is novel. Successeswith PRA as a risk management and risk communication tool aids farmers and scientists in efforts to purposefully manageand communicaterisk information. Understandingthe inherent risks in a waste managementsystem may help farmers identify and address the highest priority areas for risk management. Moreover,applying PRAinvolves developing a visual event tree for manuremanagement that can be used to explain how the farmeris managing manureto avoid risks of spills or excess nutrient applications. PROBABILISTIC RISK ASSESSMENT Probabilistic Risk Assessmenthas been used to describe potential failures of systemsthat involvecomplex relationships and poorly structured information (Kuzminskiet al., 1995). Risk is generally defined as the probability of an undesired consequence.Morespecifically, risk is describedby the expression: Risk [consequence/time] = Frequency [events/unit time] × Magnitude [consequence/event] [1] (Henley and Kumamoto, 1992). Graphical representations the causal links in a system are typically displayed on one page, demonstratingrelationships amongnatural and design characteristics, andassociatingprobabilities withkey junctures in the system or process being described (Starr, 1985). classical PRAmayfunction effectively as a descriptive tool, due in part to the pictorial nature of event tree and fault tree analyses often used. Simpleevent trees and fault trees may be used as graphicalillustrations of the level of safety associated with a feedlot-style dairy (Johnsonet. al., 1995). Probablistic Risk Assessment and Risk Management The first step in applying PRAis efficient management of hazards and efficient allocation of resources to achieve the goal of maximizingsafety by minimizingrisks. Throughdeveloping a risk assessment,a farmer and the scientists working

with him develop a morethoroughand systemic understanding of how their manure managementsystem works which, in turn, leads to better judgmentsabout the adequacyof existing processes for management of risks. Similarly, Starr (1985) observed, with regard to PRAof nuclear reactors, that "the virtue of these risk assessments is the disclosureof the system’s casual relationships and feedback mechanisms,which might lead to technical improvements in the performanceand reliability of the nuclear stations." Given an understanding of where their manuremanagementsystems are most likely to be weak, dairy farmers require a decision tool for evaluating risk managementoptions to propose improvements. Thefusion of risk analysis with cost-effectiveness criteria mayprovide a tool to inform such on-farmresource allocation decisions. Previous workby Wilkinsonand Young(1992, unpublished manuscript, OakRidge National Laboratory) investigated the ratio of costs to risk reductionsfor various nuclear waste disposal options. Theyused risk analysis as one component in a cost-benefitanalysis, rankingthe relative desirability of several options by defining the expectednet benefit as the risk reductionper dollar invested, In this application of PRA, the relative risk reductionsare quantified for potential technology changes implementedto improvethe system. The cost effectiveness of prospective improvements is depicted by dividing these risk reductions by the expectedcost of implementation for each improvement to yield a risk/cost ratio: Risk reduction ratio = Expected risk reduction [2] Cost of implementation Dueto the site specificity of manuremanagement systems, there is good reason to question whether such an approach wouldyield results with muchtransferability betweensites. However,even if results from PRAsand economicanalyses do not allow numerical comparison of the appropriate risk management strategies for various dairy farms, they mayprovide an important decision tool for farmers concerned with howbest to allocate resources to design a cost-effective manure management systemthat minimizesrisks of off-site pollution. RISK COMMUNICATION Simplegraphical representations of the major components of the systemunder consideration and evidence of deliberate attention to risk managementcan foster communicationto the public. Positive disclosure of the details of a risk analysis contributes to public acceptanceof agricultural activities simply by building public trust that somethingis being doneabout risk. Demonstratinga good-faith effort to managerisk makes a difference becausemost conflicts about risk are not driven by disagreementabout probability estimates, but rather, "the great controversies over the dangersof technologyin our time are essentially abouttrust and distrust of societal institutions" (Wildavsky and Drake, 1989). Probabilistic Risk Assessmentmayalso aid communication of risks throughthe development and disclosure of fault trees providing visual depiction of a waste and wastewatermanagementsystem. A well-designedfault tree mayprovide a useful alternative to simplyreporting numericalresults from a risk analysis. Referringto risk communication in general, Fischoff (1995) observedthat the public "mayeven feel that they can get a better feel for the degree of risk in a process from seeing howit operates than from hearing about someesoteric numbers."Certainly calculating numberswill still be an important part of conductinga PRAof an agricultural system. Yet, displaying the numbersin an easily comprehendedfault or

483

JOHNSONET AL.: DAIRY MANURE MANAGEMENT event tree has potential to clarify communication between livestock producers and their detractors. APPLICATION In the spring of 1992, research wasinitiated on three dairies to develop event trees and fault trees describing waste management systems along with associated probabilities or frequencies of system failure. Twoof the dairies were confinement dairies in central Texas. The other dairy was a smaller forage-based dairy in eastern Texas. Interviews with the dairymen, along with monitoring of nutrients in lagoon effluent, soil, and runoff from cropland, provided information about potential pathways for nutrient loss from the dairy. The computer software package CAFTA (Science Applications International, Inc.) was used to process data and aid in the design of event trees and fault trees. Event trees were constructed linking an initiating event (waste water release from milking parlor) through forward logic to subsequent events that might combine in series to result in system failure (Fig. 1). Each of the branch headings described a major component in the waste managementsystem for which a probability of success and failure was sought. Estimated probability of failure were displayed on the lower component of each branch in the event tree. Probability of success for each branch was assumed to equal 1 - P, where P was the probability of failure for that branch. Final sequence probabilities were generated by multiplying together each of the probabilities encountered by tracing a specific pathway from the initiating event (wastewater release) to the terminal event (crop uptake or leaching). Completed event trees were shown to farmers, and their commentsand questions were incorporated into revisions of the event trees. In some cases information from farmers, onsite monitoring or the literature was not sufficient to allow the quantification of failure probabilities associated with particular events in the event tree. Here fault trees were used to aid in the estimation of event tree failure probabilities.

Initiating Event

Channel System

Lagoon1

Lagoon2

In developing fault trees, a backwardlogic was used to link a particular failure event to several previous causative events (Fig. 2). Interactions were linked together by gates employing one of numerouspossible logical operators, like "and" or "or". An "and" gate was used to link two events that must both occur to yield the associated system failure. An "or" gate was used to link events that were each capable of independently leading to the associated failure. The logic was constructed backward in time until one reached a basic event for which an acceptable probability of an occurrence could be obtained (Henley and Kumamoto,1992). Probabilities of basic events were frequently estimated through on-site evaluations or application of general principles drawnfrom the literature. Once numerical values were assigned to these risks, the relative weighting of the various gates allowed calculation of a probability for the previously unknownfailure event depicted at the top of the fault tree. Probabilistic Risk Assessment: to Risk Management

An Aid

Given completed event trees, fault trees, and associated numerical estimates of failure probabilities, the farmer reviewed the depiction of pathways for manure and wastewater management. The farmer’s understanding of causal linkages between managementdecisions and seemingly distant results improved as they reviewed and discussed the tree diagrams. After viewing an event tree, one farmer determined that the largest probability of potential nutrient loss resulted from the basic events represented in the direct runoff from irrigation (Fig. 1). Uponviewing the fault tree for nutrient runoff, concluded that limited land area under irrigation was the primary cause of this unacceptable risk (Fig. 2). The high probability of nutrient loss in runoff of wastewater indicated the numberof days per year in which rates of application exceeded infiltration capacity of the soil. Basedupon these insights, the farmer purchased enough additional irrigation equipment to double the land area receiving effluent from primary and

Irrigation

Crop Use

Sequence Probability

"1 9.81x 10 Uptake Retained Contained

0.0045

0.009 Leach

8.91"3 x 10 .3 4.47x 10

Runoff Contained

-3 2.54x 10 Uptake Contained

1.0

0.00259

Wastewater

Overflow

Retained 0.0045

0.009 Leach

Runoff

"5 2.30x 10 .5 1.16x 10 9.32"6 x 10

Uptake 0.00366 Overflow

Retained 0.0045

0.00308

0.009

"8 8.47x 10

Leach

4.25 x 104 -3 3.08x 10

Runoff

Overflow Fig. 1. Eventtree detailing probabilities of N loss froma dairy manure and wastewatermanagement system.

484

J. ENVIRON. QUAL.,VOL.27, MAY-JUNE 1998 LogicalFunctions Nutrient Release in ] RunoffDuringIrrigation 0.159

/~=

OR Gate

{g~=

AND Gate I Excess Water

I I f

HighIrrigation Volume 0.109

~

I SlopingSurface 0.80

0.199

Poor Infiltration 0.101

I

I

i

i

Fig. 2. Fault tree detailing sequenceof eventsleadingto probabilityof nutrientrunoff in dairy wastewater irrigatedonto cropland. secondary anaerobic treatment lagoons. The fault tree for this part of the system was revised to account for the expansion of irrigation onto additional acreage. A subsequent reevaluation of the risk of nutrient transport in runoff indicated a substantial reduction in probability resulted from the farmer’s investment in irrigation expansion. This farmer’s reaction to probabilistic information suggests that decisions about manure and wastewater management could be improved through use of PRA.Probabilities of failure and the amount of nutrient released per failure can be used to target components of their managementsystems that need improvement. This information allows a combined estimate of the probability of failure with its associated consequences to yield a more comprehensive estimate of risk. However, under Texas law, the differences amongconsequences of various scenarios for waste and wastewater loss are not considered. The TNRCC,in keeping with the 1972 federal Clean Water Act, has established a "no discharge policy" for large dairy producers (TNRCC,1990). Accordingly, most discharges from the boundary of a dairy are treated as permit violations, regardless of volume or nutrient concentration. No-discharge compliance requirements were taken into account during subsequent calculations of risk in the study. As a result, the consequence of a nutrient release was considered to always be the same--a permit violation. A numerical value of one was assigned to the consequence of each loss scenario. Therefore, the final probability of nutrient transport in runoff from the dairy became synonymouswith the final risk associated with a discharge. Probabilistic

Risk Assessment and Cost-Effectiveness

In conjunction with PRAon a 400 cow eastern Texas dairy, a simple example assessing and comparing the relative costeffectiveness of two risk reduction strategies was calculated. Cost-effectiveness and relative risks were assessed for two

management options to reduce nutrient runoff from landapplication of dairy effluent by irrigation: (i) to recycle flush water generated by the milking parlor, thus reducing irrigation volume, or (ii) to expand the irrigation equipment to irrigate a greater land area at a correspondingly lower irrigation rate. The first option, recycling of wastewater, involved pumping water from waste-holding ponds or lagoons into the alleys and feedlanes of the dairy barn. Otherwise, potable water had been used to flush daily manureaccumulation out of the parlor area. Recycling wastewater can reduce the total amount of water discharged into holding ponds daily, thereby lowering the quantity of wastewater that must be pumped through irrigation. The cost of installing a water-recycling system on the dairy was estimated at $45 000. This cost included estimated expenses for pumps, piping, concrete, and labor. This investment was expected to result in a 30%reduction in total volume of water discharged into lagoons. An expected life span of 15 yr was estimated for the system. Annual maintenance and labor costs were assumedto be negligible. Financing the improvementfor 15 yr at an interest rate of 15%annually would add $70 436 in interest cost, resulting in a total cost of $115 436 over the life of the system. Yearly payments of $7695 would be required to service this debt. The wastewater recycling system reduced the probability of nutrient loss in wastewater runoff (Fig. 2) through a 30% reduction in the probability of high lagoon level (Fig. 3). Reduction in probability that the lagoon must be pumpedreduced the probability of runoff of nutrients in wastewater to 1.38 × 10-1, a reduction of 2.1 × 10-2 units of probability from the original value of 1.59 × 10-1 (Fig. 3). Dividing the reduction of probability by the $7695 annual cost of implementation, the risk-reduction/cost ratio was 2.73 × 10-6 (Formula 2). Similar information was collected and analyzed for the option of expanding the area of land under irrigation. Cost estimates for this option included a 1250-foot hose-reel irrigation

485

JOHNSONET AL.: DAIRYMANURE MANAGEMENT LogicalFunctions Nutrient Releasein RunoffDuringIrrigation 0.138

I Excess Water

I SlopingSurface

0.172

0.80

I

I

Poor0.101Infiltrati°n

HighIrrigation Volume 0.079

I HumanError 0.01

I High Lagoon Level

[

I

0.07

High Soil Water Content 0.l

’1

I CompactedSoil

I

0.001

Fig. 3. Fault tree detailing probabilityof nutrientrelease in runofffollowinginstallationof a flushwaterrecyclingsystem. system that would effectively double the land available for wastewater disposal through irrigation, but would similarly double the annual labor costs for irrigation. Purchasing the hose-reel system required a one-time capital outlay of $31 000 and an estimated $1500 annual increase in labor costs over the 10-yr anticipated life of the system. Financing the $31 000 irrigation system for 10 yr at an annual interest rate of 15% resulted in an additional interest expense of $30 768 with a $6177 annual loan payment. Adding the $1500 annual increase in operating expenses resulted in an annual outlay of $7677. The doubled irrigation capacity reduced the probability of high soil water content from an estimated 1.0 × 10-1 to approximately 5.5 × 10-2 (Fig. 4). This reduction subsequently -1 lowered the probability of nutrient runoff from 1.59 × 10 to 1.27 × 10-1 -2 for a total probability reduction of 3.2 × 10 units of probability. The risk-reduction/cost ratio for this investment in risk managementwas 4.17 × 10-6 (Formula 2). A comparison of the risk-reduction/cost ratios gives the producer a criterion that integrates risk and cost effectiveness in investment and managementdecision making. In this particular example comparing two simplified options (expanding land area under irrigation and wastewater recycling), expanding irrigation afforded greater reduction in risk of nutrient runoff in wastewater than installing a flushwater recycling system. Therefore, taking into consideration the risk reduction per dollar spent leads to a different decision than does an investment choice based solely on standard cost-effectiveness comparisons. Cost-weighted risks are likely to differ significantly between farms and systems. For policy analysis of risk managementin animal agriculture, this procedure is likely to be most useful as a technique to facilitate conscious deliberation of farm level options. However,while the given example is limited to an eastern Texas dairy, the general methodology for PRAand cost-effectiveness evaluation should be sufficiently robust to permit application at sites throughout the

nation. Moreover, this approach potentially could be modified and extended to encourage informed dialogue about choices among broader policy options. Probabilistic Risk Assessment: An Aid to Risk Communication Producers helped to develop and fine-tune event trees andfault trees for their dairies. These event trees and fault trees were helpful to producers because they summarized the components and events of manure and wastewater management on one page. Furthermore, these diagrams were the focal point for communication between the authors and producer. One farmer used the event tree to communicate the design and managementof his system to a federal inspector. Even without probability estimates, the EPAinspector reportedly understood and used the event tree to identify the holding pond that was critical for wast6water retention. The inspector examined the holding pond level and promptly left the property. Whether the farmer was more pleased with the inspector’s understanding of the qualitative risk analysis or his speedy departure was not clear. Whathe clearly valued was an immediate and successful use for the descriptive information about risk that PRAhelped him to communicate. Similar applications offer potential for adapting PRAand risk communication techniques to presentations of permit applications and of technology evaluations in animal agriculture. POLICY

IMPLICATIONS

Fischoff (1995) observed that even expert analysis of risk control measures have little influence if public trust of those responsible for risk management does not accompany the scientific opinions. Obsessive pursuit of technical solutions and quantitative detail about risks, without commensurate

486

J. ENVIRON. QUAL.,VOL.27, MAY-JUNE 1998 LogicalFunctions

/~ f~=

NutrientReleasein I RunoffDuringIrrigation 0.127

= ORGate AND Gate

I

I Excess Water 0.159

~

I SlopingSurface 0.80

Fig. 4. Fault tree detailing probabilityof nutrientrelease in runofffollowingexpansion of landareaunderwastewater irrigation. efforts at communicating with nonfarmers in a way that addresses their basic concerns, may undermine the development and maintenance of trust. Fischoff (1995) noted that: "every year (or, perhaps, every day), somenewindustry or institution discovers that it, too, has a risk problem. It can, if it wishes, repeat the learning process that its predecessors have undergone. Or, it can attempt to short-circuit that process, and start with its product, namely the best available approaches to risk communication." Through the development of a PRA, a farmer demonstrates willingness to devote thought and energy to explicitly addressing and reducing risks. Furthermore, as a farmer participates in developing a PRAof their manure managementsystem, they gain greater understanding of the complex interactions involved in managingrisk effectively. Integrating cost-effectiveness information into risk-based management decisions at the farm level assists producers both in making good management decisions and in communicating their choices among options for achieving environmentally sound manurehandling. For animal agriculture specifically and production agriculture in general, PRAoffers promising techniques for assessing risks and, perhaps more importantly, a fresh way of thinking about risk communication. Fischoff observed that climbing the learning curve associated with risk communication has been a slow process for manyindustries. Animal agriculture producers, through PRA, may be able to pursue a more direct.and effective path to constructive dialogue with various stakeholders. This application of PRAdemonstrates its potential as a tool for risk communication concerning farm-level managementoptions and a foundation for informed dialogue on policy options. Improved risk communication using PRAis one option for producers and policy makers interested in promoting constructive dialogue and, as a result, more satisfactory risk managementof the environmental issues facing animal agriculture.

CONCLUSIONS The results of this project support three main conclusions. First, this initial application of PRAto an environmental conflict involving agriculture suggests that risk assessment provides a new and useful way to frame complex environmental issues. Prior uses of PRA by other industries has demonstrated utility in applying this methodology to quantifying risks associated with various hazards such as nuclear reactors and chemical plants. Similar benefits are available to agriculture from deliberate quantification of risks resulting in a better understanding of the system under consideration. Second, the integration of cost-effectiveness analysis with PRA provides a new way to evaluate technology options in a manner that integrates both the potential risk reduction and the cost associated with that reduction. Such analysis is useful since evaluating technological options based solely on comparative risk reductions may lead to decisions that are not cost effective. For the example given in this paper, selecting the technical improvement with the lowest annual cost was not the most cost-effective way to reduce the risk of the manure management system failing. This suggests that analysis of a manure and wastewater management system can become more robust when both risk and economic considerations are integrated with physical management considerations in assessing options for improving a manure management system. Third, risk assessment and risk management, are new concepts for many farmers, and PRA has potential to help them communicate effectively. Many agricultural producers see technology as the preferred solution to

JOHNSON ET AL.: DAIRY MANURE MANAGEMENT

environmental problems. In contrast, neighbors and environmental activists contend that farmers have paid insufficient attention to the ecological interface between agriculture and the environment. Many current policies frame manure management issues in terms of whether or not a farmer is using best management practices for pollution abatement or whether a performance standard is met. The PRA methodology offers an avenue for improved communication with potential to expand the policy dialogue to include more fundamental issues—in particular, whether farmers are acting on environmental compliance obligations in a manner consistent with public policies and, more broadly, public interests.

487