Speakers Detail Risk Assessment, Benchmarking Practices.

This year's conference featured talks on the validity of the EPA'S approach to assessing environmental risks in foundries and the sustainability of benchmarking practices.

The Sunshine State welcomed foundry representatives from both the U.S. and abroad to the l2th AFS International Environmental, Health & Safety Conference held October 8-11 in Lake Buena Vista. The 201 registrants, which included an international group from 13 foreign countries, had an opportunity to visit with the 22 exhibitors who displayed environmental products and services at the event's expo.

In addition, the conference's 36 speakers gave presentations covering a range of topics from environmental regulation and best workplace operating parameters to foundry byproduct reuse opportunities and international environmental compliance efforts.

This article details two presentations from the conference that offer tools to cope with environmental risk assessment practices and food for thought on benchmarking limitations.

'Risk' Assessment

Because the Internet and EPA! OSHA reporting requirements allow data distribution on foundry operations to a foundry's neighbors at "light speed," who interprets environmental performance as well as the actual interpretation are becoming increasingly important, said Redmond Clark, The TDJ Group, in his discussion of strategies and tactics for environmental risk analysis. "The management of information and the perception of that information are key drivers in the political support base of EPA and advocacy groups. When perception management is used effectively, the managers have the ability to bring outrage to the table as a negotiating tool."

Clark discussed the concept of De Minimus risk and how it relates to environmental regulation. "De Minimis risks are those risks judged to be too small to be of social concern or too small to justify the use of risk management resources for control. The De Minimis risk level frequently used by government agencies (EPA, FDA) is 1-in-a-million increased risk of an adverse effect occurring over a 70-year lifetime in a large population," Clark said, stressing that this standard often is unrealistic when looking at the big picture.

EPA's risk assessment and management practices have targeted problems of "questionable magnitude," Clark said, adding that spending on "fixes" is not proprotionate to the scope of problems (Table 1).

The principal elements of risk assessment (directly applicable to foundry regulation) are problem identification, toxicity assessment, exposure assessment and risk characterization, according to Clark. Problem identification evaluates whether a substance causes harm, and is often driven by scientific or public concern, he said. Sources of evidence include animal studies (most common), in vitro studies, property comparisons and epidemiological studies (most reliable).

Exposure assessment estimates the amount of a substance that is inhaled, absorbed or ingested, Clark said. These studies also must consider length of exposure, frequency of exposure, the effects of lifestyle and environmental interactions

Toxicity assessment looks at the amount of a substance that can cause harm to humans, usually determined by extrapolation from animals, Clark said. This includes critical assumptions about the effects of high vs. low doses, size variations and biological differences.

Lastly, in risk characterization, the risk assessor assembles all available data and provides an estimate of risk, including an evaluation of the uncertainties surrounding the issue, Clark said. The quantified risk is compared against agency standards to determine whether or not regulation is required.

The issue with toxicity assessments, according to Clark, is that all substances are toxic in some dose, and effects may be acute or chronic, influencing all or part of the body with varying severity. "Most toxicity studies assume increasing dose equals increasing toxicity."

In addition, the assessment attempts to relate dose and response, Clark said. Chronic effects are estimated by dose-response studies on animals, carcinogenic effects are estimated by a carcino-genesis bioassay, and acute toxic effects are estimated by LD5O studies of observation of accidents.

With acute effects, most data comes from accidental poisonings and animal testing, according to Clark, who stressed that LD5O testing is used with animals, and the average lethal dose for an animal is different than for humans.

With chronic effects, non-carcinogenic assessments are based primarily on animal testing, yet small-to-large doses over a lifetime determine the lowest observed effect level and no observed effect level (NOEL). "NOEL is considered safe for the test animals, but this typically includes a two-order-of-magnitude added safety factor. This is commonly used as basis for regulation."

In a carcinogenesis bioassay, test animals are administered large doses over a lifetime, and at the end of the study, autopsies are used to find cancer evidence. These data and models estimate cancer incidence at lower exposures in humans, Clark said, adding that the available models often make assessment "guess work." Available models include threshold models, which are based on the premise that repeated exposures are required to reach a cancer threshold, and non-threshold models, that assume even one hit can cause cancer.

The EPA uses threshold models for carcinogens and non-threshold models for all non-carcinogens, Clark said. Non-linear extrapolation models are used for new data, and animal testing is the first method of choice, according to Clark, who stressed that there are problems with toxicology assessments.

Cancer assessments are based primarily on bioassays, but these bioassays are flawed because effects on animals do not equal effects on man and vice versa, and effects of large doses over short time spans are not the same as low doses over long time spans, he said.

"Risk assessments are being based on subjective judgments," Clark said. "The EPA's default assumption is that animal tumors predict human tumors. NAS identifies more than 50 subjective judgments required in a cancer assessment that are not science-based. The current administration has resisted use of RA because of value-based issues not considered, like environmental justice."

"To address lack of knowledge, we need validated models of toxicity and exposure (perhaps best developed in public works projects) and an informed-and-empowered public who can take the information provided and determine the level and balance of risk/benefit that they can endorse," Clark said.

Foundries can educate the public by:

* holding regular meetings with key local government officials;

* creating a community environmental working group with an open membership;

* offering ongoing training on risk perception and risk management issues;

* integrating state and regional environmental authorities into the effort;

* periodically reporting on environmental safety issues that are tied to emissions and release reports submitted to the government;

* training employees for external educational efforts;

* working with these groups well be fore permit application or renewal.

"One of the flaws in the current EPA risk assessment process is the focus on low-risk events," Clark said. "These cost the most to eliminate. There are a host of community-based programs that can more than eliminate any 'risks' associated with your operations. These programs cost little and generate great goodwill. We must communicate and negotiate to coexist."

Environmental Benchmarking

"The concept of a benchmark as used for performance comparison has an inherent optimism reflecting people's belief that it is always possible to do something better," said Martha McBarron, Castings Development Centre, Sheffield, England, in her discussion of environmental benchmarking. "Recent decades have seen an explosion in measurement, analysis and information about performance. Whatever can be measured will be measured by someone--whether for in-house controls, by industry associations for comparative purposes, or by governments for policy-making reasons."

While recommendations from the EU Thematic Network on Foundry Waste on best practices in Germany, Finland, Spain and the U.K. often aim to provide pointers to foundries on areas where they could improve (the most common objective being to increase profitability), "more recently, we have started talking about benchmarking for sustainability," McBarron said. A list of what Joachim Helber, Institute of Foundry Technology in Germany, considers to be key indicators for sustainability benchmarking is shown in Table 2.

Projects carried out in some European countries have involved the measurement of more than 200 indicators for benchmarking. One such research project is the Measuring Environmental Performance of Industry (MEPI) project, which involved partners from six European countries. Walter Wehrmeyer, Centre for Environmental Strategy at the Univ. of Surrey, England, presented information about an examination of 500 firms from six countries and in six diverse sectors. Environmental performance indicators (EPIs) developed by the project include:

* physical indicators that are concerned with mass and energy flows through the manufacturing process (i.e. amount of solid waste per unit of output);

* economic indicators that link physical data to data on business performance (i.e. emissions related to value-added activities);

* sustainability indicators that link environmental impacts to physical data on inputs, emissions and outputs (i.e. acidification potential per unit of output).

The MEPI team found that a company's performance could be described adequately using only about half a dozen indicators, McBarron said. "These provided the key measures of performance, and the addition of further data did not alter the assessment that could be arrived at using the critical short list of EPIs."

In addition, companies that had performance significantly worse than their counterparts always explained this by claiming that their operations were "special," rather than accepting that they were far adrift from the best practices adopted by their industry counterparts, she said.

According to McBarron, common data collection and recording problems within companies include:

* clerical and numerical errors arising during the transfer of figures (e.g., recording of meter readings, transfer of data from paper to computer, etc.);

* errors in the use and conversion of units (e.g. purchases may be reported as tons, liters, units, etc.);

* mistakes in spreadsheet formulas used for calculations;

* using untrained or insufficiently trained staff (e.g. to read meters);

* use of inaccurate or uncalibrated meters;

* company accounting systems not set up to generate the required information, thus turning a desirable benchmarking exercise into a daunting task of "mammoth proportions."

"Many errors occur on a random basis and are often easy to spot if data is subject to verification exercises, including comparison with previously collected information," McBarron said. "Systematic errors are, however, much more difficult to find, an obvious example being a mistake in a spreadsheet formula."

Recommendations from the U.K. Chemical Industries Assns. on how information quality could be improved include:

* involvement of the site accountant or other accounting staff members;

* use of written procedures for data collection, recording and assessment;

* independent checking of data;

* periodic auditing by a third party.

When collecting data from multiple companies, classic problems, according to McBarron, include:

* low response rates (typically 20-30% in the U.K.);

* profile of respondants (e.g. they may represent more proactive companies than the norm);

* data requested is either not available, or not available in the requested form;

* inconsistent data is provided (e.g. when new sand inputs are compared with waste sand sent for disposal, it would appear that foundry floor levels must be rising);

* data is reported in the wrong units, even where the units to be used are specified;

* mistakes are made when calculating or converting figures from raw company data to the amalgamated totals requested;

* questions are not always understood.

"The only conclusion one can reach about data quality is to acknowledge that errors are an inherent aspect of data collection," McBarron said. "In spite of the difficulties, it still is possible to gain confidence in data as more information is collected over a period of time and typical performance levels become well-established and better understood."

In-house benchmarking can be a powerful tool for companies seeking to improve their performance, communicate successes (and failures) to employees, customers and other stakeholders, provide a strong motivation to the workforce to better previous year's achievements, and give the company information that can be used for promotional purposes, as well as to satisfy regulators, McBarron said.

An example is Claas Guss, a German foundry that was able to include a chart of its transition to virtually 100% beneficial reuse of waste sand over a 9-year period in its 1999 Environmental Report. This report also contains a fully quantified input/output analysis, a description of the company's achievements, and future objectives and targets, she said.

Benchmarking information also can demonstrate improvements in working conditions and provide confidence to employees of a company's commitment to more than just the bottom line, McBarron said. An example is RHP Blackburn, a U.K. foundry that was scheduled for closure in 1993 due to a "poor business climate."

New Japanese owners, the NSK group, reorganized the operation and management, and today it is profitable, with a turnover of about $4.8 million, she said. The firm has achieved accreditation in quality and environmental management (ISO 14001 in 1997). Further, benchmarking information is used widely throughout the operation as a means of communication and motivation for improvement, McBarron said, stressing that measurement of factors that impact employee well-being are presented side-by-side with performance benchmarks and objectives.

In addition, the foundry has set ambitious goals for waste reduction, and reductions were achieved in solvent, gas, and water use (2%, 6% and 5%, respectively, between 1997-98), she said.

Environmental issues have heavily influenced the types of data that are collected and analyzed for benchmarking, with particular emphasis on reducing the consumption of materials, water and energy, McBarron said. "Eventually a point is reached where improvements using existing processes might no longer be possible."

"To have any hope of success in the drive towards sustainability, 21st century benchmarking will have to be broadened to include global considerations," McBarron said. "We must try to gauge the big picture and provide more sophisticated information to the governments, policymakers and stakeholders who are shaping the future."

A number of key questions to answer, according to McBarron, are:

* what are the environmental impacts of transport and distribution in the global economy, and how can the true "environmental costs" be accounted for?;

* what are the local and global environmental effects of shifting manufacturing production to developing countries? If the full effects are taken into consideration, is the shift acceptable? Is it sustainable?;

* how will future energy requirements be met without jeopardizing the global environment?;

* which foundry raw materials and processes are not sustainable and what will replace them in the future? What will happen when the eventual demise of petrochemicals makes phenol-based binder system use unfeasible?;

* how will the future society reconcile its consumer demands with the environmental impacts of production?

This article was adapted from two presentations at the 12th AFS International Environmental, Health & Safety Conference. Proceedings are available from AFS Publications at 800/537-4237 or from AFS' e-store at www.afsinc.org/estore/index.asp.