LIST OF PUBLICATIONS
William G.B.PHILLIPS,BSc(Eng),BSc(Econ),ARSM,AMIMM
Mineral Economics
[1] LEAD AND ZINC, Australian Mineral Industry 1968 Review, Bureau of Mineral Resources, Geology & Geophysics, Canberra,1969, 26pp.
[2] LEAD AND ZINC, Australian Mineral Industry 1969 Review, Bureau of Mineral Resources, Geology & Geophysics, Canberra,1970, 29pp.
Accounts of the Australian lead-zinc mining and refining industry in 1968 and 1969. The information was gathered by visits to the mines and refineries and discussions with the State Departments of Mines. The statistical material was provided by the Bureau of Census & Statistics. Information on prices and overseas developments was drawn from various UK, US and UN sources
[3] THE CONSUMPTION OF ZINC; AN ECONOMETRIC ANALYSIS OF THE US MARKET FOR SLAB ZINC, Australian Mineral Industry Quarterly Review, Bureau of Mineral Resources, Geology & Geophysics, Canberra, 1968, 10pp.
Demonstrates a method for pooling cross-sectional and time-series data in order to obtain better estimates of the income and price elasticities of demand for zinc. The study suggests that with prices and incomes constant the effect of technological change is to reduce the demand for zinc by about 2% per annum. This is thought to be due to the development of plastics and aluminium as substitutes.
[4] ASSESSMENT OF AUSTRALIAN TIN RESOURCES, Australian Mineral Industry Quarterly Review, vol 25, no 4, Bureau of Mineral Resources, Geology & Geophysics, Canberra, 1973, 10pp.
Uses the classification proposed by Blondel & Lasky (and accepted by the Society of Economic Geologists) to make an inventory of Australian reserves and resources of tin. This classification has two dimensions (i) certainty about the grade and location of the deposit and (ii) cost of extraction. In this framework reserves are deposits which are accurately measured and can be worked at a profit. If Australian production grew at 1.5% p.a. the reserves known in 1973 would be exhausted in about 27 years. However further exploration and changes in prices and costs would make it possible to reclassify additional tranches of resources as reserves so in practice the supply of tin would not run out in the foreseeable future.
[5] A STUDY IN ZINC PRICES 1890-1970, Proceedings of the Seminar on the Application of Statistical Methods in Mining & Mineral Exploration, University of Adelaide, South Australia, 1971, 13pp.
Demonstrates the application of several methods of time series analysis to the annual average price of zinc between 1890 and 1970. If the figures are adjusted to compensate for inflation the resulting time series can be shown to be stationary with a constant mean and variance. The application of the auto-regressive linear model suggests that the data contains a oscillation with a period of 4.28 years. Fourier analysis detects a component with a frequency of about 0.25 cycles per annum.
[6] EXTRAPOLATION OF ECONOMIC TIME SERIES, Internal discussion paper, Bureau of Mineral Resources, Geology & Geophysics, Canberra, 1972, 29pp.
Economic time-series data may be thought of as the output of a filter. Since the current state of the economic system depends on events that took place in the past this filter must have a causal transfer function. The imaginary part of a causal transfer function is the Hilbert transform of the real part. The autocorrelation function of a filter may be found by calculating the auto-regression parameters of the time-series. The parameters of the autocorrelation function may be interpreted as the coefficients of a polynomial and this polynomial can be factorised to find its real and imaginary roots. The parameters of the causal transfer function are found by selecting the complex roots of the polynomial that lie outside the unit circle. This procedure is equivalent to factorising the power spectral density function into a complex conjugate pair. The causal transfer function can be used to extrapolate an economic time-series by finding the serial product of the input and a chopped version of the filter coefficients. The filter can be used to generate random walks having the same statistical properties as the observed time-series.
[7] STATISTICAL METHODS FOR ESTIMATING LATENT MINERAL RESOURCES, Internal discussion paper, Bureau of Mineral Resources, Geology & Geophysics, Canberra, 1972, 9pp.
Geostatistics uses fragmentary knowledge about the distribution of minerals in the EarthÕs crust to make estimates of the quantities available between given grade intervals. For example, in the Tasmanian tin fields the background mineralisation of the tin-granite is known and so is the amount of metallic tin removed from the associated tin mines since operations commenced. On the assumption that the mineral is distributed log-normally, these two pieces of information can be used to estimate the parameters of the parent distribution. A relationship can then be derived between grade and reserves showing how much additional mineral becomes available as exploitable reserves as the grade declines. Similar methods could be used to estimate global resources of other minerals.
[8] MONTE CARLO TESTS OF CONCLUSION ROBUSTNESS, (Jorgen Randers & Leif K. Ervick eds.) Proceedings of the 1976 International Conference on System Dynamics, Geilo, Norway 1976.
Describes the application of Monte Carlo techniques to testing a large simulation model of the world economy. Uncertain data is encoded as probability densities and samples drawn from these densities are used as parameters for the model runs.The model output is analysed statistically and the contribution made to total uncertainty by the variance of the inputs is compared. The outputs are also analysed by multiple regression methods and it is demonstrated that variances are additive over a limited range. Due to the negative feedback loops in the model, variation in the inputs is strongly attenuated.
[9] METAL PRICES AS A FUNCTION OF ORE GRADE, (with D.P. Edwards) Resources Policy, Sept 1976, 11pp.
Reviews the evidence for resource depletion and shows that, in the case of copper in the United States, mining is now largely concentrated on large low grade disseminated ore bodies worked by open cast methods. Technological progress has largely compensated for increased costs and consequently the real price has remained almost static, on average, since 1890. A model relating the price of metals to physical parameters can explain most of the variation between the prices. The variables chosen are a representative ore grade for the metal (which is related to mining costs) and the energy required to refine the metal (which is related to the refining costs). The estimated relationship predicts 10 metal prices within the band ± 1.5 decibels and a further 7 within the band ± 3.0 decibels. The relationship holds over 5 orders of magnitude from mild steel at 14 cents per kg to platinum at $3,782 per kg. The poor prediction of the outliers (cadmium, cobalt and antimony) have convincing explanations. Since price is correlated with grade the relationship indicates how prices might change should higher grade resources be depleted.
[10] SARUM 76 GLOBAL MODELLING PROJECT, (Joint author & editor), Department of the Environment, Research Report No.19, HMSO, London 1977, 223 pp.
SARUM 76 is an attempt to build an improved model of the world economic system and its use of natural resources, and to develop new tools that will be helpful to others working in this field.The nations of the world are divided into roughly homogeneous groups whose composition is determined by the purposes of the study. Various industrial and agricultural activities are recognised within each group. It is assumed that agencies within the system behave according to standard economic behaviour patterns. For example entrepreneurs maximise profits, investors maximise returns, consumption increases as per capita income rises and falls if prices go higher. The relationships built into the model are based as far as possible on published empirical data. The adaptation of the economic system to changed circumstances is represented mathematically as a set of differential equations. The model output is the solution of these equations showing how the world economic system might be expected to develop over time (given that the data and the model relationships are valid). The model can be used to test a wide range of scenarios and policy options. Agricultural problems have been given priority and the interaction between rising population and the limits imposed by the availability of irrigation water and new land has been examined. The model demonstrates the effects of these constraints on standards of living and the fraction of income spent on different goods and services. The model also simulates changes in production methods as the relative costs of labour, capital and other inputs vary. Some preliminary experiments have been made to assess the effect that increased energy costs might have on the cost and output of fertiliser and therefore of food prices. Ancillary studies studies have been carried out to enable changes in income distribution and trading relationships to be handled. Problems arising from uncertainty about the realism of the modelÕs structure and its dynamic behaviour are discussed.
[11] STATISTICAL ESTIMATION OF GLOBAL MINERAL RESOURCES, Resources Policy, Dec 1977, 13pp.
The concentration of minerals in the crust is distributed continuously between the high grade ore reserves that are mined for industrial use and the low grade background mineralisation of the common rocks.The historical record of the U.S. copper mining industry and geochemical data on crustal abundance are consistent with the hypothesis that this distribution takes the log-normal form. A log-normal distribution could be generated by a mineralisation process which is compatible with plate tectonics theory. This suggests that the crust is mineralised by solutions generated in subduction zones. If these ore bodies are subsequently lifted above sea-level by tectonic processes they may ultimately be destroyed by erosion. A computer model of this process is demonstrated. The discussion is extended to other base metals and to the distribution of reduced carbon. Given the conditions required for the accumulation of coal, oil and natural gas the relationship between cumulative production and the unit cost of energy may also be distributed log-normally. Neither base metals or fossil fuels are likely ever to be exhausted but their cost may rise and this will encourage the development of substitutes or alternative sources of supply.
[12] UNDERGROUND STOWING, Report of the Working Party on Underground Stowing, North-East Leicestershire Project, (Joint author and editor), Department of the Environment, London 1983.
A survey of the history and practice of underground stowing in the United Kingdom and overseas with particular emphasis on pneumatic methods. A detailed description is given of the stowing methods used in the German coalfields based on a visit by the author to the Ruhr and Saar and additional information collected in Katowice, Poland. The report concludes that underground stowing is technically feasible but mining costs are substantially increased. Important health and safety aspects require closer examination. Further research on these topics is recommended as well as investigations designed to improve the integration of stowing into the longwall mining cycle. Stochastic methods might be combined with simulation methods to identify robust operating sequences that are less likely to be disrupted by delays and breakdowns elsewhere in the system.
[13] FACTORS AFFECTING THE LONG TERM AVAILABILITY OF BULK MINERALS FOR THE CONSTRUCTION INDUSTRY in Dahlem Workshop Reports, ÒResources and World DevelopmentÓ, (D.J. McLaren & B.J. Skinner Eds), Berlin, J.Wiley & Sons, pp 327-346, 1987.
With the possible exception of engineering works, buildings can be constructed from a very wide range of materials. The depletion of one particular building material raises few problems and it is relatively easy to substitute one material for another, if economic conditions make this desirable. The demand for building materials is greatest in areas of high population density and industrial activity. Quarrying often competes for the use of land with farming, housing, industry and recreation. Because of conflicts between different ways of using the land some form of statutory control is usually required. Without an adequate policy for land restoration mineral workings can have an adverse environmental impact. Only a few countries collect statistics on bulk mineral production so it is difficult to make international comparisons. Fairly reliable figures exist for cement production and cement use is strongly correlated with building activity. A statistical analysis based on cross-sectional data suggests that the consumption of cement reaches a maximum of 0.6 tonnes per capita per annum at an income level of about $7,000 per annum. Thereafter demand declines steadily and per capita consumption may be less than 0.2 tonnes per capita per annum for incomes greater than $14,000 per annum (all prices in 1982 dollars). This suggests that once an economy has passed through the initial stages of industrial development the pressures on bulk mineral resources are partially relaxed.
Risk Assessment
[14] ALTERNATIVE STRATEGIES FOR FIRE PROTECTION AT THE RAOC WAREHOUSE, DONNINGTON - AN ECONOMIC ANALYSIS, BRE Note N33/88, Fire Research Station, Borehamwood, 13pp, March 1988.
The complete destruction of a warehouse at the Central Ordnance Depot, Donnington, in June 1983 demonstrated that fires in high-racked storage spread rapidly and are extremely difficult to control. The facility at Donnington was reconstructed during 1987 incorporating a fast response sprinkler system. The design of the system was based on recommendations made by FRS following a series of full scale experimental fires. Analysis of the gas temperature profile in this type of fire shows that it follows an S-shaped logistic curve and the rate of temperature rise often exceeds 10 °C per second. Under these conditions a fast response sprinkler responds about 25 seconds earlier than the glass bulb type. Fast response sprinklers can therefore significantly reduce the probability that a fire will develop to a stage at which control becomes impossible. A fire can be regarded as developing through several discrete stages. At each stage there is a probability that the fire will either be extinguished or develop to the next stage. Values can be given to these transition probabilities on the basis of experimental data, statistical surveys and informed judgments. This method has been used to evaluate different strategies for the fire protection of the Donnington warehouse. The results suggest that the use of fast response sprinklers might reduce the expected annual loss at the new warehouse by about £1.7 million. For this warehouse alone the expected saving is an order of magnitude greater than the total cost of the research project on which FRS advice was based.
[15] THE DEVELOPMENT OF A FIRE RISK ASSESSMENT MODEL, Proceedings of the International Conference on Probabilistic Safety Assessment and Management (PSAM), Beverly Hills, Feb 1991 in Probabilistic Safety Assessment and Management G. Apostolakis Editor, Elsevier Publishing Co., Inc. 1991.
Simulation models can be used to compensate for our lack of information about real fires and to work out the fire risk implications of new materials, building designs and protection systems. A fire risk assessment model is under development which will be used to decide priorities for remedial action and test the validity of new guidelines for building control officers. The model includes mechanisms representing the physical and chemical processes of fire development as well as the behaviour of people trying to escape from or suppress the fire.
[16] A FRAMEWORK FOR THE ASSESSMENT OF FIRE SAFETY DESIGN IN BUILDINGS, BRE Client Report CR91/91, Fire Research Station, Borehamwood, 71 pp, August 1991.
A framework is needed for the assessment of fire risk in buildings, particularly for those designs that use novel materials and layouts. A review of conventional methods of risk assessment suggests that none of them is adequate for the analysis of fire hazards. Fire realm analysis which was specifically developed to quantify fire risk fails to capture important aspects of fire development and demands extensive data which is seldom available. A new approach to fire risk assessment is proposed which depends on simulations of the total fire situation. It is based on accepted physical, chemical, toxicological and behavioural relationships, allows for continuous system interaction and time dependence and uses available experimental and survey data. Probabilistic estimates of risk can be efficiently and economically derived from the model by Monte Carlo techniques. This methodology has many practical applications including building design, remedial measures, emergency planning and training for fire safety professionals and the general public.
[17] THE DEVELOPMENT OF A FIRE RISK ASSESSMENT MODEL, BRE Information Paper IP8/92, Fire Research Station, Borehamwood, May 1992, 4pp.
Simulation models can be used to compensate for our lack of information about real fires and to work out the fire risk implications of new materials, building designs and protection systems. This paper describes the fire risk assessment model which is under development at the Fire Research Station and will be used to decide priorities for remedial action and test the validity of new guide-lines for building control officers. The model includes mechanisms representing the physical and chemical processes of fire development as well as the behaviour of people trying to escape from or suppress the fire. The paper will be of interest to building control officers, fire prevention officers, fire engineers and architects.
[18] A MODEL FOR FIRE RISK ASSESSMENT IN DOMESTIC HOUSES, BRE Client Report CR159/92, Fire Research Station, Borehamwood, pp 34, September 1992.
Fire risk in buildings can be assessed by the use of simulation models representing the whole fire system. This must include both the physical and chemical fire processes and the toxicological and behavioural factors that affect the occupants. These models are based on systems of non-linear first order differential equations. They can be used to compare the effectiveness of different fire protection measures and probabilistic estimates of risk can be obtained by applying Monte Carlo methods. The technique is illustrated by the example of traditional British three bedroom house. The model has been configured to represent the layout of the house so that the effect of a variety of fires on the occupants can be observed under many different circumstances. It is intended to use the model to compare relative levels of fire protection required in two and three storey houses.
[19] AN INTRODUCTION TO CRISP (PROTOTYPE v1.0): USER'S MANUAL, BRE Client Report CR185/92/92, Fire Research Station, Borehamwood, 26 pp, November 1992.
CRISP is a method based on simulation models, for comparing fire risk in buildings. An example based on a standard two storey house is used to demonstrate one of the ways that CRISP can be applied. Instructions are provided on how to input data, run the model and display results. This prototype version of CRISP should not be used to make decisions that might affect life safety.
[20] SIMULATION MODELS FOR FIRE RISK ASSESSMENT, Proceedings of the Technical Conference on Fire Safety by Design: A Framework for the Future, Fire Research Station & Institute of Fire Safety, Borehamwood, November 1993, Fire Safety Journal 23 (1994) pp 159-169.
Fire safety engineering cannot develop into a mature discipline like civil and electrical engineering until practical tools for the comparison and measurement of fire risk are widely accepted. The Fire Research Station is developing a fire risk assessment methodology (CRISP) based on simulation models and Monte Carlo methods. Designers will be able to use these tools to meet quantified criteria of safety and cost.
[21] COMPUTER SIMULATION FOR FIRE PROTECTION ENGINEERING, Chapter 5-1 in SFPE Handbook of Fire Protection Engineering (2nd Edition), J.M. Watts Editor, National Fire Protection Association, Quincy, MA, 1995
Progress in fire protection engineering is handicapped by data limitations and the difficulty of conducting realistic experiments on complete fire systems involving human behaviour. These problems may be overcome by applying a variety of procedural simulation model originally developed in operations research. These models make efficient use of the available data and can be used to test fire protection strategies. They can be complemented by Monte Carlo methods to take account of uncertainty in the data, measure the sensitivity of the casualty rate to fire protection measures and estimate fire risk.
Back
to Home Page