OR bog/il lo/II I/fr /
7wioíio
tIar/ii 2001
The Role of Internal OR: - a personal view H GJones However, there were around 70 products br which stock might he safely macle without undue risks of being left with unsold stock. This could be produced iii the first fèw weeks of the year when order intake was very low (and the bottling hall workers on short
I have read the paper (Orrnerod, l999) by Richard Ormerod in OR Insight with in/cecil and I do not disagree with his analysis that the filure of OR appeaev to be with external
consultante. hut I be/ieee that there will v/ill he a role fir internai OR, albeit less glamorous, ifl tile fields of (a) rflomlonflg pos tems' output and performance (bY) y stem
time) and utilised in the autumn to meet the Christmas peak when overtime was generally
implementation.
required. - 00 0 o O-
It then became my task to persuade the Board of Directors of the logic of making for stock. As is not unusual in a small or medium size company, it fèll to me to draw up a stock list, implement the work and
Monitoring There are too many computer systems which may
control it until the system reached maturity and could be handed over to routine operators. I must
have been satisfactory under test conditions but have fàilecl when fàced with real life. There are probably less conspicuous installations that are fàiling to achieve true potential by not providing the information in the correct form for management to
admit I enjoyed this aspect of the work.
The reluctance to make stock was understandable. During my forays in the warehouse I discovered
As I pointed out in the OR Newsletter Jones, 1997) control the organisation more effectively.
among others 100 cases of Macnish 12 year old whisky (1200 bottles of a luxury product) which had been lying accumulating dust for a couple of years. One might ask how could this happen when bottling
there is a strong case for internal OR to be used first in establishing the facts and second in improving the situation.
was only to customer orders? There are at least three possibilities.
Implementation
The order had failed to catch a designated
I propose to illustrate the kind of problem one meets in implementation by referring to my own experience of about 34 years ago. I was at that time
shipment and the customer had gone elsewhere in a highly competitive market.
working as an internal consultant for a company
The customer had 'gone broke?.
employing
1400 people
engaged in
distilling,
The law governing the specification of the
maturing, bottling and selling Scottish whisky on a worldwide basis. At that time the company sold 1475 different products with perhaps only minor but necessary diflèrences between some of them. It was a long established company policy not to bottle for
I did not follow up to find the correct answer, but I
stock but to bottle only against a customer's firm order. I analysed the pattern of order intake and
did use the incident to ensure that we had a tight system. The three possibilities cited are also a
recognised that, in general, it would indeed be folly to make for stock because of the variable nature of the ordering pattern and often relatively small volumes of many of the items on the product list.
warning against over enthusiastic making for stock.
product in the country of destination might have changed - not an unusual situation.
In view of the earlier reluctance of the Board of Directors, it was also important that the system did not create 'dead' stock.
24
The following is an example of' possibility 3 (above).
At the time I was working on the cased stock problem customs officials in niany countries of
bet was for sixpence. A modern bottling hall, of course, will have several fully automated lines running at 500 bottles per minute, including
bottle tested contained less than the volume specified on the label: production then played safe by
stencilling, with only 2 people per line. Impressive change in a traditional Scottish industry - hut it took nearly 30 years to achieve.
overfilling by a small amount. Italy on the other hand would also reject the whole shipment if the
In practice every item of an incoming customer
destination would reject a whole shipment if one
nominal volume was exceeded in any one bottle tested; the concern was that Italian duty was being evaded. Because whisky was filled to height in the bottle, it was an easy adjustment just to alter the height of the filling nozzle. But the internal volume can vary from bottle to bottle, so production had to he doubly cautious. Rationalisation over years has led to a reduction in the number of stock items but it is still counted in hundreds.
order had to be checked against the stock list to see if stock was available and, if so, a decision was needed as to whether that stock would he allocated to the
order: certain items had a nuisance value in that they dislocated production by requiring special features such as a particular blend or strength or tint of whisky (which generally meant a small volume order). If a stock item had just been made, and the
order was for a large number of eases, and if production capacity was available, it made little sense to double handle manually fairly heavy cases
To digress further, in order to inject a little local
in and out of stock.
colour, a so-called 'high speed line in those days was operated by 25 to 30 staff for whom an output of 60
At that time, although the company had moved
bottles a minute, averaged over the week, was
away from the Diekensian quill, it was still heavily committed to pen and ink procedures. The logic of
considered a good performance. Bottles were filled
introducing computer control to help operate the cased stock was recognised, and the appropriate
automatically as described already (the fèed was automatically stopped when the whisky reached the nozzle), a screw cap was applied and tightened by another machine and yet another machine applied and rolled tight a 'lead' seal. The bottles were then
decision was made to venture into the unknown.
Following a survey of the market, I recommended
that the most suitable for our purpose was the
fed on to a conveyor belt from which four staff
Gamma 10 manufactured by De La Rue-Bull. lt may come as a surprise to the younger members of the OR fraternity that the 'most suitable' machine
working in parallel lifted a bottle, presented it to a labelling machine, pressed a foot switch and returned the bottle to the line with the main body
label now in position. Usually 2 or sometimes 3 further labels were applied by hand; two more staff polished the bottle to remove excess paste and the bottle was loaded into a cardboard case or more rarely into a wooden box. The lid of the cardboard case was pasted down, and the case loaded on a
had a core of only 4K, and the backing store (for the stock list) was held on punched cards. Punched cards carrying the order details had to be sorted oft'line into the stock list before each run. The Gamma
pallet to be transferred by forklift truck to an off-line station where 4 staff stencilled the cases. The output
was allocated, a new card was punched with the new
10 was considered to be an advanced machine because it had a second card fèed so when a stock
stock details and the old stock card was rejected. Card operations of this nature are highly dependent on competent handling. We were fortunate in
from 6 bottling lines came through this operation. On one celebrated occasion, I bet the production manager that his pet line could not produce 30,000 cases in a working week. He published the bet
recruiting a very efficient young man who is now the
computer manager of a rather more sophisticated
Our machine did have a line printer
widely and we both visited the line twice each day to
complex.
check on progress and to encourage the troops. Friday afternoon saw intense interest when it was
whose speed was geared to the rate at which cards could be handled.
touch and go who would win the bet. We were both
pleased when the final figure turned out to be
In retrospect it amazes me that we were able to
30,100 cases. Our objective of course was to create interest in a rather boring job and to establish what could be achieved. Could an external consultant have behaved like this? \Ve never disclosed that the
achieve as much as we did on the Gamma 10. In addition to running the cased stock system, we also produced each week a two page summary (Jones, I 973a) of the company's production system which 25
that at that linie we Ilad little fititil ill tile annual forecasts supplied by custOnlcrs. \\'liiskv sales \\orldwidc were in a buoyant phase. 'flic conipanv usuall' had 0111V one custonler or agent ill each
was used as the basis loi weekly coliliol nleetini4 held by the managing director.
Once a year the machine was alsi heavily involved
country. OR aggregated the sales of all proclicts for each customer and plotted several veai's o!' history 00 semi-log papc'r; this has the advantage that compound growth appears ils a sti'íught line, the slope of whicil defines the rate o! growth. Sales for nearly all the customers were growilg at rates ranging from 2 per cent per ann um to I O per ('eilt. Many o!' the customers ('ould not heheve that this could continue. (A very Iviv of' tile customers were losing sales at rates up to 2 pci' ('cnt per annum; they
in aggregating lorecast by country into whisky recipe forecasts aoci converting these into production platis
fbr the distilleries. (31 course it took time to reach this position Also, as computer technology moved on, the foithful old machine was upgraded and we became more ambitious iii our work, but always keeping one foot on the ground to ensure that our alIlhitiollS were not beyond our capabilities (Jones, 1 973b).
too had dillicultv in accepting thai this
1c) return tc) the eased stock problem. In the early days, when whisky was shipped overseas, each case was loaded manually at tile clocks intc) a net, which was then hauled by a crane into the hold of tile SlliJ). At the port oí destination the process was reversed. Each case of whisky was stencilled with a serial
voulcl
continue.)
I well remeniber the then cllairnlari o!' the company telling me that there was notllilig like a revolution for improving the sales of wlliskv ill tile country involved. It was reasoned that the wlusky had several functions; the wluskv acted as a stimulant to
Ilumber, and with the name and address of the customer. 'This information had to be carried on the punched cards, with the consequence that each stock item usually required more than one card.
bravery during tile plotting plise, as a solace ill defeat, or as an aphrodisiac in tile event of success.
Small wonder that whisky is described
'Ehe method oí shipping may appear somewhat crude. Although a fèw cases of whisky regularly disappeared en route in South America and although it was not unknown fôr a bottle to he
ill
some
circles as 'the water of lift'. I sought neither to prove nor disapprove the relationship on tile grounds that OR should not undertake a profitless investigation
that might lurIl out to be based on nothing more solid than folk lore. Anyway it would have been poor psychology as well as profitless fbi nie, a newcomer to the industry, to he doubting the word of' the chairman on information I cid not intend
broken at the docks in Scotland, over a period of five
years I am aware of only one serious mishap. This occurred when 20 cases of Ballantine destined for South Africa were shipped to Australia while on the same day 20 cases of Ballantine destined for Australia were shipped to South Africa; none of the 480 bottles involved could be sold legally and were shipped back to Scotland leaving behind two very
using.
This illustrates tile care that needs to be exercised when choosing tile first (and iildeed later) assignment if tilere is any possibility of choice, with
dissatisfied customers. Currently with the use of containers for exports life is rather more simple.
the project preferably having a clear and positive influence on the organisation. Tile cased goods problem satisfied this criteria, and some 25 years later was still a key factor when the bottling lines
A further complication was that none of the formulae normally used to calculate stock levels were applicable. The amount of stock to be made at
were automated together with the constructioml of an automated and computer controlled warehouse.
any one time was determined by an individual (initially me) taking a balanced view of several factors such as the annual forecast, the volume of
Cased goods stockmaking became part of a trilogy of models, listed below, which were used very
orders for a product received in the year to date, the
existing stock level and the production capacity available. Data on all these factors (apart from the last) had to be carried on the stock card in order to
regularly, being nlanipulated at first by hand but
produce a working printout.
1.
One of the reasons
subsequently by computer.
for initially restricting the
number of products on the stock list to just 60 was 26
A model of the industry which was used to predict shortages or successes of particular whiskies in future years. (There is in fact a considerable market between the companies,
engaged in the buying and selling of bulk whisky in ('askS.
A model for converting forccasts into distillery production plans.
A hand model involving cased stock making,
It is recognised that over the years most comparues will have sulved the more obvious 'easy' problems. A major step may be needed to meet the demands of new technology, or to conform to the latest and well advertised management concept. This is where an external consultant with a team of experts gives the appropriate line of attack.
order entry and various targets. These were updated weekly and used to monitor company performance. It is unlikely that an external consultant would
recognise the utility of the trilogy although the concepts behind the industry model have been
A good internal OR consultant will understand the workings of the organisation and be aware of areas for potential improvement: simulation problems, for example, are probably better solved locally; in the realm of soft OR, the internal consultant will know
at least some of the characteristics of the people
around Ihr years.
involved.
Over the years. we developed a forecasting system which worked well (Jones, 1975), and which I later discovered employed a form of Delphi technique. OR produced a forecast based on extrapolation of the graph (sometimes modified by market economy knowledge), and a copy was sent to the customer to
Lest it be thought thai this paper is inspired by green grape sentiment, may I he permitted another personal note. In 1960, I was invited (not pressurised) to join a firm of OR consultants. I was
either agree or modify and return the forecast.
tempted but in the end decided I preferred my relatively green life-style in the sticks to working in the traffic hound conditions of London.
These figures were then revised in a meeting with the managing director and the sales director with To the interested reader
OR present as a facilitator.
The implementation of the cased stock making has been described in outline at some length because it illustrates the far reaching problems that can arise before achieving realistic performance, even in a system which at first sight looks simple and straightforward. Who better to take charge of the implementation that an internal OR consultant?
Ormerod 1999,
'Blarkett:
the hither of OR', OR InyIit, Vol
12.
Issue 2, pp 14-20.
Jotics H G, 1997, Letter to 01? ,Vewsletter.
Joncs H G, l973a, 'The use of forecasts in target setting'. OR Quarter/n, Vol 24, pp 547-560.
Joncs H G, 19736, '(omputing in a medium sized company - a case study', OR Quarter/v. Vol 24, pp 375-389.
Jones H G, 1975, 'Suhjective probability in forecasting. \'ol. 3, No 3, J) 327.
The future It
is not pretended that the same problems that
occurred more than 30 years ago are likely to recur
today. A current tendency
is
for companies to
purchase large (and expensive) packages covering a linked spread of activities witI1n the company. Successfùl implementation requires some (often considerable) bending of the company's processes to
suit the package. Here is a situation in which an internal OR consultant should be capable of defining the limits of the problem and its linkages with other activities, and should be closely involved in testing and implementation. \Vhen the solution is running live, the internal consultant should mcnitor the output to ensure that it fits company requirements and is a help and not a hindrance.
27
Orne4a,
H G JONES. now ninety-two years old, can lay claim more (han fliOst of us to having spent a hgiime in OR. Involved from the early days of the hscipline, working first in the glass and then lie steel industry. HG has lived through some of the key shanges in OR, Ibilossing the cork promise of the early I 9Üs, when he need to overcome the post-war shortages of consumer goods gave a boost to the scientifically-based methods which had gained such pre alcnce in wartime Britain. Early work on, prumnily, cspstcity planning sowed the seeds o! a career which íinallv took him to whisky distilling not a bad place to end up!
in 1994 where he saw the future, HG unequivocally answered "the application of computers": clearly he is still
Asked
looking to the ongoing development of the discipline.
For a more detailed review of HG's lifr and times, readers are referred to the May 1994 OR Newsletter, from which the above is abstracted. Biography (Prepared by Steve Clarke CspvraIit © ()peiatiosa/ Reseaoli ,Çszetj
28