Estimating Production and Supply Costs

It seems the shed/carport industry and various others believe that if can produce workshop details will get a more accurate cost than working with documentation at less complete level. It is not so.

As I have mentioned before price is what the market is willing to pay. Business typically hopes that the price is higher than the cost of production/supply, if not then they don’t produce. So for certain want to get the cost as accurate as possible, however doing so depends on cost data. Even when a producer has finished everything and they have a summary of all associated costs they can only produce a cost estimate, they cannot determine the exact cost of the item. Why? Because they do not collect the necessary cost data, and doing so would cost more than the benefit provided.

Consider a small job shop, steel fabricator. They probably have a process layout, one area with drill presses, another with cut-off machines, an area for welding, and another with various cut-out machines. The personnel typically stick to one process area, operating one type of machine.

So their time sheets, can simply identify the number of hours spent at work, or they may be more detailed and identify the job they spent time on. The operator of an overhead crane for example probably switches which project they work on every 10 minutes or so, so not overly productive to track their time by project. So their time is an overhead to the project costing, either applied on the basis of primary hours worked or tonnage of steel on the project.

Now have computer software which determines member sizes and can produce material and cutting lists at the point of sale, from which the cost and price can be determined. However, whilst the structure may require say a 5m length of steel, it is likely that the fabricator has to buy a stock standard 6m length of steel. They thus have 1m of scrap steel. If they employ workshop detailers and steel designers on staff, then they can probably find a use for such off-cuts in the form of brackets for connections. Otherwise its waste and extra expense.

Similarly there are economic order quantities (EoQ) for various other materials. For example may need 80 bolts, but have to buy barrel of 100. Many of the items, such as bolts, flat bar, sheet, plate, angles, and welding consumables are used across multiple projects. So orders from multiple projects can be consolidated to obtain quantities to be purchased by the fabricator. The fabricator then comes up with a standard costing allowance for these items to allocate costs to individual projects.

Now with the introduction of building information modelling (BIM), there is increased talk of levels of detail (LOD) with respect to the documentation produced. The BIM industry appears to have defined some standard LOD’s: LOD 100, 200, 300, 500, which doesn’t appear to be much use for this discussion.

Structural design goes through various stages. It usually starts with stick diagrams, because whilst the basic set out of framing is known the size of the members is not. Once the member sizes are known then details sections and connection details can be drawn.

In 1996 the Australian Institute of Steel Construction (AISC now ASI) published : Costing of Steelwork from feasibility through to competition. As an industrial engineer, I thought it was a good idea and about time they moved away from $/tonne costing.

The problem with $/tonne costing is it encourages the structural designers to reduce the weight of the structure, this in turn results in increased fabrication requires, which results in increased over all cost, as labour typically costs more than materials. Thus a thicker base plate is generally preferable to a stiffened base plate. A larger beam better than one with lots of fly-bracing. More importantly the fabricators had historical data converted into $/tonne rates, and their estimating the cost of the buildings became completely out off alignment with the actual cost of supply. The labour content of fabrication was increasing but their rates were being adjusted to compensate.

Now from the consulting side of things we expect costing to be done by quantity surveyors (QS), using publications such as Rawlinsons or Cordell’s cost guides. These have cost information collected from across the country. QS’s collect information locally from projects, and suppliers, but they also use this information to calculate standard rates, for various kinds of construction. These rates maybe in $/tonne, $/sq.m, percentages of capital works.

Searching the internet can get that shed type construction likely to be somewhere between $100/sq.m to $500/sq.m, whilst house type construction (kitchens, bathroom’s etc…) likely to be somewhere between $1000/sq.m and $4000/sq.m. So as an example assume have a shed 30m span and 80m long. With first 20m used as office space. So the office space is around 25% of the total building area. Using these various rates combining minimum with minimum and maximum with maximum, get an estimate for the building between $780,000 and $3,300,000. This is a large variation in the cost and the reason to employ the services of a QS to know whether the construction lies at the low end or high end of the range. The QS should be able to produce such estimate without need for any drawings, just a rough guide to areas required and type of construction. The cost range also identifies the problems which result when owners start making changes during construction and start pushing costs closer to the top end of the price range.

With architects drawings the required areas can be refined and the type of construction better identified, as well as primary cost items costed directly, so the costs can be refined slightly. The bulk of the cost however is still based on the $/sq.m estimated rates.

The AISC/ASI article identified three stages to costing:

  1. Pre-Design Costing
  2. Indicative Costing
  3. Detailed Costing

The stage 1 costing is typically based on the $/sq.m rates and suggested to have an accuracy of some plus/minus 20%. Which I assume means that what ever is estimated, if I add 20% to it that should be the upper limit.

Stage 2, requires preliminary architectural drawings and some structural estimating of member sizes. The accuracy is plus/minus 10%. With it being considered that add 5% to allow for simply supported connections, and 10% for rigid construction.

Stage 3, requires that the engineering be completed. It does not require workshop details or cutting lists, steel fabricators typically produce as part of their supply. There is no point producing such detailed drawings if the building is not affordable. The accuracy is plus/minus 5%.

Can we get rid of the 5% and bring it to zero. The answer is no we can’t. The only way we can be certain of the costs, is if we have built the structure multiple times before and there is no variation in supply costs. So if throwing buildings of an assembly line, then have near certain costing of the product, but costs may change tomorrow.

The shed/carport industry doesn’t throw buildings of an assembly line, they attempt to design and price what the customer requests at the point-of-sale: each product is different. As a consequence of the difference, they have stop start, production processes: which is not very efficient.

When the architectural and engineering drawings are complete and ready for development approval or final contract, the drawings have all information required for stage 3 costing.

The member sizes are know. From centre line geometry member sizes can be estimated. Members maybe slightly shorter or longer than centre line geometry lengths, due to detailing at the connections. But the lengths can be estimated, and the stocks lengths which have to be purchased also estimated. The count of all members can be determined and the nominal length of all members determined.

From there can either allow for connection components and other small parts by a percentage allowance as indicated above. For example once we have the tonnage of primary steel has been determined we can add an extra 10% for connections. Whilst labour can also be added as a percentage, such as assuming the cost of the materials is 30% of total cost.

But say we want a better estimate of labour requirements, not the least of which we need to plan the labour force, can we do that? The ASI cost guide, break downs the steel work into various fabrication tasks, which are largely based on connection details. It doesn’t however break them down into specifics, it classifies into light, medium and heavy, either by description or weight of steel sections.

For example welding an end plate to the end of a universal beam (UB) isn’t so much dependent on the size of the end plate, but dependent on the size of the UB and the size of the weld. Heavier sections likely need longer preheating times, and larger welds. Whilst the original cost guide only has the 2 classes, I would expect that individual fabricators would increase the level of detail to meet their needs. So they could estimate the costs and set standard costs for each size of universal beam. At the end of a project though all they would know is that a welder spent so many hours welding on the project. Well not actually welding, most of their time may actually be spent with preparation and material handling. But they probably get paid the same hourly rate no matter what.

If there is a project manager monitoring progress, then may know at the end of each day how many hours were spent on a specific part of a project. So rather than simply knowing that 200 hours spent on welding at the end of the project may know they spent 100 hours welding rafter end connections, the remaining hours spent on cleats and minor details. We can also monitor welding consumables, simply by measuring quantities remaining at end of each day. So from hours, consumables and other materials can get an estimate for the cost of each connection. Not an exact cost, an estimate.

Consider other items. A bracket may require a steel bar be cut to length and then holes drilled in it, this bracket is then either bolted in place on site or welded in place in the workshop. A more complicated bracket may have to be cut to shape either after bar is cut to length or it may have to be cut shape directly from steel plate. A bracket may also have to be folded to shape, or welded from flats. Holes may need to be drilled before final shaping or after final shaping.

A simple item may involve a multitude of operations, and it will spend a lot of time moving around the workshop from one workstation to another to be worked on. So whilst it may take a few minutes to make a part, that part may take a few days to make as it moves around a machines. For example a days worth of parts from the cut-off machine are moved to the drill press a day later.

So whilst if have the drawings of a part can identify production operations such as cutting, drilling, and welding. And can measure hole depth, cutting lengths and length of weld, none of this is a direct measure of the full labour involved. It may take 1 minute to drill a hole through a steel plate, but 3 minutes to align each hole with the drill spindle and clamp in place.

Consider a requirement to put two holes in each flange of a UB and two holes in the web. A total of 6 holes, assume each hole takes 1 minute to drill. If drilling manually then each hole has to be drilled one at a time, and the UB has to be rotated to create a horizontal surface, so at least two rotations. Assume each rotation takes 1 minute. So total time is 8 minutes. If use an automated beam line, the beam doesn’t need rotating, and each tool has two spindles and can drill two holes at once. So it takes 3 minutes to drill the 6 holes, if only have one tool arm. If have three tool arms, then can drill top, bottom flanges and web all at the same time, so total of one minute. A beam line can have more than one workstation and can carry out multiple operations along the length of a steel work piece. We thus have the potential for three organisations with 3 different operating processes and 3 different costs. With different distribution of costs between capital equipment and labour.

Now I’ll hazard a guess that the cost of drilling a hole is less than $1, as a steel fabricator do I care whether the cost is $0.25 or $0.80 per hole. If project has 200 holes then at $1, the cost of holes is $200. For a project costing hundreds of thousands of dollars its neither here nor there. It certainly isn’t worth a few thousand dollars of workshop detailing up front to determine if have 180 holes of 200 holes. Nor would consider spending a few hundred dollars tracking to the level of detail to determine if the cost is $0.25 or $0.80.

Rather would either adopt a percentage rate. For example know the rafter costs say $200 each so add 5% to cost to allow for detailing, so fabricated cost becomes $210. Alternatively we can have a standard rate for connections say $50 each, some will cost significantly more than this others, significantly less, but not interested in the detail, we want a quick number. So we simply count all connections and multiply by our adopted rate, which would be based on historical data.

The historical data being so many hours spent on a project of so many tonnes of steel. So we get labour hours at hours/tonne. Which takes us back to original problem with $/tonne. So fabricator has to collect more data, or spend more time on standard costing. They can look at past drawings and count number of connections. So then get average hours/connection. They can classify connections, and identify them as percentage of total connections. Can also assign a rating to each type of connection, to define the effort required. For example welded cleats for cladding rails may be greatest percentage of total connections but require the lest effort.

So without going on the shop floor with a stop watch, and without part details, we can estimate minutes for each connection or fabrication process. We then define a standard cost, which allows for materials and labour. In effect a steel fabricator simply collects data to refine the ASI methodology.

Just to clarify, once have a part drawing, it is possible to define all tooling required and write up operational procedures for fabrication of the part. Based on feed rates and cutting speeds and other process data could then estimate partial production time. But it wouldn’t be total time as there are manual handling operations. An IE specialised in the use of predetermined time data could calculate the time required for the task in decimal minutes: but it would cost a fortune to get such time estimate. Better off going on the shop floor with stop watch and timing. But its not likely to improve anything.

To put another way if make 100,000 parts a year it makes a difference as to whether it takes 5 minutes to make, 2 minutes or 2.5 minutes. If its the first and only time making the part it is largely irrelevant if it takes 1 minute or 1 hour to make. The minimum cost is likely to be for 1 hours work. If make more than one, then likely to estimate number can make in 1 hour. It would mostly be a guess based on past experience, and general awareness of time taken. Unless actually been on the shop floor measuring process parameters and work times. However before can conduct a time study need to conduct work study to identify the work steps to be completed. If have a work study and an over all estimate of time for the task, then can also do some estimated break down of the over all time.

Assuming some process parameters. Say flat bar is cropped to length taking about 5 seconds for the cut. So in a 480 min day can produce 5,760 items, or a 456 minute day (38 hour week), 5472 items. Assuming $20/hr labour rate then rate is approximately $0.33/minute, to give $158.40 to $150.48 per day. Making each piece cost $0.0275. Assume the flat bar want to use costs $100 for 6m length and each item is 150 mm long, so can get approximately 40 items per stock length (depending on the importance of losses due to the cut width). So initial estimate of item cost is: $2.50 plus $0.0275.

However, it takes 5 minutes to move all needed flat bar from stock to the workstation. It takes 30 seconds to load each flat bar. If we get 40 items, then have 39 cuts. So 39×5=195 seconds of cutting plus 30 seconds of loading, to give total time of 225 seconds. So average time per item is 225/40=5.625 seconds. A mere fraction longer. But time available is now 456-5=451 minutes. So now get 451*60/5.625=4810 items at cost of $150.48, to give $0.0313 per item. Thus far labour an insignificant part of the cost.

It takes say 5 minutes to move all the cut plates to the drilling work station. Two holes are to be drilled. It takes 1 minute to set up each hole and clamp the work piece. Say the holes are drilled at a rate of 10mm/minute, the steel is say 8mm thick, so takes less than 1 minute to drill (0.8 min.). so total time is 2*0.8+2*1=3.6 minutes, time available 456-5=451 minutes. So 451/3.6=125 items per day at total cost of $150.48 to give $1.20 per item.

It then takes another 5 minutes to move all the drilled plates over to the welding bay. Where it takes say 2 minutes to locate and clamp in place each item. There is weld to each side, so assume 2*90=180mm of weld. Assume welding at the rate of 25mm/min. So takes 180/25= 7.2 minutes. The welder is paid say $40/hr, or $0.67 /minute, for 456 minutes, or $304/day. There is 451 minutes available for welding, so 451/7.2=62 items per day, at 304/62=$4.90 per item.

So total cost of fabricated and installed item is 2.50+0.0313+1.20+4.90 = 8.6313 or $8.63 per item. Now I could estimate per item, or I could estimate total:

  1. Cutting
  2. Drilling
  3. Welding
  4. Manual/Mechanical Handling

Missing from the estimate above is the 30 minutes to move the steel beam into the welding bay and then prior time to mark up the location of each cleat. Then 30 minutes to move the steel beam to next location.

Now someone could sit down and do all these calculations for a steel fabrication project, but in order to do the calculations they need to know the process parameters, the rates and times which I just made up.

The fabricator may know some rates such as:

  1. Cutting speed
  2. Feed rate (drilling/machining)
  3. Deposition Rate (welding)

But all the handling and setup times will largely be unknown, they will need to be measured for the job at hand and that is unlikely to happen, and no use for the estimate to get the current job and irrelevant for the next job.

So not going to look at the job at such level of detail to estimate a cost to win the job. Rather will get the major member sizes and quantities, and work out costs of those items. Will then work out the quantities and types of connections. Based on the type of connection will then allocate a standard cost. For routine types of connections and common sizes the standard cost will be close to the final cost assigned to the project. For one off projects the standard cost will deviate from the final cost, the objective for the fabricator is to make sure the final cost is less than the estimate, and the fee they get is greater than the cost.

The buyer on the other hand wants the cost estimate to be as low as possible and they don’t want it to change during fabrication and construction. The shed/carport industry also want their cost estimate to be as low as possible, because sales is on the basis of my rubbish is lower price than their rubbish. No one is saying my price is higher because we provide higher quality. The building industry largely works along the lines of we do it to the code, you’ll be lucky find anyone else who does. Which makes it one of the lowest quality industries we have.

So their superduper software which supposedly does structural calculations at point of sale, then determines cutting lists and costs from the cutting lists: does it really help with costing? Does it improve accuracy of costing? I doubt it, as it would need accurate cost data to be fed into the software, and accurate estimation of the labour content of fabrication.

Sure for c-section framing, mostly roll-forming cladding, cladding rails, and primary framing. Holes are punched in the flat before roll-forming, punching a hole can take less than a second. When C-section is ordered either pay a fixed fee per hole required, all all holes absorbed in the per metre rate for the section. So calculating the total linear metres of material and costing gives the main proportion of the total costs. However there are also a variety of small brackets and plates which need to be fabricated, and these need costing somehow. There are a limited number of c-section sizes so each bracket/plate could be costed based on size of c-section.

The issue isn’t whether we have a standard cost for each connection, but where we get the cost from, can having part details provide a better cost estimate than general connection detail. My contention is that part drawings don’t make a difference to the cost estimate, because the drawings do not provide any greater level of information which can be used to improve the costing process. This is because the costing process is based on crude parameters not refined details.

For many of the shed/carport suppliers all parts are purchased and nothing is made in-house, so product is just an assembly. Larger steel fabricators make everything, or outsource some fabrication. Where does the fabricator get the fee from?

For example fabricator knows they need a cleat, it is 8 thick, or maybe 90x8FL, it has two M12 bolts through it and is welded in place with 4 mm welds to each side. If they get a workshop detail which shows its length and the diameter of the holes can they improve their cost? Its doubtful because they are not costing the number of holes, the length of weld or the amount of steel. They are costing the total materials and fabrication costs, and providing an allowance for the connection details. They may not know the cost of an individual item because part of the cost is absorbed in some other component or assembly. For example could cost the cleats welded to rafter, or just include cost in over cost of cladding rails. Could cost cleats required for fly bracing, or incorporate in cost of fly braces. so whilst can reach a total cost for a building or structure, cannot provide cost for individual items within.

To get more refined cost estimates it is necessary to collect more refined data about the production processes, and collecting that information can cost more than the benefit it provides. Activity sampling and review of historical data will generally provide adequate aid to future estimating.


[01/04/2020] : Original [Draft]