Cost Estimation Software - A Comparison
LineLab can be used to estimate production costs. But is LineLab a Cost Estimation tool?
Bottom-Up, Top-Down: The Basics
The fundamental goal of cost estimation is to apply the tools and metrics of cost accounting to production that has not yet occurred. Not only does cost estimation incorporate all of the complexities of cost accounting, it also introduces the difficulties of capturing production before it even occurred. Fundamentally, cost estimation methods are classified as either of two main approaches: top-down methods and bottom-up methods. [1][2][3]
Top-down methods estimate costs based on the cost of similar, past projects. Bottom-up methods, on the other hand, estimate cost by considering all of the resources (material, labor) which will go into the final product.
Bottom-up methods arrive to a cost estimate for the final product by summing up the cost of the constituent parts. This has the subtle effect, especially when estimating manufacturing costs, that the impact of variation on production flow is ignored - sometimes the user gets to make a guess, like 30% of total costs, but of course, depending on flow behavior, rate, and product mix in the system, the real cost premium may be much different. Also, bottom-up approaches require detailed system knowledge that may be difficult to obtain in complex organizations or early-on in the design phase.
Top-down approaches, on the other hand, rely on vast libraries of historical data. This makes top-down approaches problematic for representing unprecedented system changes. Sometimes, users will specify a "percent new design" variable (or "heritage", respectively). An additional cost factor is then applied based on the "new design" percentage. Most top-down cost estimation methods are weight-based, meaning the average cost per pound is derived from past projects and applied to the expected weight of the new part.
The Association for the Advancement of Cost Engineering (AACE) sees top-down or "parametric" cost estimation methods being used in the early phases as less than 15% of the project are defined (and resulting in an accuracy range of +/- 20-50% error), and bottom-up for the final, detailed, engineering estimate with 50-100% of the project defined (resulting in an accuracy range of +/- 3-15%). [2]
Typical Cost Estimation Tools
Top-down, or "parametric" cost estimation tools are made to be used in the early design phases and are based on libraries of historical data. Often, users provide a number of subjective variables, e.g. perceived "complexity" on a 5-point scale, approximate "new design (%)", or expected part weight (often the key scaling factor). Uncertain inputs (least, likely, most) are available in many commercial software products. Based on the previous weight-cost relationships (considering complexity and other variables), an approximate cost is then derived by the software. Sensitivity analyses may be available for the quantitative inputs in the model.
Bottom-up tools provide separate costs for every single operation and sum these up. Many commercial tools have detailed process model libraries, but are not extensible, i.e. they are limited to process models defined by the software provider (to add custom process models, some providers suggest to book consulting time to have them implement an additional process model, but this is certainly not an option wherever the process information is confidential). Since detailed inputs are required, Bottom-up cost estimation products sometimes come as plugins to CAD software (e.g. SEER-3D, aPriori). After an initial setup, this allows for a rapid back-of-the-envelope cost estimation with design sensitivities. However, production flow effects are not taken into consideration. Sometimes a blanket x% margin is assumed for buffers and queueing, but the effects of variation on queueing behavior (in particular of queueing at different stages within the process chain) are not considered. Consequently, this type of cost estimation can be very helpful, but there are some important operational blind spots to be aware of. Depending on the industry, engineers may also criticize being limited to built-in process models.
However, the distinction between top-down and bottom-up today may not be as clear cut as industry standards make it seem.
Today, many cost estimation tools are based on correlations, yet rely on adding up the costs of separate parts, partly implementing a bottom-up approach (interactions like queueing effects are not considered). This way, software providers can combine the relative convenience of using historical data (vs. having to develop detailed models of the process physics) with the convenience of a CAD plugin (in the case of Facton, which extracts BOM trees from CAD data). On the flipside, of course, the accuracy of the method drops off significantly: neither are process physics captured, nor the effects of complex manufacturing systems. In particular, the key limitation of top-down approaches still exists: historical data can't predict the behavior of novel process chains. Unsurprisingly, combining the relative convenience of top-down cost estimation techniques with the simplifications and idea of linearity of bottom-up techniques, the downsides and error sources of both are also combined. This makes these tools suitable wherever +/- 20-50% error is negligible and no important investment decisions are made (e.g. limited runs).
LineLab
LineLab borrows some aspects from both top-down and bottom up cost estimation methods, but it is capable of modeling production systems with high accuracy, capturing complex production system dynamics, and precisely determining possible utilization and required inventory. This makes LineLab highly accurate in representing even unprecedented system changes.
With LineLab, engineers enter basic model inputs about the process chain and production system. In that respect, it differs from those "parametric" cost estimation tools, where users get a rough idea of costs based on historical correlations and estimated inputs for weight, complexity, and % new design. Some physics-based process models are provided for LineLab, similar to bottom-up cost estimation tools, but it's up to the user to use them or implement new, custom process models.
In addition, LineLab provides custom "parametric" inputs, and it can process uncertain (min, likely, max) inputs, making it usable long before bottom-up cost estimation tools can be used. At the same time, the production flow models make LineLab's results (and sensitivity analysis) much more exact than those of bottom-up tools. That said, LineLab does not currently plug into CAD the way some bottom-up cost estimation tools do.
LineLab also provides a number of features not typically offered in cost estimation tools: advanced trade study tools that draw relationships between any input and any output (not just cost), business model and lifecycle analysis, sustainability parameters, unlimited parametrization and optimization of any number of variables, and of course a detailed view of operations inside the factory.
The following table summarizes the key differences:
| Mostly Top-Down e.g. SEER-Space, SEER-H, Price Systems | Mostly Bottom-Up e.g. SEER-MFG, aPriori | LineLab | |
|---|---|---|---|
| Core Elements | Datasets of similar products | Built-in process models | Dynamic models of production flow |
| Typical Inputs | Weight, complexity, %-new-design, BOM | Detailed dimensions / CAD file | Process chain, optional: parametric |
| Impact of Production Flow | Unable to capture expected changes | Not captured | Automatically calculated, high accuracy |
| Custom Process Models | Some allow for custom correlation data (e.g. SEER-H, but not SEER-Space) | Not usually possible | Yes |
| Built-In Process Models | Usually included (correlation data) | Yes | Examples, but focus is on custom processes |
| Sensitivity Analysis | Depending on inputs, may just be qualitative | Yes, usually design inputs | Yes, all inputs incl. custom |
| Design-to-Cost | Most tools | Yes | Yes |
Bottom Line
You can use LineLab to estimate production costs. However, LineLab is not a classic cost estimation tool, but a new and different method. It captures the interplay of operations, not just separate steps, and permits more detailed modeling. LineLab does not use databases of historical correlation data, but it can achieve higher accuracy, and truly shines when pioneering new designs and processes.
- "NASA Cost Estimating Handbook", 2020
- Christensen & Dysert: "Cost Estimate Classification System - As Applied in Engineering, Procurement, and Costruction for the Process Industries", 2005
- Boehm et al.: "Software development cost estimation approaches - A survey", 2000